You can deploy this model and start earning money today!

Static Embeddings with BERT uncased tokenizer finetuned on various datasets

This is a sentence-transformers model trained on the gooaq, msmarco, squad, s2orc, allnli, paq, trivia_qa, msmarco_10m, swim_ir, pubmedqa, miracl, mldr and mr_tydi datasets. It maps sentences & paragraphs to a 1024-dimensional dense vector space and is designed to be used for semantic search.

Read our Static Embeddings blogpost to learn more about this model and how it was trained.

0 Active Parameters: This model does not use any active parameters, instead consisting exclusively of averaging pre-computed token embeddings.
100x to 400x faster: On CPU, this model is 100x to 400x faster than common options like all-mpnet-base-v2. On GPU, it's 10x to 25x faster.
Matryoshka: This model was trained with a Matryoshka loss, allowing you to truncate the embeddings for faster retrieval at minimal performance costs.
Evaluations: See Evaluations for details on performance on NanoBEIR, embedding speed, and Matryoshka dimensionality truncation. In short, this model is 87.4% as performant as the commonly used all-mpnet-base-v2.
Training Script: See train.py for the training script used to train this model from scratch.

See static-similarity-mrl-multilingual-v1 for a general-purpose multilingual static embedding model. It's been trained for semantic textual similarity, paraphrase mining, text classification, clustering, and more.

Model Details

Model Description

Model Type: Sentence Transformer

Maximum Sequence Length: inf tokens
Output Dimensionality: 1024 tokens
Similarity Function: Cosine Similarity
Training Datasets:
- gooaq
- msmarco
- squad
- s2orc
- allnli
- paq
- trivia_qa
- msmarco_10m
- swim_ir
- pubmedqa
- miracl
- mldr
- mr_tydi
Language: en
License: apache-2.0

Model Sources

Documentation: Sentence Transformers Documentation
Repository: Sentence Transformers on GitHub
Hugging Face: Sentence Transformers on Hugging Face

Full Model Architecture

SentenceTransformer(
  (0): StaticEmbedding(
    (embedding): EmbeddingBag(30522, 1024, mode='mean')
  )
)

Usage

Direct Usage (Sentence Transformers)

First install the Sentence Transformers library:

pip install -U sentence-transformers

Then you can load this model and run inference.

python

from sentence_transformers import SentenceTransformer

# Download from the 🤗 Hub
model = SentenceTransformer("tomaarsen/static-retrieval-mrl-en-v1")
# Run inference
sentences = [
    'Gadofosveset-enhanced MR angiography of carotid arteries: does steady-state imaging improve accuracy of first-pass imaging?',
    'To evaluate the diagnostic accuracy of gadofosveset-enhanced magnetic resonance (MR) angiography in the assessment of carotid artery stenosis, with digital subtraction angiography (DSA) as the reference standard, and to determine the value of reading first-pass, steady-state, and "combined" (first-pass plus steady-state) MR angiograms.',
    'In a longitudinal study we investigated in vivo alterations of CVO during neuroinflammation, applying Gadofluorine M- (Gf) enhanced magnetic resonance imaging (MRI) in experimental autoimmune encephalomyelitis, an animal model of multiple sclerosis. SJL/J mice were monitored by Gadopentate dimeglumine- (Gd-DTPA) and Gf-enhanced MRI after adoptive transfer of proteolipid-protein-specific T cells. Mean Gf intensity ratios were calculated individually for different CVO and correlated to the clinical disease course. Subsequently, the tissue distribution of fluorescence-labeled Gf as well as the extent of cellular inflammation was assessed in corresponding histological slices.',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 1024]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [3, 3]

This model was trained with Matryoshka loss, allowing this model to be used with lower dimensionalities with minimal performance loss (See Matryoshka Evaluations for evaluations). Notably, a lower dimensionality allows for much faster and cheaper information retrieval. You can specify a lower dimensionality with the truncate_dim argument when initializing the Sentence Transformer model:

python

from sentence_transformers import SentenceTransformer

model = SentenceTransformer("tomaarsen/static-retrieval-mrl-en-v1", truncate_dim=256)
embeddings = model.encode([
    "what is the difference between chronological order and spatial order?",
    "can lavender grow indoors?"
])
print(embeddings.shape)
# => (2, 256)

Evaluation

Metrics

Information Retrieval

Datasets: NanoClimateFEVER, NanoDBPedia, NanoFEVER, NanoFiQA2018, NanoHotpotQA, NanoMSMARCO, NanoNFCorpus, NanoNQ, NanoQuoraRetrieval, NanoSCIDOCS, NanoArguAna, NanoSciFact and NanoTouche2020
Evaluated with InformationRetrievalEvaluator

Metric	NanoClimateFEVER	NanoDBPedia	NanoFEVER	NanoFiQA2018	NanoHotpotQA	NanoMSMARCO	NanoNFCorpus	NanoNQ	NanoQuoraRetrieval	NanoSCIDOCS	NanoArguAna	NanoSciFact	NanoTouche2020
cosine_accuracy@1	0.32	0.7	0.46	0.28	0.64	0.18	0.42	0.24	0.8	0.28	0.1	0.52	0.5714
cosine_accuracy@3	0.52	0.84	0.8	0.44	0.82	0.42	0.56	0.44	0.96	0.48	0.46	0.6	0.898
cosine_accuracy@5	0.6	0.9	0.84	0.54	0.86	0.5	0.62	0.58	1.0	0.54	0.56	0.62	0.9796
cosine_accuracy@10	0.78	0.94	0.94	0.64	0.96	0.66	0.72	0.7	1.0	0.7	0.74	0.76	1.0
cosine_precision@1	0.32	0.7	0.46	0.28	0.64	0.18	0.42	0.24	0.8	0.28	0.1	0.52	0.5714
cosine_precision@3	0.1933	0.5867	0.2667	0.1933	0.3733	0.14	0.3733	0.1467	0.3867	0.2267	0.1533	0.2067	0.6054
cosine_precision@5	0.14	0.544	0.18	0.16	0.26	0.1	0.32	0.124	0.248	0.188	0.112	0.132	0.6204
cosine_precision@10	0.104	0.452	0.1	0.104	0.148	0.066	0.244	0.076	0.13	0.14	0.074	0.084	0.5306
cosine_recall@1	0.1467	0.0805	0.4367	0.1519	0.32	0.18	0.0428	0.24	0.7107	0.0597	0.1	0.485	0.0398
cosine_recall@3	0.239	0.1605	0.7467	0.2983	0.56	0.42	0.0984	0.43	0.9253	0.1417	0.46	0.57	0.1236
cosine_recall@5	0.279	0.218	0.8033	0.3793	0.65	0.5	0.1196	0.58	0.9627	0.1947	0.56	0.595	0.2095
cosine_recall@10	0.4197	0.3143	0.9033	0.4838	0.74	0.66	0.1389	0.69	0.9793	0.2887	0.74	0.75	0.337
cosine_ndcg@10	0.3309	0.5681	0.6922	0.3651	0.6547	0.4041	0.3242	0.4534	0.8951	0.2643	0.4078	0.6111	0.5703
cosine_mrr@10	0.4453	0.7854	0.6397	0.3915	0.7485	0.3245	0.5041	0.3764	0.88	0.3998	0.3034	0.5837	0.744
cosine_map@100	0.2598	0.4335	0.6205	0.3024	0.5798	0.3389	0.1449	0.389	0.8594	0.205	0.3151	0.5683	0.447

Nano BEIR

Dataset: NanoBEIR_mean
Evaluated with NanoBEIREvaluator

Metric	Value
cosine_accuracy@1	0.424
cosine_accuracy@3	0.6337
cosine_accuracy@5	0.703
cosine_accuracy@10	0.8108
cosine_precision@1	0.424
cosine_precision@3	0.2963
cosine_precision@5	0.2406
cosine_precision@10	0.1733
cosine_recall@1	0.2303
cosine_recall@3	0.398
cosine_recall@5	0.4655
cosine_recall@10	0.5727
cosine_ndcg@10	0.5032
cosine_mrr@10	0.5482
cosine_map@100	0.4203

We've evaluated sentence-transformers/static-retrieval-mrl-en-v1 on NanoBEIR and plotted it against the inference speed computed on a RTX 3090 and i7-13700K. For the inference speed tests, we calculated the number of computed query embeddings of the GooAQ dataset per second, either on CPU or GPU.

We evaluate against 3 types of models:

Attention-based dense embedding models, e.g. traditional Sentence Transformer models like all-mpnet-base-v2, bge-base-en-v1.5, and gte-large-en-v1.5.
Static Embedding-based models, e.g. static-retrieval-mrl-en-v1, potion-base-8M, M2V_base_output, and glove.6B.300d.
Sparse bag-of-words model, BM25, often a strong baseline.

Click to expand BM25 implementation details

I relied on the highly efficient bm25s implementation, using model.get_scores() on tokens after tokenization and stemming with the English PyStemmer.

NOTE: Many of the attention-based dense embedding models are finetuned on the training splits of the (Nano)BEIR evaluation datasets. This gives the models an unfair advantage in this benchmark and can result in lower downstream performance on real retrieval tasks.

static-retrieval-mrl-en-v1 is purposefully not trained on any of these datasets.

GPU

CPU

We can draw some notable conclusions from these figures:

static-retrieval-mrl-en-v1 outperforms all other Static Embedding models.
static-retrieval-mrl-en-v1 is the only Static Embedding model to outperform BM25.
static-retrieval-mrl-en-v1 is
- 87.4% as performant as the commonly used all-mpnet-base-v2,
- 24x faster on GPU,
- 397x faster on CPU.
static-retrieval-mrl-en-v1 is quicker on CPU than on GPU: This model can run extraordinarily quickly everywhere, including consumer-grade PCs, tiny servers, phones, or in-browser.

Matryoshka Evaluations

We experimented with the results on NanoBEIR performance when we performed Matryoshka-style dimensionality reduction by truncating the output embeddings to a lower dimensionality.

Dimensionality	NanoBEIR_mean	NanoArguAna	NanoClimateFEVER	NanoDBPedia	NanoFEVER	NanoFiQA2018	NanoHotpotQA	NanoMSMARCO	NanoNFCorpus	NanoNQ	NanoQuoraRetrieval	NanoSCIDOCS	NanoSciFact	NanoTouche2020
1024	0.5031	0.4077	0.3308	0.5681	0.6921	0.3651	0.6547	0.4040	0.3241	0.4533	0.8950	0.2642	0.6111	0.5702
512	0.4957	0.3878	0.3360	0.5626	0.6945	0.3517	0.6280	0.3892	0.3206	0.4505	0.8986	0.2657	0.5953	0.5635
256	0.4819	0.3855	0.3203	0.5407	0.6734	0.3518	0.6027	0.4144	0.2860	0.4254	0.8948	0.2466	0.5620	0.5605
128	0.4622	0.4001	0.2982	0.5266	0.6273	0.3188	0.5606	0.4025	0.2693	0.4021	0.8930	0.2283	0.5447	0.5368
64	0.4176	0.3424	0.2809	0.5022	0.5480	0.2831	0.4680	0.3739	0.2153	0.3845	0.8525	0.1680	0.5045	0.5050
32	0.3532	0.2866	0.1870	0.4292	0.4193	0.2292	0.3602	0.3587	0.1444	0.3525	0.8325	0.1525	0.3983	0.4408

NanoBEIR performance vs Matryoshka dimensionality reduction

These findings show that reducing the dimensionality by e.g. 2x only has a 1.47% reduction in performance (0.5031 NDCG@10 vs 0.4957 NDCG@10), while realistically resulting in a 2x speedup in retrieval speed.

Training Details

Training Datasets

gooaq

Dataset: gooaq at b089f72
Size: 3,012,496 training samples
Columns: question and answer
Approximate statistics based on the first 1000 samples:
question answer
type string string
details
min: 18 characters
mean: 43.23 characters
max: 96 characters
min: 55 characters
mean: 253.36 characters
max: 371 characters

Samples:

question	answer
`what is the difference between broilers and layers?`	`An egg laying poultry is called egger or layer whereas broilers are reared for obtaining meat. So a layer should be able to produce more number of large sized eggs, without growing too much. On the other hand, a broiler should yield more meat and hence should be able to grow well.`
`what is the difference between chronological order and spatial order?`	`As a writer, you should always remember that unlike chronological order and the other organizational methods for data, spatial order does not take into account the time. Spatial order is primarily focused on the location. All it does is take into account the location of objects and not the time.`
`is kamagra same as viagra?`	`Kamagra is thought to contain the same active ingredient as Viagra, sildenafil citrate. In theory, it should work in much the same way as Viagra, taking about 45 minutes to take effect, and lasting for around 4-6 hours. However, this will vary from person to person.`

Loss: MatryoshkaLoss with these parameters:
json
1{ 2 "loss": "MultipleNegativesRankingLoss", 3 "matryoshka_dims": [ 4 1024, 5 512, 6 256, 7 128, 8 64, 9 32 10 ], 11 "matryoshka_weights": [ 12 1, 13 1, 14 1, 15 1, 16 1, 17 1 18 ], 19 "n_dims_per_step": -1 20}

msmarco

Dataset: msmarco at 84ed2d3
Size: 502,939 training samples
Columns: query, positive, and negative

Approximate statistics based on the first 1000 samples:

	query	positive	negative
type	string	string	string
details	min: 11 characters mean: 33.26 characters max: 197 characters	min: 96 characters mean: 356.24 characters max: 1006 characters	min: 68 characters mean: 327.52 characters max: 995 characters

Samples:

query	positive	negative
`when was the sullivan acts`	`Sullivan Act Tim Sullivan, a major Irish criminal passed the Sullivan Act in 1911 to help his constituents rob strangers or to help them against Italian incomers. That is the crux of story that goes with a very early gun control law.`	`Sullivan Act Tim Sullivan, a major Irish criminal passed the Sullivan Act in 1911 to help his constituents rob strangers or to help them against Italian incomers. That is the crux of story that goes with a very early gun control law.`
`can lavender grow indoors`	`Growing Lavender Indoors. People ALWAYS ask if you can grow lavender indoors. Well, you can, but most Lavender does best outside. Here is our winter experiment to show you what it would look like. This is one of our 4 Lavender Babies from Fall 2010. Our test specimen is L. x intermedia 'Grosso'.`	`Lavender can be grown indoors with a bit of effort to keep it in the conditions it loves to thrive. First off begin with choosing a variety that is better able to tolerate the conditions inside a home. To successfully grow Lavender indoors you need to create optimal growing conditions which is hard to do inside a house.`
`what kind of barley do you malt`	`Barley is a wonderfully versatile cereal grain with a rich nutlike flavor and an appealing chewy, pasta-like consistency. Its appearance resembles wheat berries, although it is slightly lighter in color. Sprouted barley is naturally high in maltose, a sugar that serves as the basis for both malt syrup sweetener.`	`Specialty grains that can be used in this way are usually barley, malted or unmalted, that has been treated differently at the malting company. Crystal malt is one of the specialty grains. It is available in a whole range of colors, from 20 to 120 Lovibond. Crystal malt is malted barley that is heated while wet.`

Loss: MatryoshkaLoss with these parameters:
json
1{ 2 "loss": "MultipleNegativesRankingLoss", 3 "matryoshka_dims": [ 4 1024, 5 512, 6 256, 7 128, 8 64, 9 32 10 ], 11 "matryoshka_weights": [ 12 1, 13 1, 14 1, 15 1, 16 1, 17 1 18 ], 19 "n_dims_per_step": -1 20}

squad

Dataset: squad at d84c8c2
Size: 87,599 training samples
Columns: question and answer
Approximate statistics based on the first 1000 samples:
question answer
type string string
details
min: 14 characters
mean: 59.66 characters
max: 150 characters
min: 156 characters
mean: 769.53 characters
max: 3706 characters

Samples:

question	answer
`What did Business Insider call San Diego in 2013?`	San Diego was ranked as the 20th-safest city in America in 2013 by Business Insider. According to Forbes magazine, San Diego was the ninth-safest city in the top 10 list of safest cities in the U.S. in 2010. Like most major cities, San Diego had a declining crime rate from 1990 to 2000. Crime in San Diego increased in the early 2000s. In 2004, San Diego had the sixth lowest crime rate of any U.S. city with over half a million residents. From 2002 to 2006, the crime rate overall dropped 0.8%, though not evenly by category. While violent crime decreased 12.4% during this period, property crime increased 1.1%. Total property crimes per 100,000 people were lower than the national average in 2008.
`What did the Spanish call this region?`	The name Montana comes from the Spanish word Montaña, meaning "mountain", or more broadly, "mountainous country". Montaña del Norte was the name given by early Spanish explorers to the entire mountainous region of the west. The name Montana was added to a bill by the United States House Committee on Territories, which was chaired at the time by Rep. James Ashley of Ohio, for the territory that would become Idaho Territory. The name was successfully changed by Representatives Henry Wilson (Massachusetts) and Benjamin F. Harding (Oregon), who complained that Montana had "no meaning". When Ashley presented a bill to establish a temporary government in 1864 for a new territory to be carved out of Idaho, he again chose Montana Territory. This time Rep. Samuel Cox, also of Ohio, objected to the name. Cox complained that the name was a misnomer given that most of the territory was not mountainous and that a Native American name would be more appropriate than a Spanish one. Other names such as...
`Small missiles were designed that could be mounted on what?`	As this process continued, the missile found itself being used for more and more of the roles formerly filled by guns. First to go were the large weapons, replaced by equally large missile systems of much higher performance. Smaller missiles soon followed, eventually becoming small enough to be mounted on armored cars and tank chassis. These started replacing, or at least supplanting, similar gun-based SPAAG systems in the 1960s, and by the 1990s had replaced almost all such systems in modern armies. Man-portable missiles, MANPADs as they are known today, were introduced in the 1960s and have supplanted or even replaced even the smallest guns in most advanced armies.

Loss: MatryoshkaLoss with these parameters:
json
1{ 2 "loss": "MultipleNegativesRankingLoss", 3 "matryoshka_dims": [ 4 1024, 5 512, 6 256, 7 128, 8 64, 9 32 10 ], 11 "matryoshka_weights": [ 12 1, 13 1, 14 1, 15 1, 16 1, 17 1 18 ], 19 "n_dims_per_step": -1 20}

s2orc

Dataset: s2orc at 8cfc394
Size: 90,000 training samples
Columns: title and abstract
Approximate statistics based on the first 1000 samples:
title abstract
type string string
details
min: 31 characters
mean: 80.02 characters
max: 185 characters
min: 84 characters
mean: 635.31 characters
max: 1023 characters

Samples:

title	abstract
`Modeling Method of Flow Diversion of the Three Outlets in Jingjiang Reach Under Unsteady Flow Conditions`	The Yangtze River Flood Protection Physical Model is built under the financial support of World Bank loan.Based on theoretical analysis and experimental study,a modeling method of flow diversion of the three outlets in Jingjiang Reach under unsteady flow conditions was established for the model.Validation tests under both steady and unsteady flow conditions manifested that with this modeling method,the experimental flow diversion proves to be consistent with that of the prototype and therefore meets the requirements for precision.Being validated,this modeling method has been applied to Yangtze River Flood Protection Physical Model to study the flood routing features in Jingjiang reach.
`Enlightening on medical administration by clinical governance in British`	`Medical quality and safety were the responsibilities of medical system in view of British clinical governance. Medical regulation institutes were considered to be built and be authorized regulation rights. British medical administration was introduced and its enlightening in China was mentioned.`
`APPLICATION OF A FUZZY MULTI-CRITERIA DECISION-MAKING MODEL FOR SHIPPING COMPANY PERFORMANCE EVALUATION`	Combining fuzzy set theory, Analytic Hierarchy Process (AHP) and concept of entropy, a fuzzy Multiple Criteria Decision-Making (MCDM) model for shipping company performance evaluation is proposed. First, the AHP is used to construct subjective weights for all criteria and sub-criteria. Then, linguistic values characterized by triangular fuzzy numbers and trapezoidal fuzzy numbers are used to denote the evaluation values of all alternatives with respect to various subjective and objective criteria. Finally, the aggregation fuzzy assessment of different shipping companies is ranked to determine the best selection. Utilizing this fuzzy MCDM model, the decision-maker's fuzzy assessment and the trade-off between various evaluations criteria can be taken into account in the aggregation process, thus ensuring more effective and accurate decision-making.

Loss: MatryoshkaLoss with these parameters:
json
1{ 2 "loss": "MultipleNegativesRankingLoss", 3 "matryoshka_dims": [ 4 1024, 5 512, 6 256, 7 128, 8 64, 9 32 10 ], 11 "matryoshka_weights": [ 12 1, 13 1, 14 1, 15 1, 16 1, 17 1 18 ], 19 "n_dims_per_step": -1 20}

allnli

Dataset: allnli at d482672
Size: 557,850 training samples
Columns: anchor, positive, and negative

Approximate statistics based on the first 1000 samples:

	anchor	positive	negative
type	string	string	string
details	min: 18 characters mean: 34.88 characters max: 193 characters	min: 15 characters mean: 46.49 characters max: 181 characters	min: 16 characters mean: 50.47 characters max: 204 characters

Samples:

anchor	positive	negative
`A person on a horse jumps over a broken down airplane.`	`A person is outdoors, on a horse.`	`A person is at a diner, ordering an omelette.`
`Children smiling and waving at camera`	`There are children present`	`The kids are frowning`
`A boy is jumping on skateboard in the middle of a red bridge.`	`The boy does a skateboarding trick.`	`The boy skates down the sidewalk.`

Loss: MatryoshkaLoss with these parameters:
json
1{ 2 "loss": "MultipleNegativesRankingLoss", 3 "matryoshka_dims": [ 4 1024, 5 512, 6 256, 7 128, 8 64, 9 32 10 ], 11 "matryoshka_weights": [ 12 1, 13 1, 14 1, 15 1, 16 1, 17 1 18 ], 19 "n_dims_per_step": -1 20}

paq

Dataset: paq at 74601d8
Size: 64,371,441 training samples
Columns: query and answer
Approximate statistics based on the first 1000 samples:
query answer
type string string
details
min: 25 characters
mean: 50.56 characters
max: 104 characters
min: 509 characters
mean: 620.96 characters
max: 773 characters

Samples:

query	answer
`in veetla visheshanga ganesh is the husband of`	Veetla Visheshanga a song which reminds Ganga's memory. She is actually not Ganga but Gowri and her lover is the groom named Ganesh. When both were about to marry they were stopped by some goons because of which Gowri fell from the mountain but survived with injuries. Gopal who found the truth brought Ganesh to unite them. Gopal insists Gowri to marry Ganesh as both of them are lovers to which Gowri unwillingly accepts. But while Ganesh tries to tie the Mangal Sutra, Gowri stops him and she goes to Gopal saying that he may not need her but she needs him
`when did simon property group became a publicly traded company`	of the S&P 100. Simon Property Group has been the subject of several lawsuits and investigations regarding civil rights and discrimination. Simon Property Group was formed in 1993 when the majority of the shopping center interests of Melvin Simon & Associates became a publicly traded company. Melvin Simon & Associates, owned by brothers Melvin Simon and Herbert Simon, was founded in 1960 in Indianapolis, Indiana, and had long been one of the top shopping center developers in the United States. In 1996, Simon DeBartolo Group was created when Simon Property merged with former rival DeBartolo Realty Corp. This was shortly
`what was the nationality of antoine faivre`	Theosophy (Boehmian) below. "Theosophy": The scholar of esotericism Wouter Hanegraaff described Christian theosophy as "one of the major currents in the history of Western esotericism". Christian theosophy is an under-researched area; a general history of it has never been written. The French scholar Antoine Faivre had a specific interest in the theosophers and illuminists of the eighteenth and nineteenth centuries. He wrote his doctoral thesis on Karl von Eckartshausen and Christian theosophy. Scholars of esotericism have argued that Faivre's definition of Western esotericism relies on his own specialist focus on Christian theosophy, Renaissance Hermeticism, and Romantic "Naturphilosophie" and therefore creates an "ideal"

Loss: MatryoshkaLoss with these parameters:
json
1{ 2 "loss": "MultipleNegativesRankingLoss", 3 "matryoshka_dims": [ 4 1024, 5 512, 6 256, 7 128, 8 64, 9 32 10 ], 11 "matryoshka_weights": [ 12 1, 13 1, 14 1, 15 1, 16 1, 17 1 18 ], 19 "n_dims_per_step": -1 20}

trivia

_qa

Dataset: trivia_qa at a7c36e3
Size: 73,346 training samples
Columns: query and answer
Approximate statistics based on the first 1000 samples:
query answer
type string string
details
min: 21 characters
mean: 76.91 characters
max: 455 characters
min: 136 characters
mean: 3273.89 characters
max: 4096 characters

Samples:

query	answer
`What type of rock is formed by the solidification of molten magma?`	igneous rock - Dictionary Definition : Vocabulary.com igneous rock n rock formed by the solidification of molten magma Types: a rare type of peridotite that sometimes contains diamonds; found in South Africa and Siberia Type of: material consisting of the aggregate of minerals like those making up the Earth's crust Word Family Usage Examples Sign up, it's free! Whether you're a student, an educator, or a life-long learner, Vocabulary.com can put you on the path to systematic vocabulary improvement.
`Which river flows through the town of Shrewsbury?`	`River Severn`
`Which band's name was inspired by a novel by Herman Hesse?`	`23 Band Names Inspired by Literature :: Books :: Lists :: Paste 23 Band Names Inspired by Literature By Wyndham Wyeth`

Loss: MatryoshkaLoss with these parameters:
json
1{ 2 "loss": "MultipleNegativesRankingLoss", 3 "matryoshka_dims": [ 4 1024, 5 512, 6 256, 7 128, 8 64, 9 32 10 ], 11 "matryoshka_weights": [ 12 1, 13 1, 14 1, 15 1, 16 1, 17 1 18 ], 19 "n_dims_per_step": -1 20}

msmarco_10m

Dataset: msmarco_10m at 8c5139a
Size: 10,000,000 training samples
Columns: query, positive, and negative

Approximate statistics based on the first 1000 samples:

	query	positive	negative
type	string	string	string
details	min: 10 characters mean: 33.47 characters max: 158 characters	min: 53 characters mean: 353.76 characters max: 948 characters	min: 67 characters mean: 343.74 characters max: 1063 characters

Samples:

query	positive	negative
`what is ged equivalent`	`A GED is equivalent to a high school diploma however colleges do not look at it the same. A huge part of high school is commitment and dedication, therefore if you choose to drop out of high school you will probably not get accepted into any top colleges. However, you could always start at a community college and work your way up. Good Luck`	`If you are not far along in your Army career, education is an especially good way to boost your chances of promotion. Continuing your education at any level can earn you points. For completing your GED or bachelor's degree, you earn 10 points.`
`foods that help with diverticulitis`	`Diet for Diverticulitis. Gradually you can ease back into a regular diet. Your doctor may advise you to start with low-fiber foods (white bread, meat, poultry, fish, eggs, and dairy products) before introducing high-fiber foods. Fiber softens and adds bulk to stools, helping them pass more easily through the colon.`	`During an attack of diverticulitis, your doctor may recommend a clear liquid diet or a low-fiber diet. This helps the area of infection to heal. Foods allowed on a clear-liquid diet include: 1 Plain water.iverticulitis occurs when small, bulging pouches (diverticula) in your colon become infected and inflamed â causing severe abdominal pain, nausea, and fever. The treatment of a diverticulitis attack will depend on the severity of the symptoms and whether this is your first attack.`
`calories burned in turbo kick class`	`The American Council on Exercise did a study to find out how many calories are burned during a turbo kick class. The study looked at 15 women in a turbo kick class who weighed about 135 lbs and found that they burned between 6.45 to 8.3 calories a minute. This comes out to about 350 to 450 calories an hour.`	`Popular Calories Burned Searches: 1 Calories Burned For Intervals of run/walk or walk/jog: 6 mph or slower (> 10 minutes per mile) 2 Calories Burned For Walking: 3.5 mph (17 minutes per mile) 3 Calories Burned For Walking: 4 mph (15 minutes per mile) Calories Burned For Walking: 4.6 mph (13 minutes per mile)`

Loss: MatryoshkaLoss with these parameters:
json
1{ 2 "loss": "MultipleNegativesRankingLoss", 3 "matryoshka_dims": [ 4 1024, 5 512, 6 256, 7 128, 8 64, 9 32 10 ], 11 "matryoshka_weights": [ 12 1, 13 1, 14 1, 15 1, 16 1, 17 1 18 ], 19 "n_dims_per_step": -1 20}

swim_ir

Dataset: swim_ir at 834c20f
Size: 501,538 training samples
Columns: query and text
Approximate statistics based on the first 1000 samples:
query text
type string string
details
min: 0 characters
mean: 59.98 characters
max: 189 characters
min: 208 characters
mean: 525.9 characters
max: 2743 characters

Samples:

query	text
`How many blocked kicks did Williams have in his second year at Bowling Green State University?`	`Williams accepted a football scholarship from Bowling Green State University, where he became one of the best special teams players in school history. As a redshirt freshman, he was a wide receiver on the scout team. The next year, he played mainly on special teams and had 3 blocked kicks.`
`How many town councils are there in the metropolitan borough?`	`Horwich, Westhoughton and Blackrod are now constituted as civil parishes. There are three town councils in the metropolitan borough, Westhoughton Town Council, Horwich Town Council and Blackrod Town Council. The rest of the metropolitan borough, Bolton, Farnworth, Kearsley, Little Lever, and South Turton, have remained unparished areas since 1974.`
`What is the name of the person selected to lead BART’s 296-member police force?`	In 2009, the hiring of two independent organizations reviewed BART's policies and procedures in the process of overseeing the BART Police. The two independent firms investigated the matters of BART Police Shooting of Oscar Grant and were charged with making recommendations to the board. Ward Allen formulated and chaired BART's first Police Department Review Committee, and as a result, BART made sweeping changes on many security measures, as well as corrected and implemented several policies and procedures. BPD Review Committee has led to the re-training of all officers on use of force, diversity re-training and other issues. Ward Allen hired Kenton Rainey, the person selected to lead BART’s 296-member police force, to take command as Chief of Police.

Loss: MatryoshkaLoss with these parameters:
json
1{ 2 "loss": "MultipleNegativesRankingLoss", 3 "matryoshka_dims": [ 4 1024, 5 512, 6 256, 7 128, 8 64, 9 32 10 ], 11 "matryoshka_weights": [ 12 1, 13 1, 14 1, 15 1, 16 1, 17 1 18 ], 19 "n_dims_per_step": -1 20}

pubmedqa

Dataset: pubmedqa at a1ef0b5
Size: 1,660 training samples
Columns: anchor, positive, negative_1, negative_2, negative_3, negative_4, negative_5, negative_6, negative_7, negative_8, negative_9, negative_10, negative_11, negative_12, negative_13, negative_14, negative_15, negative_16, negative_17, negative_18, negative_19, and negative_20

Approximate statistics based on the first 1000 samples:

	anchor	positive	negative_1	negative_2	negative_3	negative_4	negative_5	negative_6	negative_7	negative_8	negative_9	negative_10	negative_11	negative_12	negative_13	negative_14	negative_15	negative_16	negative_17	negative_18	negative_19	negative_20
type	string	string	string	string	string	string	string	string	string	string	string	string	string	string	string	string	string	string	string	string	string	string
details	min: 25 characters mean: 94.06 characters max: 213 characters	min: 5 characters mean: 409.42 characters max: 1582 characters	min: 5 characters mean: 325.57 characters max: 1300 characters	min: 17 characters mean: 299.5 characters max: 1352 characters	min: 21 characters mean: 317.37 characters max: 1590 characters	min: 13 characters mean: 334.43 characters max: 1536 characters	min: 13 characters mean: 335.49 characters max: 1247 characters	min: 13 characters mean: 336.16 characters max: 1383 characters	min: 14 characters mean: 319.98 characters max: 1501 characters	min: 16 characters mean: 337.33 characters max: 1493 characters	min: 15 characters mean: 324.82 characters max: 1058 characters	min: 13 characters mean: 336.76 characters max: 1457 characters	min: 10 characters mean: 355.4 characters max: 1748 characters	min: 13 characters mean: 344.26 characters max: 1705 characters	min: 16 characters mean: 335.11 characters max: 1593 characters	min: 18 characters mean: 353.83 characters max: 1374 characters	min: 13 characters mean: 328.01 characters max: 1755 characters	min: 12 characters mean: 337.74 characters max: 1579 characters	min: 16 characters mean: 336.94 characters max: 1325 characters	min: 15 characters mean: 319.49 characters max: 1410 characters	min: 12 characters mean: 340.91 characters max: 1680 characters	min: 20 characters mean: 330.34 characters max: 1509 characters

Samples:

anchor	positive	negative_1	negative_2	negative_3	negative_4	negative_5	negative_6	negative_7	negative_8	negative_9	negative_10	negative_11	negative_12	negative_13	negative_14	negative_15	negative_16	negative_17	negative_18	negative_19	negative_20
`Visceral adipose tissue area measurement at a single level: can it represent visceral adipose tissue volume?`	`Measurement of visceral adipose tissue (VAT) needs to be accurate and sensitive to change for risk monitoring. The purpose of this study is to determine the CT slice location where VAT area can best reflect changes in VAT volume and body weight.`	`46 patients with psoriasis and 46 sex- and age-matched control patients were included in this study. The abdominal fat area [visceral fat area (VFA), subcutaneous fat area (SFA) and total fat area (TFA)] at the level of the umbilicus was evaluated by computed tomography.`	`A retrospective review of CRC patients who received adjuvant chemotherapy at a single center during the period 2006-2009 identified from a prospectively maintained database. Visceral adiposity was determined by measuring visceral fat area (VFA) on preoperative staging CT. All patients were followed up to study completion or death.`	`A total of 1941 participants without known cardiovascular disease were enrolled from the Korean Genome and Epidemiology Study. Visceral fat area (VFA) was assessed by computed tomography. Appendicular skeletal muscle mass (ASM) was estimated by dual-energy X-ray absorptiometry and was used as a percentage of body weight (ASM/Wt). LV structure and function were assessed by tissue Doppler imaging (TDI) echocardiography.`	`One hundred and forty nonobese patients (BMI <25 kg/m2) were enrolled. EFV and visceral fat area were measured by MDCT. Patients were classified according to the plaque components (noncalcified, mixed and calcified) and severity of CAD. Inflammatory biomarkers were also measured, and compared with each CT parameter.`	`The blood gas level in each pulmonary vein (PV) was measured in supine subjects with diverse body mass index (BMI) values, to determine whether there was a regional insufficiency in gas exchange depending on the subject's BMI.`	Magnetic resonance imaging (MRI) of 163 patients with cholecystolithiasis and 163 non-cholecystolithiasis control subjects admitted to our institution between March 2011 and September 2013 were included in this cross-sectional evaluation. There were 98 women and 65 men in cholecystolithiasis group with an average age of 57±16 years (range 25-86 years). There were 87 women and 76 men in the control group with an average age of 41±16 years (range 14-77 years). Visceral adipose tissue (VAT), abdominal subcutaneous adipose tissue (SAT) and total abdominal adipose tissue (TAT) of all the subjects at navel level were measured on abdominal MRI. According to the visceral adipose area (cut-off point VAT = 100 cm2), study subjects were divided into 1) increased accumulation of intra-abdominal fat and 2) normal distribution of intra-abdominal fat. Logistic regression was used to assess the association of fat with the presence of cholecystolithiasis, adjusted for age and sex.	`We reviewed the medical records of 106 patients undergoing LA or LESS-A at our institution. Total fat area (TFA) and visceral fat area (VFA) were measured at the level of the L4 vertebra by computed tomography. To categorize the type of obesity, the VFA/TFA ratio was calculated. Multiple logistic regression analyses were performed to identify independent predictors of prolonged operative time.`	`The weight gain was 5.2% greater in rats exposed to fructose than in controls (P = 0.042). Total and visceral adipose tissue volumes were 5.2 cm3 (P = 0.017) and 3.1 cm3 (P = 0.019) greater, respectively, while lean tissue volumes did not differ. The level of triglycerides and apolipoprotein A-I was higher (P = 0.034, P = 0.005, respectively) in fructose-exposed rats.`	A total of 1593 middle-aged to older patients participated in this cross-sectional study. Brachial-to-ankle pulse wave velocity (baPWV) was measured as an index of arterial stiffness. Second PP (PP2) at the second peak of radial SBP was used to estimate central PP. Radial augmentation index was calculated as PP2/PP. Thigh muscle cross-sectional area and abdominal visceral fat area were quantified by computed tomography. Patients were classified as sarcopenic if their hand grip strength or skeletal muscle mass (measured by bioelectrical impedance) was more than 1 SD lower than the mean of those in a reference group aged below 50 years, or in the lowest 20% of the studied population. Visceral obesity was defined as visceral fat area greater than 100 cm.	`Visceral adiposity is linked with sleep-disordered breathing (SDB) (called Syndrome Z), and both correlate with coronary artery disease (CAD). The aim of the present study was to determine the significance of excess visceral fat, SDB and circulating levels of biomarkers in CAD in Japanese men.`	`There are no published studies on the impact of visceral adipose tissue (VAT) change on outcomes of restorative proctocolectomy and ileal pouch-anal anastomosis (IPAA). The aim of this historic cohort study was to evaluate the impact of excessive VAT gain on the outcomes of inflammatory bowel disease (IBD) patients with IPAA.`	`A subgroup of 46 men (n = 20, aged 29.1-33.4 years) and women (n = 26, aged 29.1-33.8 years) were recruited from an ongoing population study at our institution. Anthropometric variables including weight, height, and waist circumference were measured using standard procedures, and body mass index was calculated (kg/m(2)). Visceral adipose tissue (VAT) was measured with magnetic resonance imaging. Plasma apolipoproteins, lipids, glucose, and insulin were measured after an overnight fasting.`	`Our aim was to describe adipose tissue content and distribution in ALS patients.`	This was a cross-sectional study of 140 Japanese patients with type 2 diabetes (mean age 65 ± 11 year; 44.6% women). Visceral fat area (VFA; cm(2) ) and liver attenuation index (LAI) were assessed by abdominal computed tomography. The patients were divided into four groups by VFA and body mass index (BMI; kg/m(2) ) as follows: BMI <25 kg/m(2) and VFA <100 cm(2) (OB[-]VA[-]), BMI ≥25 kg/m(2) and VFA <100 cm(2) (OB[+]VA[-]), BMI <25 kg/m(2) and VFA ≥100 cm(2) (OB[-]VA[+]), and BMI ≥25 kg/m(2) and VFA ≥100 cm(2) (OB[+]VA[+]). Multivariate linear regression and logistic regression analysis were carried out to determine the impact of OB(-)VA(+) on LAI.	`Epicardial adipose tissue represents visceral adiposity, the early detection of which could be helpful for assessing subclinical target organ damage. Although previous studies have reported a relationship between epicardial fat thickness (EFT) and carotid intima-media thickness, there have been no studies detailing the relationship between EFT and brachial-ankle pulse wave velocity (baPWV).`	Higher adiponectin levels were associated with lower risk of diabetes (P < 0.001). Visceral fat was the only adiposity measure associated with diabetes after adjusting for BMI (odds ratio 3.0 [2.1-4.3] in women and 1.3 [1.0-1.6] in men, P < 0.001 between-sex comparison). Adipocytokines attenuated the association between visceral fat and diabetes for both sexes but more strongly in men (women 2.3 [1.5-3.3], men 1.1 [0.9-1.4]). In men, adiponectin, IL-6, and PAI-1 remained independently associated with diabetes after adjusting for fat depots; in women, adiponectin was the only independently associated adipocytokine. Controlling for insulin, HDL, triglycerides, and blood pressure did not change these results.	Ninety obesity patients and 95 non-obesity Uygur individuals were enrolled in this study. CD68 levels in abdominal subcutaneous and omental adipose tissues were detected by immunohistochemistry. The cytokine expression levels of adiponectin (APMI) and visfatin in serum were measured by enzyme-linked immunosorbent assay. Infection of 3T3-L1 cells with Ad36 was performed. Real-time PCR was performed to determine expression levels of APMI and Visfatin genes in the 3T3-L1 preadipocytes infected with Ad36.	`Serum CEA levels correlated with visceral fat area, fasting glucose, and triglyceride levels after adjusting for age and BMI. The mean visceral fat area increased significantly with the increasing CEA tirtiles. In a step-wise multiple regression analysis, age (β = 0.26, p<0.01) and visceral fat area (β = 0.19, p = 0.03) were identified as explanatory variables for serum CEA level.`	`The visceral adiposity index (VAI) has proved to be a marker of visceral adipose dysfunction, strongly associated with insulin sensitivity in both the general and specific populations of patients at metabolic risk.`	`Visceral adiposity is associated with hepatic steatosis, inflammation, and fibrosis in non-alcoholic fatty liver disease (NAFLD). The visceral adiposity index (VAI), a novel marker of visceral fat distribution and dysfunction, has been correlated with histology in hepatitis C. We assessed the ability of VAI to predict disease severity in NAFLD and hence its role as a non-invasive marker of liver damage.`
`Do general practitioner hospitals reduce the utilisation of general hospital beds?`	`Observational study comparing the total rates of admissions and of occupied bed days in general hospitals between populations with and without access to GP hospitals. Comparisons were also made separately for diagnoses commonly encountered in GP hospitals.`	`To study the prevalence of GERD comorbidities in a tertiary care hospital.`	`The population studied was a sample of 10% of the patients 65 years or older registered with a general practitioner contributing to the General Practice Research Database between 1988 and 1996.`	`Tertiary University Hospitals.`	`University hospital and district general hospital.`	`Inpatient rehabilitation facilities.`	`Inpatient rehabilitation facilities.`	`General medical service at a teaching hospital.`	`To examine the hypothesis that nursing homes responding to these changes in demand shifted the balance of resources from hotel to clinical activities.`	`Outpatient practices of general practitioners in the United Kingdom who contribute to the General Practice Research Database.`	`Hospital rehabilitation programs.`	`Surgical department of a large district general hospital.`	`Hospital-based case-control study.`	`Patients' homes.`	`the relationship between proximity to death and the amount of care provided by general practitioners (GPs) is largely unknown.`	`Public hospital, primary care clinic.`	`A teaching hospital and a district general hospital.`	`District General Hospital in the UK.`	`Academic general internal medicine practice.`	`A hospital-based study was conducted.`	`To evaluate the impact of participation in a trial on General Practitioners management and patient behaviour.`
`"Occult" posttraumatic lesions of the knee: can magnetic resonance substitute for diagnostic arthroscopy?`	We identified three types of occult post-traumatic injuries by morpho-topographic and signal intensity patterns: bone bruises (no. 25), subchondral (no. 33) and osteochondral (no. 35) injuries. Arthroscopy depicted 45 osteochondral and 19 chondral injuries. A bone bruise was defined as a typical subcortical area of signal loss, with various shapes, on T1-weighted images and of increased signal intensity on T2-weighted and FIR images. The cortical bone and articular cartilage were normal in all cases, while osteochondral injuries exhibited associated bone and cartilage damage with the same abnormal MR signal intensity. Sprain was the mechanism of injury in 52 cases, bruise in 12 and stress in 6. In 52 sprains (30 in valgus), the injury site was the lateral compartment in 92.3% of cases (100% in valgus), associated with meniscal damage in 73% of cases (90% in valgus) and with ligament injury in 90.4% (100% in valgus). In 12 bruises, the injury site was the lateral compartment in 58.3% of...	`Preoperative range of motion (ROM) has been regarded as one of the most important factors in predicting postoperative ROM following total knee arthroplasty (TKA). Mobile-bearing TKA designs have been suggested to possibly improve the knee kinematics compared to fixed-bearing designs. The purpose of this study was to examine the difference in postoperative flexion as a function of preoperative flexion in a consecutive series of TKAs done using a posterior-stabilized rotating-platform prosthesis.`	`To assess the association of underlying diagnosis with outcomes after revision total knee arthroplasty (TKA).`	`We identified 37 knees diagnosed with osteoarthritis with a preoperative knee flexion ≥120° but a 12-month postoperative range of motion (ROM) ≤110°. A random sample of 111 patients (1:3) from the same database, whose knees had a preoperative and 12-month postoperative ROM ≥120°, based on a diagnosis of primary osteoarthritis and no previous open knee surgery, were selected as the controls.`	`This study reports a series of patients operated on by anterior cruciate ligament (ACL) reconstruction combined with valgus high tibial osteotomy (HTO) for chronic anterior knee instability associated with medial tibiofemoral osteoarthritis. It was hypothesized that the combined surgery would enable return to sport, stabilize the knee and relieve medial pain.`	`To determine the prevalence and factors associated with knee osteoarthritis (OA) defined by magnetic resonance imaging (MRI) and specific OA features on MRI 1 year after anterior cruciate ligament reconstruction (ACLR).`	`In anterior ankle arthroscopy, the anterior working area (AWA) is restricted by the presence of the dorsalis pedis artery (DPA) and tendons. Pseudoaneurysms caused by iatrogenic damage to the DPA are difficult to identify intraoperatively. In knee arthroscopy, risk of popliteal artery damage is reduced in the flexed position [1]. This study investigates how DPA movement is affected by dorsiflexion and plantarflexion with the aim of identifying the positions providing the greatest AWA.`	`To investigate whether sex affects the trajectory of functional recovery after total knee arthroplasty (TKA).`	`To determine whether magnetic resonance imaging (MRI) evidence of tendinopathy in early rheumatoid arthritis (RA) could be used to predict the course of tendon involvement in later disease and specifically the risk of tendon rupture.`	With the advent of MRI (Magnetic Resonance Imaging), Synovial lesions around knee are being more and more easily detected. Synovial lesions of knee present with boggy swelling, effusion, pain, and restriction of motion. Differential diagnoses of such lesions include pigmented villonodular synovitis, synovial lipoma, synovial chondromatosis, rheumatoid arthritis, synovial hemangioma, amyloid arthropathy, xanthomata and lipoma arborescens. CT and MRI often help in diagnosis of such lesions. MRI of Lipoma Arborescens has been regarded to have characteristic diagnostic appearance - it includes a synovial mass with frond-like architecture and fat signal intensity on all pulse sequences. Sometimes Lipoma Arborescens can present in conjunction with inflammatory arthritis. Synovectomy is often curative for such conditions.	`Joint trauma can lead to a spectrum of acute lesions, including cartilage degradation, ligament or meniscus tears, and synovitis, all potentially associated with osteoarthritis (OA). This study was undertaken to generate and validate a murine model of knee joint trauma following noninvasive controlled injurious compression in vivo.`	`To determine which subregions of the knee joint have a high prevalence of pre-radiographic osteoarthritic changes, i.e., cartilage damage and osteophytes that can only be detected by magnetic resonance imaging (MRI), in radiographically normal knees.`	`78 of initially 84 patients (80 of 86 knees) were clinically and radiographically reassessed 5 (5.1-5.9) years after conventional, image-based, and image-free total knee arthroplasty. The methodology was identical to that used preoperatively and at 2 years, including the Knee Society score (KSS) and the functional score (FS), and AP and true lateral standard radiographs.`	`In adult patients with trauma, an increase in the thickness of the retropharyngeal soft tissues is commonly used as a potential indicator of occult injury, but no studies have examined this parameter using computed tomography (CT) as a screening modality.`	`To evaluate the thickness of cartilage at the posterior aspect of the medial and lateral condyle in Osteoarthritis (OA) knees compared to non-OA knees using computed tomography arthrography (CTA).`	`Total knee arthroplasty (TKA) successfully alleviates pain from knee osteoarthritis; but deficits in function can persist long term. Despite these well-known deficits, there is little evidence supporting the use of rehabilitation interventions following TKA.`	`The anterior intermeniscal ligament of the knee is at risk during knee arthroscopy, anterior cruciate ligament reconstruction, and tibial nail insertion.`	It has been previously demonstrated that radiographic severity of arthritis predicts outcome following knee replacement. In certain circumstances, patients may undergo arthroplasty without severe radiographic disease. An example may be the patient with significant chondral damage unsuccessfully treated with arthroscopy. This patient may proceed to joint replacement when their radiographs would not normally merit such intervention. We investigated whether these findings were also applicable to total ankle replacements (TARs).	`There is an increasing body of evidence that magnetic resonance imaging-occult tissue damage is an important component of primary progressive multiple sclerosis (PPMS) pathology. Proton magnetic resonance spectroscopy (1H-MRS) can be used to measure in vivo whole-brain N-acetylaspartate (WBNAA) concentrations, the decrease of whose levels is considered a marker of neuronal-axonal injury.`	`Meniscectomy and articular cartilage damage have been found to increase the prevalence of osteoarthritis after anterior cruciate ligament reconstruction, but the effect of knee range of motion has not been extensively studied.`	`Patients 50 years and older with knee osteoarthritis who underwent arthroscopy between 1998 and 2010 were retrospectively identified and an annual arthroscopy rate was calculated from 1998 through 2002 and from 2006 through 2010. Patients who underwent knee arthroplasty within 2 years of arthroscopy during each period were identified, and a 2-year conversion to arthroplasty rate was calculated.`

Loss: MatryoshkaLoss with these parameters:
json
1{ 2 "loss": "MultipleNegativesRankingLoss", 3 "matryoshka_dims": [ 4 1024, 5 512, 6 256, 7 128, 8 64, 9 32 10 ], 11 "matryoshka_weights": [ 12 1, 13 1, 14 1, 15 1, 16 1, 17 1 18 ], 19 "n_dims_per_step": -1 20}

miracl

Dataset: miracl at 07e2b62
Size: 789,900 training samples
Columns: anchor, positive, and negative

Approximate statistics based on the first 1000 samples:

	anchor	positive	negative
type	string	string	string
details	min: 11 characters mean: 39.23 characters max: 129 characters	min: 72 characters mean: 745.86 characters max: 4292 characters	min: 18 characters mean: 649.52 characters max: 3570 characters

Samples:

anchor	positive	negative
`Who created The Walking Dead comic books?`	Days Gone Bye (The Walking Dead)[object Object]Robert Kirkman, the creator of the eponymous series of comic books, considered the idea of creating a television show based on the comic series, but did not move forward. Frank Darabont expressed interest in developing the series for television. In January 2010, AMC formally announced that it had ordered a pilot for a possible series adapted from "The Walking Dead" comic book. In the announcement, the executives stated that Darabont would serve as writer, director, and an executive producer alongside Gale Anne Hurd.	`Living Dead[object Object]The Walking Dead`
`When was the first car invented?`	Car[object Object]In 1879, Benz was granted a patent for his first engine, which had been designed in 1878. Many of his other inventions made the use of the internal combustion engine feasible for powering a vehicle. His first "Motorwagen" was built in 1885 in Mannheim, Germany. He was awarded the patent for its invention as of his application on 29 January 1886 (under the auspices of his major company, Benz & Cie., which was founded in 1883). Benz began promotion of the vehicle on 3 July 1886, and about 25 Benz vehicles were sold between 1888 and 1893, when his first four-wheeler was introduced along with a model intended for affordability. They also were powered with four-stroke engines of his own design. Emile Roger of France, already producing Benz engines under license, now added the Benz car to his line of products. Because France was more open to the early cars, initially more were built and sold in France through Roger than Benz sold in Germany. In August 1888 Bertha Benz, the wife of Karl B...	Elwood Haynes[object Object]In 1905, three years after the Apperson brothers split from Haynes, Haynes-Apperson was renamed the Haynes Automobile Company and Haynes launched a series of publicity campaigns. A parade of 2,000 cars was organized in New York City during 1908 and Haynes, whom many recognized as the inventor of the American automobile, led the parade down Broadway riding in the "Pioneer". He was followed by ten Haynes cars, a model from each year to display the advancement in technology. On his way to the parade, Haynes was unaware of the city's newly established speeding laws and was arrested for driving too fast—in a car with a top speed of 15 mph (17 km/h)—and taken to jail. He was soon able to see a magistrate who released him after learning that he was Elwood Haynes and had come to lead the parade. The celebration was intended to be a ten-year commemoration of the invention of the automobile, although earlier self-vehicles dated back nearly twenty years in Europe. Haynes donated the...
`How many doctors are in Doctor Who?`	The Doctor (Doctor Who)[object Object]The Doctor is the title character in the long-running BBC science fiction television programme "Doctor Who". Since the show's inception in 1963, the character has been portrayed by thirteen lead actors. In the programme, "the Doctor" is the alias assumed by a centuries-old alien—a Time Lord from the planet Gallifrey—who travels through space and time in the TARDIS, frequently with companions. The transition to each succeeding actor is explained within the show's narrative through the plot device of "regeneration", a biological function of the Time Lord race that allows a change of cellular structure and appearance with recovery following a potentially fatal injury.	Sixth Doctor[object Object]The Sixth Doctor is an incarnation of the Doctor, the protagonist of the BBC science fiction television series "Doctor Who". He is portrayed by Colin Baker. Although his televisual time on the series was comparatively brief and turbulent, Baker has continued as the Sixth Doctor in Big Finish's range of original "Doctor Who" audio adventures. Within the series' narrative, the Doctor is a centuries-old Time Lord alien from the planet Gallifrey who travels in time and space in his TARDIS, frequently with companions. At the end of life, the Doctor can regenerate his body; in doing so, his physical appearance and personality change. Baker portrays the sixth such incarnation, an arrogant, flamboyant character in brightly coloured, mismatched clothes whose brash, often patronising personality set him apart from all his previous incarnations.

Loss: MatryoshkaLoss with these parameters:
json
1{ 2 "loss": "MultipleNegativesRankingLoss", 3 "matryoshka_dims": [ 4 1024, 5 512, 6 256, 7 128, 8 64, 9 32 10 ], 11 "matryoshka_weights": [ 12 1, 13 1, 14 1, 15 1, 16 1, 17 1 18 ], 19 "n_dims_per_step": -1 20}

mldr

Dataset: mldr at 40ad767
Size: 200,000 training samples
Columns: anchor, positive, and negative

Approximate statistics based on the first 1000 samples:

	anchor	positive	negative
type	string	string	string
details	min: 17 characters mean: 65.31 characters max: 210 characters	min: 2432 characters mean: 20354.29 characters max: 123500 characters	min: 3035 characters mean: 16236.77 characters max: 166364 characters

Samples:

anchor	positive	negative
`When was the art museum in Santa Barbara first opened to the public?`	The Santa Barbara Museum of Art (SBMA) is an art museum located in downtown Santa Barbara, California.[object Object][object Object]Founded in 1941, it is home to both permanent and special collections, the former of which includes Asian, American, and European art that spans 4,000 years from ancient to modern.[object Object][object Object]History[object Object][object Object]The Santa Barbara Museum of Art opened to the public on June 5, 1941, in a building that was at one time the Santa Barbara Post Office (1914–1932). The idea for an art museum first came from the local artist Colin Campbell Cooper when he learned that the post office was going to be sold. In a letter to the editor published in the Santa Barbara News-Press in July 1937, Cooper proposed that the impressive Italianate structure should be transformed into a museum. After gaining momentum in town and with the support of local businesses, politicians and art collectors the Santa Barbara Museum of Art was officially established just four years after Cooper's letter was published. The renowned Chicago arc...	The Knott's Berry Farm amusement park in Orange County, California, originated from a berry farm owned by Walter Knott (1889–1981). In the 1920s, Knott and his wife, Cordelia, sold berries, berry preserves and pies from a roadside stand beside State Route 39, near the small town of Buena Park.[object Object]In 1932, on a visit to Rudolph Boysen's farm in nearby Anaheim, Walter Knott was introduced to a new hybrid berry of a blackberry, a red raspberry, and a loganberry cross-bred by Boysen, who gave Walter his last six wilted berry-hybrid plants. Walter planted and cultivated them, then the family sold the berries at their roadside stand. When people asked what kind they were, he called them "boysenberries".[object Object][object Object]In 1934, to make ends meet, Knott's wife Cordelia (1890–1974) reluctantly began serving fried chicken dinners on their wedding china. For dessert, Knott's signature Boysenberry Pie was also served to guests dining in the small tea room. As Southern California developed, Highway 39 became the ma...
`What is the objective of Opération Chammal?`	Opération Chammal is a French military operation in Iraq and Syria in an attempt to contain the expansion of the Islamic State of Iraq and the Levant and to support the Iraqi Army. Its name comes from the Shamal (Chammal in French), a northwesterly wind that blows over Iraq and the Persian Gulf states.[object Object][object Object]Airstrikes over Iraq started 19 September 2014 and airstrikes over Syria started by the end of September 2015. The French operation is limited to airstrikes; French president François Hollande has reiterated that no ground troops would be used in the conflict. Additionally, the French frigate has joined the United States Navy's Commander Task Force 50 (CTF 50) as an escort.[object Object][object Object]On 14 November 2015, ISIL claimed that the attacks that took place in Paris the previous day were retaliation for Opération Chammal. In response, French forces increased their attacks against ISIL in Syria.[object Object][object Object]Background [object Object][object Object]On 10 June 2014, the terrorist group of the Islamic State of Iraq and the Levant and several ot...	CMA CGM S.A. is a French container transportation and shipping company. It is the world’s 3rd largest container shipping company, using 257 shipping routes between 420 ports in 160 different countries. Its headquarters are in Marseille, France and its North American headquarters are in Norfolk, Virginia, United States.[object Object][object Object]The name is an acronym of two predecessor companies, Compagnie Maritime d'Affrètement (CMA) and Compagnie Générale Maritime (CGM), which translate as "Maritime Freighting Company" and "General Maritime Company".[object Object][object Object]History[object Object][object Object]The history of CMA CGM can be traced back to the middle of the 19th century, when two major French shipping lines were created, respectively Messageries Maritimes (MM) in 1851 and Compagnie Générale Maritime (CGM) in 1855, soon renamed Compagnie Générale Transatlantique in 1861. Both companies were created partly with the backing of the French State, through the award of mail contracts to various destinations, French colonies and overseas territories a...
`What was the reason for Williams-Franklin's decision to become a vegan during her time as a player?`	Taj McWilliams-Franklin (born October 20, 1970) is a former American professional women's basketball player.[object Object][object Object]A two-time WNBA champion with the Detroit Shock and Minnesota Lynx and six-time all-star, McWilliams-Franklin's professional career has spanned three decades, and began before the WNBA was founded. She retired from the WNBA after the 2012 season.[object Object][object Object]College years[object Object]After attending T. W. Josey High School in Augusta, Georgia, McWilliams-Franklin attended Georgia State University in 1989 and played on the school's basketball team for one season. However, she had become pregnant during her senior year in high school, and after the coach who recruited her to Georgia State was let go, the incoming staff told her "school was no place for kids." McWilliams-Franklin moved to Austin, Texas, where a friend connected her with St. Edward's University coach Dave McKey. She enrolled at St. Edwards as a Rhetoric major.[object Object][object Object]While at St. Edward's, she set school records and individual achievements, in...	Spider-Woman (Gwendolyne Maxine Stacy; colloquial: "Spider-Gwen" or "Ghost-Spider") is a superhero appearing in American comic books published by Marvel Comics. She was created by Jason Latour and Robbi Rodriguez. The character debuted in Edge of Spider-Verse issue #2 as part of the 2014–15 "Spider-Verse" comic book storyline, leading to the ongoing series Spider-Gwen that began in 2015.[object Object][object Object]Spider-Woman is a variant of Spider-Man and an alternate-universe version of Gwen Stacy. She lives on Earth-65, where Gwen Stacy is bitten by a radioactive spider and becomes a superhero instead of Peter Parker becoming Spider-Man. The character's various enemies include Earth-65 versions of Matt Murdock and Frank Castle. Gwen Stacy's Spider-Woman harbors much of Peter's personality and conflicts along with his powers and abilities.[object Object][object Object]Spider-Woman was met with positive reviews from critics, with them applauding her design—cited as a popular choice for cosplay—and a feminist perspective. For promotion, ...

Loss: MatryoshkaLoss with these parameters:
json
1{ 2 "loss": "MultipleNegativesRankingLoss", 3 "matryoshka_dims": [ 4 1024, 5 512, 6 256, 7 128, 8 64, 9 32 10 ], 11 "matryoshka_weights": [ 12 1, 13 1, 14 1, 15 1, 16 1, 17 1 18 ], 19 "n_dims_per_step": -1 20}

mr_tydi

Dataset: mr_tydi at abbdf55
Size: 354,700 training samples
Columns: anchor, positive, and negative

Approximate statistics based on the first 1000 samples:

	anchor	positive	negative
type	string	string	string
details	min: 12 characters mean: 38.85 characters max: 95 characters	min: 64 characters mean: 645.85 characters max: 4067 characters	min: 21 characters mean: 626.85 characters max: 2870 characters

Samples:

anchor	positive	negative
`Is amnesia real?`	Amnesia[object Object]Amnesia is a deficit in memory caused by brain damage, disease, or psychological trauma.[1] Amnesia can also be caused temporarily by the use of various sedatives and hypnotic drugs. The memory can be either wholly or partially lost due to the extent of damage that was caused.[2] There are two main types of amnesia: retrograde amnesia and anterograde amnesia. Retrograde amnesia is the inability to retrieve information that was acquired before a particular date, usually the date of an accident or operation.[3] In some cases the memory loss can extend back decades, while in others the person may lose only a few months of memory. Anterograde amnesia is the inability to transfer new information from the short-term store into the long-term store. People with this type of amnesia cannot remember things for long periods of time. These two types are not mutually exclusive; both can occur simultaneously.	Amnesia[object Object]Head trauma is a very broad range as it deals with any kind of injury or active action toward the brain which might cause amnesia. Retrograde and anterograde amnesia is more often seen from events like this, an exact example of a cause of the two would be electroconvulsive therapy, which would cause both briefly for the receiving patient. Traumatic events are more subjective. What is traumatic is dependent on what the person finds to be traumatic. Regardless, a traumatic event is an event where something so distressing occurs that the mind chooses to forget rather than deal with the stress. A common example of amnesia that is caused by traumatic events is dissociative amnesia, which occurs when the person forgets an event that has deeply disturbed them.[8] An example would be a person forgetting a fatal and graphic car accident involving their loved ones. Physical deficiencies are different from head trauma because physical deficiencies lean more toward passive physical issues.
`What is the largest naval base in the world?`	Naval Station Norfolk[object Object]Naval Station Norfolk, is a United States Navy base in Norfolk, Virginia. It supports naval forces in the United States Fleet Forces Command,[1] those operating in the Atlantic Ocean, Mediterranean Sea, and the Indian Ocean. The installation occupies about 4 miles (6.4km) of waterfront space and 11 miles (18km) of pier and wharf space of the Hampton Roads peninsula known as Sewell's Point. It is the world's largest naval station, with the largest concentration of U.S. Navy forces through 75 ships alongside 14 piers and with 134 aircraft and 11 aircraft hangars at the adjacently operated Chambers Field and [2] Port Services controls more than 3,100 ships' movements annually as they arrive and depart their berths.	`Clark Air Base[object Object]Clark Air Base was arguably the most urbanized military facility in history and was the largest American base overseas. At its peak around 1990, it had a permanent population of 15,000. It had a base exchange, a large commissary, a small shopping arcade, a branch department store, cafeterias, teen centers, a hotel, miniature golf, riding stables, zoo, and other concessions.`
`What is the power of the Red Lantern?`	Red Lantern Corps[object Object]In Final Crisis: Rage of the Red Lanterns, Atrocitus is shown in a flashback as having apparently formed a central power battery by using the blood of the other Inversions in blood magic rituals. The battery stands before a great lake of blood from which he forms his red power ring (crystallized by his anger), as well as other rings and batteries used to form the Red Lantern Corps. Harnessing the red light of rage, he sends his rings out into the universe; however, upon accepting the rings, his recruits' hearts are rendered useless. Their blood spoils from within, forcing them to expel the violently flammable and corrosive material from their mouths. Additionally, the Red Lanterns are reduced to an almost animalistic state, with only Atrocitus appearing to be in full control of himself. Once Atrocitus assembles a sufficient force, he leads them on a mission to capture Sinestro (who is being transferred to Korugar for his execution). Coincidentally, the Sinestro Corps ...	Green Lantern in other media[object Object]John Stewart is a member of the Justice League in the "Justice League" animated series. In this series, Stewart's ring was initially constrained to permitting him to fly, generating a protective force field, creating walls, and firing energy blasts; this limitation was established as being due to Stewart's mindset, not an inherent limitation of the ring itself (the series' version of Stewart is a former U.S. Marine, not an architect). After being berated by Katma Tui for his unimaginative use of the ring, Stewart has learned to generate complex tools (to defuse a bomb in one instance) and weapons. (He was also shown to be more creative when transformed into a child in the episode "Kids Stuff".) In a development not seen in any other version of the Green Lantern mythos, Stewart's eyes glow green when wearing his charged power ring. The glow fades when the ring runs out of power. The series has been inconsistent about the ring's effectiveness against yellow; ...

Loss: MatryoshkaLoss with these parameters:
json
1{ 2 "loss": "MultipleNegativesRankingLoss", 3 "matryoshka_dims": [ 4 1024, 5 512, 6 256, 7 128, 8 64, 9 32 10 ], 11 "matryoshka_weights": [ 12 1, 13 1, 14 1, 15 1, 16 1, 17 1 18 ], 19 "n_dims_per_step": -1 20}

Evaluation Datasets

gooaq

Dataset: gooaq at b089f72
Size: 3,012,496 evaluation samples
Columns: question and answer
Approximate statistics based on the first 1000 samples:
question answer
type string string
details
min: 18 characters
mean: 43.17 characters
max: 98 characters
min: 51 characters
mean: 254.12 characters
max: 360 characters

Samples:

question	answer
`how do i program my directv remote with my tv?`	`['Press MENU on your remote.', 'Select Settings & Help > Settings > Remote Control > Program Remote.', 'Choose the device (TV, audio, DVD) you wish to program. ... ', 'Follow the on-screen prompts to complete programming.']`
`are rodrigues fruit bats nocturnal?`	`Before its numbers were threatened by habitat destruction, storms, and hunting, some of those groups could number 500 or more members. Sunrise, sunset. Rodrigues fruit bats are most active at dawn, at dusk, and at night.`
`why does your heart rate increase during exercise bbc bitesize?`	`During exercise there is an increase in physical activity and muscle cells respire more than they do when the body is at rest. The heart rate increases during exercise. The rate and depth of breathing increases - this makes sure that more oxygen is absorbed into the blood, and more carbon dioxide is removed from it.`

Loss: MatryoshkaLoss with these parameters:
json
1{ 2 "loss": "MultipleNegativesRankingLoss", 3 "matryoshka_dims": [ 4 1024, 5 512, 6 256, 7 128, 8 64, 9 32 10 ], 11 "matryoshka_weights": [ 12 1, 13 1, 14 1, 15 1, 16 1, 17 1 18 ], 19 "n_dims_per_step": -1 20}

msmarco

Dataset: msmarco at 84ed2d3
Size: 502,939 evaluation samples
Columns: query, positive, and negative

Approximate statistics based on the first 1000 samples:

	query	positive	negative
type	string	string	string
details	min: 10 characters mean: 33.36 characters max: 137 characters	min: 67 characters mean: 347.87 characters max: 906 characters	min: 57 characters mean: 318.18 characters max: 906 characters

Samples:

query	positive	negative
`is cabinet refacing worth the cost?`	`Fans of refacing say this mini-makeover can give a kitchen a whole new look at a much lower cost than installing all-new cabinets. Cabinet refacing can save up to 50 percent compared to the cost of replacing, says Cheryl Catalano, owner of Kitchen Solvers, a cabinet refacing franchise in Napierville, Illinois. From.`	`Most cabinet refacing projects cost about $4,000 to $10,000. The price varies based on the materials you select and the size and configuration of your kitchen. Wood veneer doors, for example, will cost less than solid wood doors.`
`is the fovea ethmoidalis a bone`	`Ethmoid bone/fovea ethmoidalis. The medial portion of the ethmoid bone is a cruciate membranous bone composed of the crista galli, cribriform plate, and perpendicular ethmoidal plate. The crista is a thick piece of bone, shaped like a âcock's comb,â that projects intracranially and attaches to the falx cerebri.`	`Ethmoid bone/fovea ethmoidalis. The medial portion of the ethmoid bone is a cruciate membranous bone composed of the crista galli, cribriform plate, and perpendicular ethmoidal plate. The crista is a thick piece of bone, shaped like a âcock's comb,â that projects intracranially and attaches to the falx cerebri.`
`average pitches per inning`	`The likelihood of a pitcher completing nine innings if he throws an average of 14 pitches or less per inning is reinforced by the totals of the 89 games in which pitchers did actually complete nine innings of work.`	`The likelihood of a pitcher completing nine innings if he throws an average of 14 pitches or less per inning is reinforced by the totals of the 89 games in which pitchers did actually complete nine innings of work.`

Loss: MatryoshkaLoss with these parameters:
json
1{ 2 "loss": "MultipleNegativesRankingLoss", 3 "matryoshka_dims": [ 4 1024, 5 512, 6 256, 7 128, 8 64, 9 32 10 ], 11 "matryoshka_weights": [ 12 1, 13 1, 14 1, 15 1, 16 1, 17 1 18 ], 19 "n_dims_per_step": -1 20}

squad

Dataset: squad at d84c8c2
Size: 87,599 evaluation samples
Columns: question and answer
Approximate statistics based on the first 1000 samples:
question answer
type string string
details
min: 1 characters
mean: 60.25 characters
max: 161 characters
min: 152 characters
mean: 761.88 characters
max: 2525 characters

Samples:

question	answer
`When did the Russian Empire begin to question the existence of the Ottoman Empire?`	`In 1853 the Russian Empire on behalf of the Slavic Balkan states began to question the very existence of the Ottoman Empire. The result was the Crimean War, 1853–1856, in which the British Empire and the French Empire supported the Ottoman Empire in its struggle against the incursions of the Russian Empire. Eventually, the Ottoman Empire lost control of the Balkan region.`
`How would one describe the control of universities before nation-states in the 17th century?`	The propagation of universities was not necessarily a steady progression, as the 17th century was rife with events that adversely affected university expansion. Many wars, and especially the Thirty Years' War, disrupted the university landscape throughout Europe at different times. War, plague, famine, regicide, and changes in religious power and structure often adversely affected the societies that provided support for universities. Internal strife within the universities themselves, such as student brawling and absentee professors, acted to destabilize these institutions as well. Universities were also reluctant to give up older curricula, and the continued reliance on the works of Aristotle defied contemporary advancements in science and the arts. This era was also affected by the rise of the nation-state. As universities increasingly came under state control, or formed under the auspices of the state, the faculty governance model (begun by the University of Paris) became more and m...
`When did Jewish law recognize copyright?`	The concept's origins can potentially be traced back further. Jewish law includes several considerations whose effects are similar to those of modern intellectual property laws, though the notion of intellectual creations as property does not seem to exist – notably the principle of Hasagat Ge'vul (unfair encroachment) was used to justify limited-term publisher (but not author) copyright in the 16th century. In 500 BCE, the government of the Greek state of Sybaris offered one year's patent "to all who should discover any new refinement in luxury".

Loss: MatryoshkaLoss with these parameters:
json
1{ 2 "loss": "MultipleNegativesRankingLoss", 3 "matryoshka_dims": [ 4 1024, 5 512, 6 256, 7 128, 8 64, 9 32 10 ], 11 "matryoshka_weights": [ 12 1, 13 1, 14 1, 15 1, 16 1, 17 1 18 ], 19 "n_dims_per_step": -1 20}

s2orc

Dataset: s2orc at 8cfc394
Size: 10,000 evaluation samples
Columns: title and abstract
Approximate statistics based on the first 1000 samples:
title abstract
type string string
details
min: 31 characters
mean: 80.04 characters
max: 198 characters
min: 96 characters
mean: 653.93 characters
max: 1023 characters

Samples:

title	abstract
`Screen Printing Ink Film Thickness Analysis of the Passive RFID Tag Antenna`	The relationship between the screen mesh and the theoretical and practical ink film thickness was analyzed based on the main influencing factors of the ink film thickness by screen printing.A calculation model for the ink thickness was established based on the screen under static and compressive deformation.The relation curve between the screen mesh and the ink film thickness was fitted and the suitable printing craft parameter was chosen to print two kinds of RFID tag antennas.The fluctuation of the antenna resistance was analyzed to demonstrate the reliability of the passive RFID tag antenna manufactured by screen printing technology.
`Subclinical organ damage and cardiovascular risk prediction`	AbstractTraditional cardiovascular risk factors have poor prognostic value for individuals and screening for subclinical organ damage has been recommended in hypertension in recent guidelines. The aim of this review was to investigate the clinical impact of the additive prognostic information provided by measuring subclinical organ damage. We have (i) reviewed recent studies linking markers of subclinical organ damage in the heart, blood vessels and kidney to cardiovascular risk; (ii) discussed the evidence for improvement in cardiovascular risk prediction using markers of subclinical organ damage; (iii) investigated which and how many markers to measure and (iv) finally discussed whether measuring subclinical organ damage provided benefits beyond risk prediction. In conclusion, more studies and if possible randomized studies are needed to investigate (i) the importance of markers of subclinical organ damage for risk discrimination, calibration and reclassification; and (ii) the econom...
`A Novel Approach to Simulate Climate Change Impacts on Vascular Epiphytes: Case Study in Taiwan`	In the wet tropics, epiphytes form a conspicuous layer in the forest canopy, support abundant coexisting biota, and are known to have a critical influence on forest hydrology and nutrient cycling. Since canopy-dwelling plants have no vascular connection to the ground or their host plants, they are likely more sensitive to environmental changes than their soil-rooted counterparts, subsequently regarded as one of the groups most vulnerable to global climate change. Epiphytes have adapted to life in highly dynamic forest canopies by producing many, mostly wind-dispersed, seeds or spores. Consequently, epiphytes should colonize trees rapidly, which, in addition to atmospheric sensitivity and short life cycles, make epiphytes suitable climate change indicators. In this study, we assess the impact of climate change on Taiwanese epiphytes using a modeling approach.

Loss: MatryoshkaLoss with these parameters:
json
1{ 2 "loss": "MultipleNegativesRankingLoss", 3 "matryoshka_dims": [ 4 1024, 5 512, 6 256, 7 128, 8 64, 9 32 10 ], 11 "matryoshka_weights": [ 12 1, 13 1, 14 1, 15 1, 16 1, 17 1 18 ], 19 "n_dims_per_step": -1 20}

allnli

Dataset: allnli at d482672
Size: 6,584 evaluation samples
Columns: anchor, positive, and negative

Approximate statistics based on the first 1000 samples:

	anchor	positive	negative
type	string	string	string
details	min: 15 characters mean: 72.82 characters max: 300 characters	min: 12 characters mean: 34.11 characters max: 126 characters	min: 11 characters mean: 36.38 characters max: 121 characters

Samples:

anchor	positive	negative
`Two women are embracing while holding to go packages.`	`Two woman are holding packages.`	`The men are fighting outside a deli.`
`Two young children in blue jerseys, one with the number 9 and one with the number 2 are standing on wooden steps in a bathroom and washing their hands in a sink.`	`Two kids in numbered jerseys wash their hands.`	`Two kids in jackets walk to school.`
`A man selling donuts to a customer during a world exhibition event held in the city of Angeles`	`A man selling donuts to a customer.`	`A woman drinks her coffee in a small cafe.`

Loss: MatryoshkaLoss with these parameters:
json
1{ 2 "loss": "MultipleNegativesRankingLoss", 3 "matryoshka_dims": [ 4 1024, 5 512, 6 256, 7 128, 8 64, 9 32 10 ], 11 "matryoshka_weights": [ 12 1, 13 1, 14 1, 15 1, 16 1, 17 1 18 ], 19 "n_dims_per_step": -1 20}

paq

Dataset: paq at 74601d8
Size: 64,371,441 evaluation samples
Columns: query and answer
Approximate statistics based on the first 1000 samples:
query answer
type string string
details
min: 25 characters
mean: 51.3 characters
max: 108 characters
min: 504 characters
mean: 623.09 characters
max: 835 characters

Samples:

query	answer
`when did season 3 of the voice brasil start`	The Voice Brasil (season 3) The third season of "The Voice Brasil", premiered on Rede Globo on September 18, 2014 in the 10:30 p.m. (BRT/AMT) slot immediately following the primetime telenovela "Império". The 22- and 24-year-old sertanejo duo Danilo Reis e Rafael won the competition on December 25, 2014 with 43% of the votes cast. This marked Lulu Santos' first win as a coach, the first stolen artist to win a Brazilian season of "The Voice", and the first time in any "The Voice" franchise that a duo won the competition. Online applications for "The Voice Brasil" were open on
`when did the little ranger first come out`	Gang" theme song was an instrumental medley of "London Bridge", "Here We Go Round the Mulberry Bush" and "The Farmer in the Dell". It remained in use until the series ended in 1944. The Little Ranger The Little Ranger is a 1938 "Our Gang" short comedy film directed by Gordon Douglas. It was the 169th short in the "Our Gang" series, and the first produced by Metro-Goldwyn-Mayer, who purchased the rights to the series from creator Hal Roach. Snubbed by his girlfriend Darla, Alfalfa accepts the invitation of tomboyish Muggsy to attend the local picture show. While watching the adventures
`what is the name of rachel's sister in ninjaaiden`	her among ten female characters who have never been featured on their games' cover arts, Samir Torres of VentureBeat wrote that while "Team Ninja sexualy exploits all of their female characters, yet Rachel somehow got axed from every modern "Ninja Gaiden" box art." Rachel (Ninja Gaiden) In 2004's "Ninja Gaiden", Rachel is a fiend hunter whom the game's protagonist Ryu Hayabusa meets in the Holy Vigoor Empire, where she is on a mission to destroy the fiends, as well as find her missing sister, Alma, who has become a Greater Fiend. Soon after they first meet, she is captured but

Loss: MatryoshkaLoss with these parameters:
json
1{ 2 "loss": "MultipleNegativesRankingLoss", 3 "matryoshka_dims": [ 4 1024, 5 512, 6 256, 7 128, 8 64, 9 32 10 ], 11 "matryoshka_weights": [ 12 1, 13 1, 14 1, 15 1, 16 1, 17 1 18 ], 19 "n_dims_per_step": -1 20}

trivia_qa

Dataset: trivia_qa at a7c36e3
Size: 73,346 evaluation samples
Columns: query and answer
Approximate statistics based on the first 1000 samples:
query answer
type string string
details
min: 26 characters
mean: 77.62 characters
max: 258 characters
min: 135 characters
mean: 3169.71 characters
max: 4096 characters

Samples:

query	answer
`In which country is 'Ninety Mile Beach'?`	Ninety (90) Mile Beach, Gippsland, Victoria - Tourism Australia Gippsland Find travel information on Ninety Mile Beach, one of the longest uninterrupted beaches in the world, located outside of Melbourne at Gippsland Lakes. Ninety Mile Beach, located in the Gippsland region on Victoria's south-eastern coastline, is one of the longest uninterrupted beaches in the world. Stand on the beach and watch the beach disappear into the salty sea spray in the distance. You might find that your footprints are the only ones in the sand that day. This is one of the most natural and unspoilt beaches in the world and is ideal for activities from beach fishing and swimming to walking, whale and dolphin-spotting or just lazing in the sun. Sun, sand and lush national parks all create the perfect holiday environment. Victoria's Ninety Mile Beach is a 90-mile long stretch of pristine golden sand that separates the Gippsland Lakes from Bass Strait. Stretching as far as the eye can see it is one of the most ...
`What country gets nearly 75% of its electricity from nuclear power?`	`Nuclear Power in France`
`Which Spaniard led an expedition which reached Tenochtitlan, the Aztec capital in 1519?`	`The Spanish Conquest (1519-1521) : Mexico History History`

Loss: MatryoshkaLoss with these parameters:
json
1{ 2 "loss": "MultipleNegativesRankingLoss", 3 "matryoshka_dims": [ 4 1024, 5 512, 6 256, 7 128, 8 64, 9 32 10 ], 11 "matryoshka_weights": [ 12 1, 13 1, 14 1, 15 1, 16 1, 17 1 18 ], 19 "n_dims_per_step": -1 20}

msmarco_10m

Dataset: msmarco_10m at 8c5139a
Size: 10,000,000 evaluation samples
Columns: query, positive, and negative

Approximate statistics based on the first 1000 samples:

	query	positive	negative
type	string	string	string
details	min: 10 characters mean: 33.98 characters max: 131 characters	min: 56 characters mean: 353.39 characters max: 1029 characters	min: 85 characters mean: 339.79 characters max: 983 characters

Samples:

query	positive	negative
`what does a dental hygienist do`	`During a dental appointment, a hygienist typically removes soft and hard deposits from a patient's teeth; examines the gums and teeth to discern the presence of disease or oral abnormality; and strips the teeth of calculus (tartar), stains and plaque. dental hygienist takes on a somewhat academic role as well; he or she educates dental patients on how to establish and maintain suitable oral hygiene, often with the aid of teeth models to give the patient a visual sense.`	`Michelly in Tennessee said: Can anyone tell me how much you get paid hourly working as a dental hygienist. I make $36 dollars an hour and I just graduated from school. Should I be making more. Try to visit this link [object Object] so you will know how much dental hygienist usually get based on experience.`
`average annual temperature by florida county`	`Lake County Weather. The average temperature of Lake County is 70.90Â°F, which is about the same as the Florida average temperature of 71.80Â°F and is much higher than the national average temperature of 54.45Â°F. Historical Weather. Heating Cost Index, #29.`	`average rn salary in fl the average annual salary for a registered nurse in the state of florida in 2011 was $ 64020 the average does fluctuate throughout the state with median rn salaries at their highest in metro areas`
`what does amortization mean`	`What is 'Amortization'. Amortization is the paying off of debt with a fixed repayment schedule in regular installments over a period of time for example with a mortgage or a car loan.`	`First off, your EBIT is the same as your operating profit, but you can also calculate it by subtracting interest and tax from net income: $100,000 / ($10,000 + $25,000) = $65,000 To get EBITDA, you need to add back in depreciation and amortization:`

Loss: MatryoshkaLoss with these parameters:
json
1{ 2 "loss": "MultipleNegativesRankingLoss", 3 "matryoshka_dims": [ 4 1024, 5 512, 6 256, 7 128, 8 64, 9 32 10 ], 11 "matryoshka_weights": [ 12 1, 13 1, 14 1, 15 1, 16 1, 17 1 18 ], 19 "n_dims_per_step": -1 20}

swim_ir

Dataset: swim_ir at 834c20f
Size: 501,538 evaluation samples
Columns: query and text
Approximate statistics based on the first 1000 samples:
query text
type string string
details
min: 4 characters
mean: 59.74 characters
max: 165 characters
min: 206 characters
mean: 522.53 characters
max: 3079 characters

Samples:

query	text
`Where was he born?`	`He was born in Brownsville, Edmonson County, Kentucky, March 28, 1890; attended the public schools, Western Kentucky State Teachers College at Bowling Green, and the law department of the University of Kentucky at Lexington; was admitted to the bar in 1915 and commenced practice in Brownsville, Ky.; county judge of Edmonson County, Ky., 1916-1918.`
`What was Channon's National Service?`	`Channon completed his National Service in the Royal Horse Guards (the Blues) from 1955 to 1956, serving in Cyprus during the 1956 Cyprus emergency. In London, he was a member of the set around Princess Margaret, and then attended Christ Church, Oxford from 1956. He was president of the Oxford University Conservative Association.`
`What is the role of Immunoglobulin A in the immune system?`	Immunoglobulin heavy constant alpha 1 is a immunoglobulin gene with symbol "IGHA1". It encodes a constant (C) segment of Immunoglobulin A heavy chain. Immunoglobulin A is an antibody that plays a critical role in immune function in the mucous membranes. IgA shows the same typical structure of other antibody classes, with two heavy chains and two light chains, and four distinct domains: one variable region, and three variable regions. As a major class of immunoglobulin in body secretions, IgA plays a role in defending against infection, as well as preventing the access of foreign antigens to the immunologic system.

Loss: MatryoshkaLoss with these parameters:
json
1{ 2 "loss": "MultipleNegativesRankingLoss", 3 "matryoshka_dims": [ 4 1024, 5 512, 6 256, 7 128, 8 64, 9 32 10 ], 11 "matryoshka_weights": [ 12 1, 13 1, 14 1, 15 1, 16 1, 17 1 18 ], 19 "n_dims_per_step": -1 20}

pubmedqa

Dataset: pubmedqa at a1ef0b5
Size: 1,660 evaluation samples
Columns: anchor, positive, negative_1, negative_2, negative_3, negative_4, negative_5, negative_6, negative_7, negative_8, negative_9, negative_10, negative_11, negative_12, negative_13, negative_14, negative_15, negative_16, negative_17, negative_18, negative_19, and negative_20

Approximate statistics based on the first 1000 samples:

	anchor	positive	negative_1	negative_2	negative_3	negative_4	negative_5	negative_6	negative_7	negative_8	negative_9	negative_10	negative_11	negative_12	negative_13	negative_14	negative_15	negative_16	negative_17	negative_18	negative_19	negative_20
type	string	string	string	string	string	string	string	string	string	string	string	string	string	string	string	string	string	string	string	string	string	string
details	min: 38 characters mean: 88.92 characters max: 140 characters	min: 12 characters mean: 372.72 characters max: 1113 characters	min: 26 characters mean: 371.1 characters max: 1185 characters	min: 20 characters mean: 326.89 characters max: 1084 characters	min: 23 characters mean: 334.62 characters max: 1477 characters	min: 13 characters mean: 347.64 characters max: 1310 characters	min: 13 characters mean: 337.96 characters max: 1221 characters	min: 13 characters mean: 303.39 characters max: 953 characters	min: 14 characters mean: 328.25 characters max: 1168 characters	min: 16 characters mean: 318.59 characters max: 989 characters	min: 18 characters mean: 284.81 characters max: 853 characters	min: 13 characters mean: 333.89 characters max: 875 characters	min: 26 characters mean: 327.33 characters max: 1041 characters	min: 19 characters mean: 354.74 characters max: 1705 characters	min: 16 characters mean: 334.33 characters max: 1593 characters	min: 18 characters mean: 346.0 characters max: 1374 characters	min: 13 characters mean: 367.52 characters max: 1625 characters	min: 17 characters mean: 357.96 characters max: 1126 characters	min: 23 characters mean: 322.84 characters max: 1060 characters	min: 25 characters mean: 312.1 characters max: 805 characters	min: 22 characters mean: 362.97 characters max: 1002 characters	min: 22 characters mean: 328.62 characters max: 1310 characters

Samples:

anchor	positive	negative_1	negative_2	negative_3	negative_4	negative_5	negative_6	negative_7	negative_8	negative_9	negative_10	negative_11	negative_12	negative_13	negative_14	negative_15	negative_16	negative_17	negative_18	negative_19	negative_20
`Chronic functional somatic symptoms: a single syndrome?`	`Observational study, with a comparison control group.`	`Nasal nitric oxide (NO) and olfactory function are decreased in patients with chronic inflammatory sinonasal disease, suggesting a link between these two parameters. The aim of the study was to investigate nasal NO levels in patients with olfactory dysfunction due to different causes.`	`Fibromyalgia (FM) is a worldwide diffuse musculoskeletal chronic pain condition that affects up to 5% of the general population. Many symptoms associated with mitochondrial diseases are reported in patients with FM such as exercise intolerance, fatigue, myopathy and mitochondrial dysfunction. In this study, we report a mutation in cytochrome b gene of mitochondrial DNA (mtDNA) in a family with FM with inflammasome complex activation.`	`Chronic fatigue syndrome (CFS) has an uncertain pathogenesis. Allergies have been suggested as one cause.`	`Patient's self-reported symptoms on a structured case history questionnaire.`	`Chronic multisymptom illness (reporting at least one symptom in at least two of the following symptom constructs: general fatigue; mood and cognition problems; and musculoskeletal discomfort) was assessed, differentiating by potential burn pit exposure, among deployers who completed 2004 and 2007 Millennium Cohort questionnaires.`	`functional decline and death.`	Olfactory loss is a debilitating symptom of chronic rhinosinusitis (CRS). The pathophysiology of inflammatory olfactory dysfunction likely involves both conductive and sensorineural components. To study the interaction of CRS-associated inflammatory cytokines with the olfactory epithelium (OE), a transgenic mouse model was developed that allows temporally-controlled local gene expression. Interferon-gamma (IFN-γ) is a prototypical T helper 1 (Th1) cytokine linked to nonpolypoid CRS (CRSsNP), as well as sinonasal viral and bacterial infections. In this study, the effects of chronic IFN-γ expression on olfactory histology and function were investigated.	`Monosomy 1p36 syndrome is the most commonly observed subtelomeric deletion syndrome. Patients with this syndrome typically have common clinical features, such as intellectual disability, epilepsy, and characteristic craniofacial features.`	`Migraine is frequently accompanied by symptoms consistent with functional gastrointestinal disorders (FGIDs). This study evaluated the prevalence of functional gastrointestinal symptoms and assessed the symptoms' relationship with the concomitant functional symptoms of anxiety, depression, and headache-related disability.`	`The 2003 Canadian Consensus Criteria for chronic fatigue syndrome (CFS) are often assumed to suggest low-grade systemic inflammation, but have never been formally validated. This study explored the content validity of the Criteria in a sample of adolescents with CFS selected according to a wide case definition.`	Irritable pouch syndrome (IPS) is a functional disorder in patients with ileal pouch-anal anastomosis (IPAA), which presents with symptoms in the absence of structural abnormalities of the pouch. Thus, it resembles other functional disorders, such as irritable bowel syndrome characterized by visceral hypersensitivity in the presence of normal rectal biomechanics. The aim was to assess pouch biomechanics and perception of balloon distension in different groups of subjects with IPAA and to correlate the findings with clinical features.	`Fibromyalgia (FM) and chronic fatigue syndrome (CFS) frequently overlap clinically and have been considered variants of one common disorder. We have recently shown that CFS is associated with a short corrected electrocardiographic QT interval (QTc). In the present study, we evaluated whether FM and CFS can be distinguished by QTc.`	`Single case with clinical follow-up over 2 years.`	`Fibromyalgia syndrome (FMS) is a disease of unknown pathogenesis characterized by chronic musculoskeletal pain. FMS has been also associated with altered endocrinological responses, but findings are inconsistent.`	`Primary Sjögren's syndrome (pSS) is a systemic rheumatic disease in which gastrointestinal (GI) symptoms are common. Faecal calprotectin (FC) is a non-invasive biomarker that has been suggested to discriminate organic intestinal disease from functional disorders. The purpose of this study was to explore the usefulness of FC testing in patients with pSS.`	`An inactive lifestyle has been associated with functional somatic symptoms (FSS), but findings are contradictory. Moreover, mediating factors in this relationship are unclear. We examined whether low physical activity was related to FSS in adolescents, and whether this association was mediated by low physical fitness.`	Nephrotic syndrome is a common kidney disease in both children and adults that is characterized by dramatic structural changes in the actin-rich foot processes of glomerular podocytes. Although glucocorticoids are the primary treatment for nephrotic syndrome, neither the target cell nor mechanism of action of glucocorticoids in nephrotic syndrome is known. For the last 30 years glucocorticoids have been presumed to act by reducing the release of soluble mediators of disease by circulating lymphocytes. In contrast, we hypothesized that glucocorticoids exert their beneficial effects in nephrotic syndrome by direct action on podocytes.	Chronic sclerosing sialadenitis is a fibroinflammatory disease of the salivary glands, characteristically of the submandibular gland. One prior Asian study proposed that chronic sclerosing sialadenitis is a part of the spectrum of IgG4-associated disease. This association has not been confirmed in Western populations. We therefore, investigated the relationship between IgG4 and chronic sclerosing sialadenitis, and compared the histomorphologic features of this condition with those of chronic sialadenitis-not otherwise specified, Sjögren syndrome, and lymphoepithelial sialadenitis.	`Presence of GI symptoms.`	Disrupted-in schizophrenia 1 (DISC1), identified in a pedigree with a familial psychosis with the chromosome translocation (1:11), is a putative susceptibility gene for psychoses such as schizophrenia and major depressive disorder (MDD). Patients with chronic fatigue syndrome (CFS) report having continuous severe fatigue and many overlapping symptoms with MDD; however, the mechanism and effective treatment of CFS are still unclear. We focused on the overlapping symptoms between CFS and MDD and performed an association study of the functional single-nucleotide polymorphism (SNP) in the DISC1 gene with CFS.
`Does sonographic needle guidance affect the clinical outcome of intraarticular injections?`	In total, 148 painful joints were randomized to IA triamcinolone acetonide injection by conventional palpation-guided anatomic injection or sonographic image-guided injection enhanced with a one-handed control syringe (the reciprocating device). A one-needle, 2-syringe technique was used, where the first syringe was used to introduce the needle, aspirate any effusion, and anesthetize and dilate the IA space with lidocaine. After IA placement and synovial space dilation were confirmed, a syringe exchange was performed, and corticosteroid was injected with the second syringe through the indwelling IA needle. Baseline pain, procedural pain, pain at outcome (2 weeks), and changes in pain scores were measured with a 0-10 cm visual analog pain scale (VAS).	`We used a blinded, longitudinal observational design of effectiveness in an effort to determine the accuracy of intra-articular injections and the effect of that accuracy on pain and functional outcomes in patients with various shoulder pathologies.`	Data from standardized procedure notes and postprocedure chest radiographs were extracted and individually reviewed to verify the presence of pneumothorax or misplacement, and any intervention performed for either complication. The overall success rate of ultrasound-guided right internal jugular vein central venous catheter placement was 96.9% with an average of 1.3 attempts. There was only one pneumothorax (0.1% [95% CI, 0-0.4%]), and the rate of catheter misplacement requiring repositioning or replacement was 1.0% (95% CI, 0.6-1.7%). There were no arterial placements found on chest radiographs. Multivariate regression analysis showed no correlation between high-risk patient characteristics and composite complication rate.	`Real-time ultrasound-guided techniques allow for improved cannulation of the internal jugular vein and femoral vein for hemodialysis; however, these techniques require extra sterilization procedures, specialized probes, or needle guides. A simpler ultrasound vessel localization method was performed to investigate whether this alternative approach would aid in the cannulation of the femoral vein for patients in whom temporary angioaccess was required for hemodialysis.`	`To determine whether duration or venue of intravenous antibiotic administration affect lung function.`	`Needle electromyography (NEE) would be more valuable if it could predict outcomes after lumbar epidural steroid injections (LESIs) in lumbosacral radiculopathy (LSR).`	`Among patients with Wilkes stage III and IV disease undergoing arthroscopic lysis and lavage, does the use of an intra-articular injection of sodium hyaluronate (SH), when compared with Ringer lavage, result in better postoperative pain control and temporomandibular joint (TMJ) function?`	`Intraoperative MR imaging and sonography are used for navigation during neurosurgical procedures. The purpose of this experimental study was to evaluate the potential of high-resolution sonography using superparamagnetic iron oxide (SPIO) particles as a contrast medium to delineate brain tumors and to relate these findings with those of MR imaging.`	`Academic general internal medicine practice.`	`To evaluate how guidance on water-intake impacts the degree of nocturia.`	`Intra-articular knee injections are commonly performed in clinical practice for treating various knee joint disorders such as osteoarthritis and rheumatoid arthritis. When selecting the portal for injection, not only intra-articular needle accuracy but also procedural pain should be taken into consideration. The purpose of this study was to determine whether injection through anterolateral portal provokes less pain and provides better pain relief compared to superolateral portal.`	`Ultrasound guidance reduces the required local anesthetic volume for successful peripheral nerve blockade, but it is unclear whether this impacts postoperative analgesia. This prospective, randomized, observer-blinded study tested the hypothesis that a low-volume ultrasound-guided ankle block would provide similar analgesia after foot surgery compared with a conventional-volume surface landmark technique.`	`The purpose of this study was to evaluate how intravascular ultrasound-determined thickness and reflectivity of the inner echogenic layer of coronary artery plaque is affected by changes in collagen, elastin, proteoglycan, calcium and lipid content in the intima and media.`	`The anesthetic requirement is decreased in animals with head injury, but there are no data regarding the effect of intracranial tumor on the potency for intravenous anesthetics. The authors compared the quantal dose-response curves for propofol in patients having large (> or = 30 mm, mass effect) brain tumor with those having smaller (< 30 mm) lesions and with control patients undergoing noncranial surgery.`	`To develop digital multimedia instruction on intraoral suturing.`	`To evaluate a long term efficiency of a deep sclerectomy with T-Flux implant on intraocular pressure`	In bilateral cochlear implant users, electrodes mapped to the same frequency range in each ear may stimulate different places in each cochlea due to an insertion depth difference of electrode arrays. This interaural place of stimulation mismatch can lead to problems with auditory image fusion and sensitivity to binaural cues, which may explain the large localization errors seen in many patients. Previous work has shown that interaural place of stimulation mismatch can lead to off-centered auditory images being perceived even though interaural time and level differences (ITD and ILD, respectively) were zero. Large interaural mismatches reduced the ability to use ITDs for auditory image lateralization. In contrast, lateralization with ILDs was still possible but the mapping of ILDs to spatial locations was distorted. This study extends the previous work by systematically investigating the effect of interaural place of stimulation mismatch on ITD and ILD sensitivity directly and examining...	`Mathematical modeling and use of real world clinical inputs.`	`The use of ultrasonography in regional anesthetic blocks has rapidly evolved over the past few years. It has been speculated that ultrasound guidance might increase success rates and reduce complications. The aim of our study is to compare the success rate and quality of interscalene brachial plexus blocks performed either with direct ultrasound visualization or with the aid of nerve stimulation to guide needle placement.`	Intra-operative nerve monitoring (IONM) of the recurrent laryngeal nerve (RLN) during thyroid and parathyroid surgery is thought to aid in identification and dissection of the RLN. While utilization of IONM is increasing, one area of variability in its application is the assessment of adequate endotracheal tube electrode placement for IONM during the case. The main objective of this study is to assess the overall success of utilizing respiratory variation to confirm proper endotracheal tube placement for RLN monitoring.	Time-dependent development of intracochlear impedances was investigated in 4 different groups of adult patients up to 4 years after implantation. Additionally, during rehabilitation period just after first fitting, impedances before and after stimulation were measured as to investigate the influence of electrical stimulation on the impedances. Results from standard Nucleus 24 Contour (control), standard Nucleus 24 Contour with intraoperative application of steroids, iridium-coated Nucleus 24 Contour, and iridium-coated Nucleus 24 Contour with intraoperative application of steroids were compared.
`Does type 1 diabetes mellitus affect Achilles tendon response to a 10 km run?`	Participants were 7 individuals with T1DM and 10 controls. All regularly ran distances greater than 5 km and VISA-A scores indicated good tendon function (T1DM = 94 ± 11, control = 94 ± 10). There were no diabetic complications and HbA1c was 8.7 ± 2.6 mmol/mol for T1DM and 5.3 ± 0.4 mmol/mol for control groups. Baseline tendon structure was similar in T1DM and control groups - UTC echo-types (I-IV) and anterior-posterior thickness were all p > 0.05. No response to load was seen in either T1DM or control group over the 4-days post exercise.	The inability to produce insulin endogenously precipitates the clinical symptoms of type 1 diabetes mellitus. However, the dynamic trajectory of beta cell destruction following onset remains unclear. Using model-based inference, the severity of beta cell destruction at onset decreases with age where, on average, a 40% reduction in beta cell mass was sufficient to precipitate clinical symptoms at 20 years of age. While plasma C-peptide provides a surrogate measure of endogenous insulin production post-onset, it is unclear as to whether plasma C-peptide represents changes in beta cell mass or beta cell function. The objective of this paper was to determine the relationship between beta cell mass and endogenous insulin production post-onset.	`Seven healthy individuals with type 1 diabetes were tested on two separate occasions, during which either a 30-min MOD or IHE protocol was performed. MOD consisted of continuous exercise at 40% Vo(2peak), while the IHE protocol involved a combination of continuous exercise at 40% Vo(2peak) interspersed with 4-s sprints performed every 2 min to simulate the activity patterns of team sports.`	It is unclear how genetic type 1 diabetes mellitus (DM) influences infarct size when blood glucose is tightly controlled. The aim of this study was to determine the effect of genetic type 1 DM, as occurs in BB rats, on infarct size after transient unilateral middle cerebral artery occlusion (MCAO) in male and female rats. In addition, studies suggest that male type 1 DM rats have a higher incidence of end-organ complications than do females. A second aim of this study was to determine the effect of chronic 17beta-estradiol (E(2)) administration on infarct size in male BB rats.	`Studies on bone mineral characteristics in children with type 1 diabetes mellitus (T1DM) have generated conflicting results.`	`It has recently been reported that the risk of developing nephropathy in patients with insulin dependent (type 1) diabetes mellitus is strongly associated with synergism between poor glycaemic control and carriage of the hypertension associated angiotensin II (type 1) receptor C1166 allele. The same report also revealed an increase in risk of nephropathy in diabetic patients carrying a specific angiotensin II (type 1) receptor haplotype, i.e. C1166/140-bp microsatellite allele (major allele).`	The angiotensin II (type 1) (AT1) receptor mediates many biological effects of the renin-angiotensin system (RAS), leading to remodelling of cardiac tissue. The present study was designed to analyse changes in the function and expression of the AT1 receptor as principal effector of the RAS in myocardium from type 2 diabetic patients compared with non-diabetic myocardium as control. In addition, we determined the effect of treatment with ACE inhibitors or AT1 receptor blockers on expression levels of the receptor in diabetic patients.	`Type 2 diabetes mellitus increases the risk of atherosclerotic cardiovascular disease. Antioxidative properties of high density lipoprotein (HDL) are important for atheroprotection. This study investigated whether the antioxidative functionality of HDL is altered in type 2 diabetes mellitus and aimed to identify potential determinants of this parameter.`	`Life-long insulin is the standard treatment for type 1 diabetes mellitus (T1DM). The role of traditional Chinese medicine (TCM) in T1DM is still not clear. The aim of this study is to explore the prescription pattern of TCM and its impact on the risk of diabetic ketoacidosis (DKA) in patients with T1DM.`	`In type 1 diabetes, lung diffusing capacity for carbon monoxide (DL(CO)) may be impaired, and insulin has been shown to be beneficial in cases in which near-normal metabolic control is achieved. An influence of insulin, per se, on the alveolar-capillary membrane conductance is unexplored. We aimed at testing this possibility.`	aIMT, but not cIMT, was significantly greater in the children with type 1 diabetes mellitus than in control subjects (P < .001). In children with type 1 diabetes mellitus, aIMT correlated with glycosylated hemoglobin (r = 0.31, P = .01) and was independently associated with age (beta = 0.38, P = .001) and low-density lipoprotein cholesterol level (beta = 0.38, P = .001). Vascular function (GTN) was worse in children with type 1 diabetes mellitus who had an aIMT >95th percentile, as defined with the control subjects.	Type 1 diabetes mellitus (T1DM) is caused by specific destruction of the pancreatic beta cells in the islets of Langerhans. Increased sensitivity to cytokines, in particular to interleukin-1beta (IL-1beta) seems to be an acquired trait during beta-cell maturation. In response to cytokines both protective and deleterious mechanisms are induced in beta cells, and when the deleterious prevail, T1DM develops. The aims of this study were to identify perturbation in protein patterns (PiPP) associated with beta-cell maturation, and compare these changes to previous analyses of IL-1beta exposed rat islets. For this purpose, proteome analyses were carried out using a cell-line, which matures from a glucagon-producing pre-beta-cell phenotype (NHI-glu) to an insulin-producing beta-cell phenotype (NHI-ins). We have previously shown that this maturation is accompanied by acquired sensitivity to the toxic effects of IL-1beta.	Type 1 diabetes mellitus is a T-cell-mediated autoimmune disease that leads to a major loss of insulin-secreting beta cells. The further decline of beta-cell function after clinical onset might be prevented by treatment with CD3 monoclonal antibodies, as suggested by the results of a phase 1 study. To provide proof of this therapeutic principle at the metabolic level, we initiated a phase 2 placebo-controlled trial with a humanized antibody, an aglycosylated human IgG1 antibody directed against CD3 (ChAglyCD3).	`Type 2 diabetes mellitus (T2DM) affects approximately 10% of Americans, while 79 million Americans are estimated to have glucose intolerance or prediabetes (pre-DM). The present study was designed to determine whether obese patients with pre-DM or T2DM would lose weight as effectively as obese normoglycemic patients, in a medically supervised high-protein, low-calorie-weight management program.`	`We have previously shown that long-term type 1 diabetes affects the structural organization, contractile apparatus and extracellular matrix (ECM) of the myometrium during early pregnancy in mice.`	`In Achilles injured participants, there was a significant difference between soleus and lateral gastrocnemius offset times during running compared to the asymptomatic controls (p<0.05). There were no significant differences in triceps surae muscle activity between the footwear only and footwear and orthoses condition in the Achilles injured participants.`	Nine adolescents with type 1 diabetes mellitus (five males, four females, aged 16 +/- 1.8 yr, diabetes duration 8.2 +/- 4.1 yr, hemoglobin A1c 7.8 +/- 0.8%, mean +/- SD) were subjected on two different occasions to a rest or 45 min of exercise at 95% of their lactate threshold. Insulin was administered iv at a rate based on their usual insulin dose, with similar plasma insulin levels for both studies (82.1 +/- 19.0, exercise vs. 82.7 +/- 16.4 pmol/liter, rest). Glucose was infused to maintain euglycemia for 18 h.	`Type 1 diabetes mellitus (T1DM) is influenced by genetic as well as environmental factors. Its incidence has risen considerably since the 1950s.`	`Insulin resistance has been associated with hypertension and with renal complications in patients with type 1 diabetes mellitus. Causal relationships have not been fully explained.`	`Diabetes mellitus (DM) is a common metabolic disease among the middle-aged and older population, which leads to an increase of stroke incidence and poor stroke recovery. The present study was designed to investigate the impact of DM on brain damage and on ischemic brain repair after stroke in aging animals.`	`Type 1 diabetes (T1D) is an autoimmune disease with multiple susceptibility genes. The aim of this study was to determine whether combining IDDM1/HLA and IDDM2/ insulin( INS) 5' variable number of tandem repeat locus (VNTR) genotypes improves T1D risk assessment.`

Loss: MatryoshkaLoss with these parameters:
json
1{ 2 "loss": "MultipleNegativesRankingLoss", 3 "matryoshka_dims": [ 4 1024, 5 512, 6 256, 7 128, 8 64, 9 32 10 ], 11 "matryoshka_weights": [ 12 1, 13 1, 14 1, 15 1, 16 1, 17 1 18 ], 19 "n_dims_per_step": -1 20}

miracl

Dataset: miracl at 07e2b62
Size: 789,900 evaluation samples
Columns: anchor, positive, and negative

Approximate statistics based on the first 1000 samples:

	anchor	positive	negative
type	string	string	string
details	min: 11 characters mean: 38.08 characters max: 86 characters	min: 93 characters mean: 746.58 characters max: 3851 characters	min: 18 characters mean: 648.12 characters max: 2554 characters

Samples:

anchor	positive	negative
`What was Paul Heyman's first production?`	Paul Heyman[object Object]On May 23, 2005, Heyman returned in a segment with Vince McMahon and Eric Bischoff announcing ECW One Night Stand, with Heyman in charge. On the May 22 episode of "Raw", Heyman appeared as ECW Representative promoting "One Night Stand". On May 25, 2006 it was announced that ECW would relaunch, as a third WWE brand. Heyman was in charge of the new brand on-camera but had minimal creative input off-camera as well. On the May 29 episode of "Raw", during a face-off with Mick Foley, Heyman announced that he was granted a draft pick from both Raw and SmackDown! by Vince McMahon. His Raw draft pick was former ECW wrestler (and Money in the Bank contract holder) Rob Van Dam, and his SmackDown! draft pick was Kurt Angle. Heyman predicted that Van Dam would defeat John Cena at "One Night Stand" for the WWE Championship and then declare himself the new ECW World Heavyweight Champion.	Jay Wolpert[object Object]Wolpert left Goodson-Todman to form his own production company, and his first game show was the 1979 series "Whew!" for CBS. Wolpert produced the series with Burt Sugarman for most of its run. "Whew!" was canceled in 1980 and Wolpert did not return to television with a series until January 1983, despite shooting several pilots in the interim. On January 3, 1983, Wolpert's "Hit Man" debuted on NBC with Peter Tomarken as its host. "Hit Man" lasted thirteen weeks on the air.
`When was Locke born?`	`Adria Locke Langley[object Object]Locke was born in Iowa, 1899, as the youngest of three children. When she was young her family moved to Stanton, Nebraska and that is where she grew up. Her father William Locke, was president of the Omaha livestock market and a Quaker. He had certain ideas of what a woman should be. Because of this Adria's grandfather, Thomas Glendenning, took over in teaching her what would later be a great social consciousness.`	Keith Locke[object Object]Locke was born and grew up in Christchurch, to Jack and Elsie Locke, prominent lifelong political activists for a wide variety of causes. Their four children were brought up in this environment and followed their parents into a life of activism, (as well as Keith, his sister Maire Leadbeater is a well-known activist and former city councillor for Auckland City Council). His father Jack was under surveillance during the 1951 New Zealand waterfront dispute.[object Object]Former Prime Minister Robert Muldoon is said to have described the Lockes as the most "notorious Communist family in New Zealand". The Lockes lived in the Avon Loop area of the Christchurch Central City and were very active in the community notably organising Avon River clean-ups and native tree planting and arguing against development of the area, and in favour of retaining the character of the area.
`How do plants get blight?`	Bacterial blight of soybean[object Object]Bacterial blight of soybeans can enter leaves through wounds or natural openings such as stomata. After gaining entrance to the host leaves, "Pseudomonas syringae pv. glycinea" multiplies in the leaf intercellular fluid. The pathogen must then overcome the plants defenses. "Pseudomonas syringae pv. glycinea" accomplishes this by using the type three secretion system to inject a variety of pathogenicity effector proteins (Hrp proteins) into the plant cell cytoplasm. These proteins act by interfering with effector-triggered immunity and producing phytohormones/toxins that suppress plant defenses. The expression of these virulence factors depends on the environmental conditions at the time of infection (see "environment section). Furthermore, expression of virulence factors will only take place when a sufficiently large population of bacteria is present, which is determined through quorum sensing. When successful, the common symptoms of bacterial blight will be...	Gummy stem blight[object Object]Gummy Stem Blight is a cucurbit-rot disease caused by the fungal plant pathogen "Didymella bryoniae" (anamorph "Phoma cucurbitacearum").[object Object]Gummy Stem Blight can affect a host at any stage of growth in its development and affects all parts of the host including leaves, stems and fruits. Symptoms generally consist of circular dark tan lesions that blight the leaf, water soaked leaves, stem cankers, and gummy brown ooze that exudes from cankers, giving it the name Gummy Stem Blight. Gummy Stem Blight reduces yields of edible cucurbits by devastating the vines and leaves and rotting the fruits.

Loss: MatryoshkaLoss with these parameters:
json
1{ 2 "loss": "MultipleNegativesRankingLoss", 3 "matryoshka_dims": [ 4 1024, 5 512, 6 256, 7 128, 8 64, 9 32 10 ], 11 "matryoshka_weights": [ 12 1, 13 1, 14 1, 15 1, 16 1, 17 1 18 ], 19 "n_dims_per_step": -1 20}

mldr

Dataset: mldr at 40ad767
Size: 200,000 evaluation samples
Columns: anchor, positive, and negative

Approximate statistics based on the first 1000 samples:

	anchor	positive	negative
type	string	string	string
details	min: 16 characters mean: 66.52 characters max: 527 characters	min: 3223 characters mean: 20386.58 characters max: 131268 characters	min: 3611 characters mean: 15772.93 characters max: 145135 characters

Samples:

anchor	positive	negative
`What is an Aboriginal reserve and how was it created?`	An Aboriginal reserve, also called simply reserve, was a government-sanctioned settlement for Aboriginal Australians, created under various state and federal legislation. Along with missions and other institutions, they were used from the 19th century to the 1960s to keep Aboriginal people separate from the white Australian population, for various reasons perceived by the government of the day. The Aboriginal reserve laws gave governments much power over all aspects of Aboriginal people’s lives.[object Object][object Object]Protectors of Aborigines and (later) Aboriginal Protection Boards were appointed to look after the interests of the Aboriginal people.[object Object][object Object]History[object Object]Aboriginal reserves were used from the nineteenth century to keep Aboriginal people separate from the white Australian population, often ostensibly for their protection.[object Object][object Object]Protectors of Aborigines had been appointed from as early as 1836 in South Australia (with Matthew Moorhouse as the first permanent appointment as Chief Protector in 1839), with the G...	The Nature Reserve () is a nature reserve located in the municipalities of Sorsele and Storuman in Västerbotten County of Swedish Lapland. It is the largest natural reserve in Sweden and one of the largest protected areas in Europe, totaling 562,772 ha (approx. 5,628 km2).[object Object][object Object]Most of the reserve is made up of several Scandinavian Mountains, the main ones being Artfjället, Norra Storfjället, Ammarfjället and Björkfjället. Most of the landscapes of the Swedish mountains are represented. This ranges from the pronounced alpine character of Norra Storfjället, which includes the highlight of the reserve, the Norra Sytertoppen (1,768 m), to the plateau and plains near the base of the mountains. The differences in elevation highlight the diversity of rocks in the mountains. Among the mountains are the valleys and waterways of the Ume River drainage basin. This includes a portion of the Vindel River, after which the reserve is named. Towards the east, the elevation decreases and the mountains gi...
`What is the English translation of the song "Bosanac"?`	`This is a list of Bosnian patriotic songs.[object Object][object Object]{`	class="wikitable sortable"
`Who were the representatives that declined to run for reelection to the Vermont House of Representatives in 2016?`	Selene Colburn is an American politician currently serving in the Vermont House of Representatives from the Chittenden-6-4 district since 2017 as a member of the Vermont Progressive Party. Prior to her tenure in the State House she served on the city council in Burlington, Vermont. She is the first female chair of the House Progressive Caucus.[object Object][object Object]Colburn was born in Burlington, and educated at Burlington High School, Bennington College, and Simmons University. She became active in politics in her youth when she joined anti-war demonstrations.[object Object][object Object]Colburn was first elected to office with her election to the Burlington city council in the 2014 election and she won reelection in the 2015 and 2017 elections. She was elected to the state house alongside Brian Cina in the 2016 election with the nominations of the Progressive and Democratic parties and was reelected in the 2018 and 2020 elections. She was selected to serve as assistant chair of the Vermont Progressive Party's caucus in the state h...	The 2016 United States Senate election in Florida was held November 8, 2016 to elect a member of the United States Senate to represent the State of Florida, concurrently with the 2016 U.S. presidential election, as well as other elections to the United States Senate in other states and elections to the United States House of Representatives and various state and local elections. The primary elections for both the Republicans and Democrats took place on August 30, 2016.[object Object][object Object]Incumbent Republican Senator Marco Rubio ran for another term but faced well-funded Republican primary opposition after initially announcing he would not seek re-election to his Senate seat. He had openly considered whether to seek re-election or run for president in 2016. He stated in April 2014 that he would not run for both the Senate and president in 2016, as Florida law prohibits a candidate from simultaneously appearing twice on a ballot, but did not rule out running for either office.[object Object][object Object]However, in April 2015, Ru...

Loss: MatryoshkaLoss with these parameters:
json
1{ 2 "loss": "MultipleNegativesRankingLoss", 3 "matryoshka_dims": [ 4 1024, 5 512, 6 256, 7 128, 8 64, 9 32 10 ], 11 "matryoshka_weights": [ 12 1, 13 1, 14 1, 15 1, 16 1, 17 1 18 ], 19 "n_dims_per_step": -1 20}

mr_tydi

Dataset: mr_tydi at abbdf55
Size: 354,700 evaluation samples
Columns: anchor, positive, and negative

Approximate statistics based on the first 1000 samples:

	anchor	positive	negative
type	string	string	string
details	min: 13 characters mean: 38.6 characters max: 101 characters	min: 71 characters mean: 667.4 characters max: 21297 characters	min: 10 characters mean: 619.16 characters max: 3134 characters

Samples:

anchor	positive	negative
`What is the largest city in Sardinia?`	Sardinia[object Object]Sardinia is politically a region of Italy, whose official name is Regione Autonoma della Sardegna / Regione Autònoma de Sardigna (Autonomous Region of Sardinia),[3] and enjoys some degree of domestic autonomy granted by a specific Statute.[4] It is divided into four provinces and a metropolitan city, with Cagliari being the region's capital and its largest city as well. Sardinia's indigenous language and the other minority languages (Sassarese, Corsican Gallurese, Algherese Catalan and Ligurian Tabarchino) spoken on the island are recognized by the regional law and enjoy "equal dignity" with Italian.[5]	History of Cagliari[object Object]Caralis (or Karales) was the capital of the Roman province of Sardinia and Corsica and was elevated to the rank of "Municipium", a result of the civil war between Julius Caesar and Pompey when Caesar himself granted this status in gratitude to the city for its fidelity during the bloody war. All Caralitani obtained Roman citizenship and were enrolled in the tribe Quirina. The territory of the city included the campidano plain, likely becoming Sanluri.[object Object]With about 20,000 inhabitants Caralis was the largest and most populous city of the island and the most important of the western Mediterranean basin of the Republic, and later of the Roman Empire. The city was equipped with important road links to the main towns of the island such as Sulki, with the coastal road and with that running through the valley of the Cixerri, Olbia and Tibula along the east coast, and Turris and Tibula along the road modeled on the current "Carlo Felice", and finally a road through the centre ...
`What activism did Katherine Schmidt participate in?`	Katherine Schmidt[object Object]Katherine Schmidt (February 6, 1899 – April 18, 1978) was an American artist and art activist. Early in her career the figure studies, landscapes, and still lifes she painted drew praise for their "purity and clarity of color," "sound draftsmanship," and "individual choice of subject and its handling."[3] During the 1930s she was known mainly for the quality of her still life paintings which showed, one critic said, "impeccable artistry."[4] At the end of her career, in the 1960s and 1970s, she produced specialized and highly disciplined still lifes of objects such as dead leaves and pieces of crumpled paper, which, said a critic, approached a "magical realism."[5] As an art activist she helped promote the rights of artists for fair remuneration.[6]	Caroline Katzenstein[object Object]Caroline Katzenstein (1888–1968) was an American suffragist, activist, advocate for equal rights, insurance agent, and author. She was active in the local Philadelphia suffragist movement through the Pennsylvania branch of the National American Woman Suffrage Association and the Equal Franchise Society of Philadelphia. She played a role in the formation of the Congressional Union for Women Suffrage, which later became the National Women's Party. Katzenstein was also active in the movement for equal rights, serving on the Women's Joint Legislative Committee with Alice Paul, and championing the cause for the Equal Rights Amendment. She was the author of "Lifting the Curtain: the State and National Woman Suffrage Campaigns in Pennsylvania as I Saw Them" (1955).
`Who made the first wristwatch?`	History of watches[object Object]From the beginning, wristwatches were almost exclusively worn by women, while men used pocketwatches up until the early 20th century. The concept of the wristwatch goes back to the production of the very earliest watches in the 16th century. Some people say the world's first wristwatch was created by Abraham-Louis Breguet for Caroline Murat, Queen of Naples, in 1810.[21][22][23][24][25] Elizabeth I of England received a wristwatch from Robert Dudley in 1571, described as an arm watch. By the mid nineteenth century, most watchmakers produced a range of wristwatches, often marketed as bracelets, for women.[26]	Rolex[object Object]In 1908, Wilsdorf registered the trademark "Rolex" and opened an office in La Chaux-de-Fonds, Switzerland.[14][15] The book The Best of Time: Rolex Wristwatches: An Unauthorized History by Jeffrey P. Hess and James Dowling says that the name was just made up.[16] One story, never confirmed by Wilsdorf, recounts that the name came from the French phrase horlogerie exquise, meaning "exquisite clockwork"[9] or as a contraction of "horological excellence". Wilsdorf was said to want his watch brand's name to be easily pronounceable in any language.[17] He also thought that the name "Rolex" was onomatopoeic, sounding like a watch being wound. It is easily pronounceable in many languages and, as all its upper-case letters have the same size, can be written symmetrically. It was also short enough to fit on the face of a watch.[17]

Loss: MatryoshkaLoss with these parameters:
json
1{ 2 "loss": "MultipleNegativesRankingLoss", 3 "matryoshka_dims": [ 4 1024, 5 512, 6 256, 7 128, 8 64, 9 32 10 ], 11 "matryoshka_weights": [ 12 1, 13 1, 14 1, 15 1, 16 1, 17 1 18 ], 19 "n_dims_per_step": -1 20}

Training Hyperparameters

Non-Default Hyperparameters

eval_strategy: steps
per_device_train_batch_size: 2048
per_device_eval_batch_size: 2048
learning_rate: 0.2
num_train_epochs: 1
warmup_ratio: 0.1
bf16: True
batch_sampler: no_duplicates

All Hyperparameters

Click to expand

overwrite_output_dir: False
do_predict: False
eval_strategy: steps
prediction_loss_only: True
per_device_train_batch_size: 2048
per_device_eval_batch_size: 2048
per_gpu_train_batch_size: None
per_gpu_eval_batch_size: None
gradient_accumulation_steps: 1
eval_accumulation_steps: None
torch_empty_cache_steps: None
learning_rate: 0.2
weight_decay: 0.0
adam_beta1: 0.9
adam_beta2: 0.999
adam_epsilon: 1e-08
max_grad_norm: 1.0
num_train_epochs: 1
max_steps: -1
lr_scheduler_type: linear
lr_scheduler_kwargs: {}
warmup_ratio: 0.1
warmup_steps: 0
log_level: passive
log_level_replica: warning
log_on_each_node: True
logging_nan_inf_filter: True
save_safetensors: True
save_on_each_node: False
save_only_model: False
restore_callback_states_from_checkpoint: False
no_cuda: False
use_cpu: False
use_mps_device: False
seed: 42
data_seed: None
jit_mode_eval: False
use_ipex: False
bf16: True
fp16: False
fp16_opt_level: O1
half_precision_backend: auto
bf16_full_eval: False
fp16_full_eval: False
tf32: None
local_rank: 0
ddp_backend: None
tpu_num_cores: None
tpu_metrics_debug: False
debug: []
dataloader_drop_last: False
dataloader_num_workers: 0
dataloader_prefetch_factor: None
past_index: -1
disable_tqdm: False
remove_unused_columns: True
label_names: None
load_best_model_at_end: False
ignore_data_skip: False
fsdp: []
fsdp_min_num_params: 0
fsdp_config: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}
fsdp_transformer_layer_cls_to_wrap: None
accelerator_config: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
deepspeed: None
label_smoothing_factor: 0.0
optim: adamw_torch
optim_args: None
adafactor: False
group_by_length: False
length_column_name: length
ddp_find_unused_parameters: None
ddp_bucket_cap_mb: None
ddp_broadcast_buffers: False
dataloader_pin_memory: True
dataloader_persistent_workers: False
skip_memory_metrics: True
use_legacy_prediction_loop: False
push_to_hub: False
resume_from_checkpoint: None
hub_model_id: None
hub_strategy: every_save
hub_private_repo: False
hub_always_push: False
gradient_checkpointing: False
gradient_checkpointing_kwargs: None
include_inputs_for_metrics: False
eval_do_concat_batches: True
fp16_backend: auto
push_to_hub_model_id: None
push_to_hub_organization: None
mp_parameters:
auto_find_batch_size: False
full_determinism: False
torchdynamo: None
ray_scope: last
ddp_timeout: 1800
torch_compile: False
torch_compile_backend: None
torch_compile_mode: None
dispatch_batches: None
split_batches: None
include_tokens_per_second: False
include_num_input_tokens_seen: False
neftune_noise_alpha: None
optim_target_modules: None
batch_eval_metrics: False
eval_on_start: False
use_liger_kernel: False
eval_use_gather_object: False
batch_sampler: no_duplicates
multi_dataset_batch_sampler: proportional

Training Logs

Click to expand

Epoch	Step	Training Loss	gooaq loss	msmarco loss	squad loss	s2orc loss	allnli loss	paq loss	trivia qa loss	msmarco 10m loss	swim ir loss	pubmedqa loss	miracl loss	mldr loss	mr tydi loss	NanoClimateFEVER_cosine_ndcg@10	NanoDBPedia_cosine_ndcg@10	NanoFEVER_cosine_ndcg@10	NanoFiQA2018_cosine_ndcg@10	NanoHotpotQA_cosine_ndcg@10	NanoMSMARCO_cosine_ndcg@10	NanoNFCorpus_cosine_ndcg@10	NanoNQ_cosine_ndcg@10	NanoQuoraRetrieval_cosine_ndcg@10	NanoSCIDOCS_cosine_ndcg@10	NanoArguAna_cosine_ndcg@10	NanoSciFact_cosine_ndcg@10	NanoTouche2020_cosine_ndcg@10	NanoBEIR_mean_cosine_ndcg@10
0	0	-	-	-	-	-	-	-	-	-	-	-	-	-	-	0.0741	0.3518	0.2118	0.0793	0.3538	0.3200	0.1954	0.1589	0.6759	0.1532	0.0945	0.4296	0.1455	0.2495
0.0000	1	32.2074	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.0064	250	22.7851	8.3992	17.7191	17.6791	16.6296	18.7913	12.1404	18.9992	12.1891	11.6795	26.3440	8.5795	19.3571	9.5985	0.2366	0.5129	0.6004	0.1960	0.6334	0.3941	0.2713	0.3392	0.7977	0.2416	0.3819	0.5448	0.4679	0.4321
0.0127	500	9.6296	4.6987	13.6254	11.7605	12.5290	17.2038	6.9342	12.0873	7.3539	9.1374	23.1663	4.2482	14.5991	3.3365	0.2929	0.5509	0.6529	0.2890	0.6495	0.4244	0.2873	0.3690	0.8830	0.2373	0.3815	0.5802	0.5422	0.4723
0.0191	750	6.7008	3.6302	11.7061	10.1299	11.2366	15.1612	5.5833	10.6967	5.7074	8.8117	23.2404	3.5115	12.5734	2.4346	0.3101	0.5565	0.6684	0.3406	0.6354	0.4111	0.2972	0.3894	0.8611	0.2513	0.3613	0.5840	0.5578	0.4788
0.0255	1000	5.8282	2.9789	11.2050	9.5095	10.6029	14.7717	5.0173	9.6170	5.1146	9.0596	22.3746	3.0912	12.2982	2.2626	0.3066	0.5514	0.6654	0.3252	0.6390	0.4139	0.2917	0.4168	0.8678	0.2590	0.3884	0.6214	0.5614	0.4852
0.0318	1250	5.3975	2.8335	11.0393	9.3407	9.6014	14.5350	4.8262	9.4577	4.9009	8.9271	21.8053	3.2513	12.2634	1.6880	0.3090	0.5449	0.6607	0.3432	0.6243	0.4145	0.3026	0.4392	0.8801	0.2608	0.3760	0.6102	0.5768	0.4879
0.0382	1500	5.3077	2.7030	10.6366	8.9914	9.8588	14.6669	4.6253	9.3728	4.5863	9.1788	22.0617	2.8378	10.9618	1.8702	0.3240	0.5421	0.7026	0.3507	0.6227	0.4134	0.3150	0.3996	0.8776	0.2493	0.3625	0.6120	0.5642	0.4874
0.0445	1750	4.9354	2.6691	10.6339	8.9606	9.7095	14.9174	4.5880	9.3114	4.1786	8.2898	22.8332	2.6850	11.3781	1.6352	0.3092	0.5591	0.6615	0.3253	0.6363	0.3926	0.3165	0.4057	0.9019	0.2600	0.3685	0.6030	0.5563	0.4843
0.0509	2000	4.8017	2.5867	10.0547	8.8155	9.6765	14.7973	4.3931	9.2721	4.0193	7.7955	23.4468	1.9884	11.4315	1.8009	0.3274	0.5615	0.7024	0.3531	0.6481	0.3959	0.3134	0.4183	0.8849	0.2505	0.3694	0.5991	0.5664	0.4916
0.0573	2250	4.8193	2.4974	9.9855	8.8389	9.6763	14.4220	4.3112	9.1019	4.0176	8.4064	22.7034	2.7534	11.5256	2.3585	0.3149	0.5392	0.6689	0.3344	0.6495	0.4080	0.3058	0.3953	0.8857	0.2588	0.3426	0.5986	0.5756	0.4829
0.0636	2500	4.8773	2.7116	10.2180	8.6935	9.7664	14.3161	4.2722	8.9829	4.2454	9.1911	22.8367	2.6666	11.6110	2.0147	0.3377	0.5455	0.6547	0.3130	0.6396	0.4259	0.3256	0.4226	0.8825	0.2491	0.3908	0.5852	0.5656	0.4875
0.0700	2750	4.5856	2.7758	9.9754	8.6197	9.6282	14.4828	4.1534	8.8766	4.2312	9.5281	21.9368	2.9119	9.7259	1.9405	0.3384	0.5763	0.6646	0.3068	0.6740	0.4195	0.3155	0.4301	0.8809	0.2366	0.3965	0.5963	0.5714	0.4928
0.0764	3000	4.3725	2.5993	10.1291	8.7361	8.9502	14.8227	4.1163	8.8371	4.1014	9.2259	23.4047	3.2715	10.3051	2.3604	0.3040	0.5703	0.6836	0.3006	0.6355	0.3993	0.3318	0.4236	0.9001	0.2579	0.3966	0.5902	0.5770	0.4900
0.0827	3250	4.4409	2.5753	10.0879	8.5131	8.7106	14.8015	4.0560	8.8296	4.1868	9.3069	22.5793	2.3810	8.6639	2.0435	0.3147	0.5806	0.6766	0.3342	0.6293	0.4134	0.3208	0.4089	0.8834	0.2656	0.3784	0.6119	0.5731	0.4916
0.0891	3500	4.6192	2.4352	9.9932	8.5716	9.2016	14.1559	4.0585	8.9413	3.8278	8.6089	22.9941	2.5541	9.4271	1.7271	0.3052	0.5531	0.6921	0.3284	0.6391	0.4027	0.3288	0.4235	0.8938	0.2565	0.3928	0.5848	0.5741	0.4904
0.0955	3750	4.4805	2.5370	10.0723	8.4652	8.8024	14.4678	4.0045	8.8487	3.6855	8.4129	22.5177	2.5961	9.1362	1.6572	0.2996	0.5669	0.7051	0.3007	0.6433	0.3822	0.3127	0.4419	0.8853	0.2741	0.3696	0.5911	0.5796	0.4886
0.1018	4000	4.3246	2.4130	10.0235	8.4203	9.1794	14.4445	3.9667	8.8021	3.6692	8.0637	22.1590	2.2690	9.5487	1.4625	0.2987	0.5659	0.6905	0.3103	0.6280	0.4004	0.3025	0.4273	0.9026	0.2728	0.3710	0.6004	0.5746	0.4881
0.1082	4250	4.547	2.2318	10.0773	8.4970	8.6540	13.9845	3.9893	8.7805	3.4480	8.1038	21.2066	2.4344	9.2932	1.4761	0.3037	0.5760	0.7118	0.3239	0.6135	0.4056	0.3170	0.4323	0.8860	0.2620	0.3887	0.6211	0.5824	0.4941
0.1145	4500	4.3008	2.3243	9.8733	8.5042	9.0158	14.0935	3.9014	8.8306	3.5557	8.4240	21.2823	2.6280	9.4869	1.8310	0.3076	0.5788	0.6867	0.3187	0.6190	0.3936	0.3181	0.4160	0.8895	0.2564	0.3960	0.6148	0.5736	0.4899
0.1209	4750	4.2386	2.4259	9.8799	8.3964	9.1116	13.9412	3.8572	8.7955	3.6524	9.6881	21.3812	2.2282	8.9280	1.5408	0.3393	0.5783	0.7190	0.3137	0.6239	0.3953	0.3044	0.4231	0.8768	0.2636	0.3828	0.6043	0.5662	0.4916
0.1273	5000	4.141	2.4005	9.9973	8.2741	9.1627	14.4273	3.7931	8.7825	3.6856	9.0001	21.6595	2.2374	9.2771	1.4845	0.3243	0.5705	0.6858	0.3304	0.6328	0.3888	0.3145	0.4096	0.8775	0.2492	0.3769	0.6001	0.5600	0.4862
0.1336	5250	4.3221	2.4200	9.8792	8.2559	9.0431	13.9564	3.8055	8.5773	3.6137	8.1900	21.6272	2.2271	8.1229	1.6308	0.3207	0.5876	0.6945	0.3449	0.6232	0.4072	0.3011	0.4084	0.8894	0.2557	0.3668	0.5905	0.5582	0.4883
0.1400	5500	4.2121	2.3857	10.1277	8.3257	9.0878	13.8545	3.7696	8.6034	3.5613	8.7845	21.5562	2.3611	7.5145	1.9243	0.3160	0.5683	0.7024	0.3382	0.6244	0.4038	0.3075	0.4316	0.8784	0.2603	0.3876	0.5866	0.5650	0.4900
0.1464	5750	4.1071	2.4265	9.9555	8.0947	9.1289	14.0017	3.7337	8.6306	3.4562	8.3132	21.7894	2.1157	8.1967	1.5567	0.3317	0.5604	0.7101	0.3645	0.6460	0.3857	0.2987	0.4236	0.8830	0.2535	0.3867	0.5767	0.5612	0.4909
0.1527	6000	4.1189	2.3586	10.0799	8.0905	9.0291	14.0232	3.7064	8.5220	3.4742	8.3858	21.5903	2.1663	7.6242	1.4405	0.3264	0.5614	0.6825	0.3668	0.6296	0.3972	0.2863	0.4296	0.8869	0.2482	0.3809	0.6004	0.5556	0.4886
0.1591	6250	4.0873	2.2906	9.9813	8.1351	8.5907	13.8665	3.7028	8.5648	3.5042	8.1623	21.6688	2.2940	7.6652	1.5228	0.3444	0.5666	0.7035	0.3415	0.6188	0.3992	0.2989	0.4318	0.8816	0.2504	0.4014	0.6042	0.5637	0.4928
0.1654	6500	3.9586	2.2378	9.9318	8.0887	8.7977	14.1260	3.6614	8.5028	3.3178	8.3118	21.5718	2.2074	8.0905	1.7266	0.3299	0.5828	0.6994	0.3505	0.6375	0.3988	0.3225	0.4173	0.8891	0.2448	0.3930	0.6034	0.5614	0.4946
0.1718	6750	4.1981	2.2863	9.5475	7.6881	8.5462	13.6929	3.6704	8.6660	3.3401	8.9262	21.8933	2.0578	8.4832	1.6796	0.3456	0.5600	0.6983	0.3549	0.6445	0.3844	0.3139	0.4263	0.8952	0.2626	0.3877	0.5820	0.5597	0.4935
0.1782	7000	4.0528	2.3292	9.5129	7.8273	8.7743	13.6930	3.6284	8.6346	3.3430	8.5204	21.3129	2.3350	8.8695	1.9034	0.3457	0.5673	0.6850	0.3274	0.6321	0.3981	0.3171	0.4252	0.8830	0.2643	0.3901	0.5888	0.5590	0.4910
0.1845	7250	4.0547	2.2386	9.5373	7.9214	8.7896	13.6151	3.6172	8.5316	3.3128	9.3566	21.4568	2.3743	9.1696	1.7235	0.3528	0.5597	0.6931	0.3369	0.6327	0.3951	0.3111	0.4368	0.8787	0.2552	0.3758	0.5911	0.5515	0.4900
0.1909	7500	4.3005	2.2273	9.4397	7.9013	8.8606	13.3847	3.6401	8.4134	3.2583	8.3415	21.4206	2.4573	9.2348	1.4832	0.3557	0.5642	0.7145	0.3380	0.6412	0.3772	0.3085	0.4278	0.8792	0.2522	0.3738	0.5843	0.5587	0.4904
0.1973	7750	4.0054	2.2277	9.4653	7.9297	8.5999	13.6106	3.6049	8.3861	3.2335	9.3198	21.6595	2.4730	8.7335	1.6145	0.3396	0.5535	0.6901	0.3556	0.6311	0.3867	0.3182	0.4308	0.8692	0.2590	0.3654	0.5925	0.5558	0.4883
0.2036	8000	3.8426	2.2970	9.4352	7.9532	8.6501	13.9004	3.5835	8.3664	3.2109	8.4302	21.0340	2.1047	9.0103	1.1751	0.3420	0.5695	0.6868	0.3746	0.6434	0.4042	0.3193	0.4259	0.8847	0.2623	0.3785	0.5945	0.5702	0.4966
0.2100	8250	3.9404	2.2417	9.5349	7.8978	8.6899	13.8131	3.5697	8.3664	3.1548	8.6003	21.6214	2.0881	9.1829	0.9559	0.3306	0.5651	0.6959	0.3448	0.6455	0.3857	0.3123	0.4431	0.9009	0.2580	0.3981	0.6073	0.5748	0.4971
0.2164	8500	3.9522	2.2103	9.6575	7.9030	8.3617	14.0083	3.5433	8.3198	3.2148	8.5004	20.8166	2.3194	8.0428	1.2475	0.3343	0.5688	0.7054	0.3411	0.6625	0.3919	0.3148	0.4213	0.8965	0.2665	0.3454	0.6133	0.5782	0.4954
0.2227	8750	3.9665	2.1659	9.7216	7.8772	8.6394	13.9406	3.5289	8.3360	3.1863	8.7835	21.2033	2.1874	8.4683	1.2399	0.3231	0.5715	0.6794	0.3378	0.6684	0.3966	0.3171	0.4048	0.8887	0.2660	0.3461	0.6094	0.5559	0.4896
0.2291	9000	4.0217	2.1042	9.6765	7.8951	8.4255	13.8092	3.5314	8.3896	3.1097	8.0204	21.4246	2.0600	8.7244	1.3343	0.3218	0.5696	0.6931	0.3569	0.6654	0.3978	0.3159	0.4193	0.9000	0.2827	0.3750	0.5890	0.5796	0.4974
0.2354	9250	4.0008	2.0865	9.4154	7.9689	8.5298	13.6352	3.5371	8.4191	3.0414	8.4828	21.5173	1.9966	7.6465	1.1097	0.3282	0.5675	0.6934	0.3476	0.6559	0.3907	0.3272	0.4132	0.9038	0.2712	0.3891	0.5951	0.5716	0.4965
0.2418	9500	3.8041	2.0969	9.4478	7.9720	8.6298	13.7493	3.5003	8.4702	3.0939	8.5108	21.6929	1.9457	7.9947	1.2784	0.3291	0.5655	0.6915	0.3533	0.6495	0.3949	0.3291	0.4313	0.9007	0.2641	0.3910	0.5838	0.5710	0.4965
0.2482	9750	3.9483	2.1627	9.5085	7.8994	8.7048	13.4591	3.4941	8.3342	3.1202	8.7011	20.9101	1.8594	7.8214	1.1181	0.3312	0.5665	0.6870	0.3608	0.6530	0.4038	0.3267	0.4511	0.8973	0.2613	0.4063	0.5832	0.5660	0.4996
0.2545	10000	3.9843	2.1229	9.4348	7.9006	8.2377	13.5086	3.4905	8.2574	3.0114	8.3314	20.5157	1.9033	6.7485	1.1358	0.3365	0.5704	0.6858	0.3616	0.6561	0.3953	0.3263	0.4322	0.8833	0.2690	0.3995	0.6004	0.5691	0.4989
0.2609	10250	3.8779	2.1442	9.4343	7.9411	8.4677	13.5196	3.4558	8.2314	3.0592	8.6126	20.3879	1.9378	6.8287	1.1270	0.3192	0.5688	0.7051	0.3669	0.6577	0.3913	0.3167	0.4479	0.8978	0.2619	0.4030	0.5966	0.5699	0.5002
0.2673	10500	3.8708	2.1487	9.6030	7.9352	8.4657	13.3417	3.4574	8.3005	3.0167	8.4265	20.5354	1.9044	6.9217	1.1079	0.3255	0.5730	0.7020	0.3445	0.6527	0.3861	0.3178	0.4322	0.8859	0.2644	0.3997	0.5969	0.5739	0.4965
0.2736	10750	3.8153	2.1486	9.2440	7.8432	8.5271	13.4366	3.4702	8.3222	2.9680	8.4780	20.6855	1.8790	6.6261	1.0838	0.3240	0.5727	0.6970	0.3487	0.6546	0.3897	0.3294	0.4421	0.8908	0.2548	0.3802	0.6028	0.5744	0.4970
0.2800	11000	3.9693	2.1542	9.4072	7.8122	8.6226	13.1083	3.4426	8.2961	3.0086	8.6255	20.5001	1.9112	6.6305	1.1554	0.3273	0.5631	0.6912	0.3502	0.6431	0.4012	0.3173	0.4422	0.8834	0.2740	0.3829	0.6117	0.5745	0.4971
0.2864	11250	3.7596	2.1288	9.4115	7.8479	8.6295	13.1814	3.4282	8.2609	2.9531	8.6024	20.7827	1.8339	7.0503	1.0273	0.3301	0.5714	0.6814	0.3457	0.6447	0.3962	0.3142	0.4545	0.8751	0.2687	0.3827	0.5921	0.5492	0.4928
0.2927	11500	3.7377	2.1764	9.2284	7.7482	8.6753	13.2556	3.4186	8.2092	2.9631	8.2251	20.9522	1.8887	6.8783	1.1493	0.3184	0.5710	0.6896	0.3620	0.6411	0.4016	0.3141	0.4598	0.8871	0.2660	0.4035	0.5946	0.5680	0.4982
0.2991	11750	3.645	2.1757	9.2386	7.7988	8.4091	13.3105	3.3986	8.1770	3.0392	8.3319	20.8279	1.8464	7.2340	1.2369	0.3066	0.5680	0.6938	0.3413	0.6498	0.4035	0.3119	0.4692	0.8776	0.2697	0.4018	0.5937	0.5725	0.4969
0.3054	12000	3.8302	2.1730	9.2542	7.8031	8.3773	13.3083	3.4245	8.2024	2.9451	8.4451	20.3890	1.8992	7.1868	1.3149	0.3022	0.5639	0.7047	0.3580	0.6585	0.3953	0.3193	0.4512	0.8757	0.2667	0.4016	0.5815	0.5653	0.4957
0.3118	12250	3.7341	2.1580	9.1449	7.7487	8.2782	13.4871	3.4325	8.1531	2.8524	8.0765	20.4420	1.8084	7.4004	1.1942	0.3217	0.5648	0.7022	0.3658	0.6597	0.4010	0.3204	0.4470	0.8778	0.2668	0.4062	0.5841	0.5764	0.4995
0.3182	12500	3.6937	2.2003	9.0298	7.7776	8.3088	13.3345	3.4198	8.0772	2.8139	8.7163	20.4754	1.8802	7.2714	1.2016	0.3086	0.5676	0.6930	0.3609	0.6532	0.4067	0.3228	0.4341	0.8737	0.2723	0.4014	0.5783	0.5799	0.4964
0.3245	12750	3.6917	2.1808	9.0285	7.7770	8.3062	13.5737	3.3953	8.1291	2.8232	8.3426	20.9837	1.9420	7.1199	1.2568	0.3288	0.5620	0.6855	0.3512	0.6575	0.4069	0.3309	0.4372	0.8906	0.2709	0.4106	0.5928	0.5874	0.5009
0.3309	13000	3.6376	2.1621	9.0474	7.7871	8.1721	13.4874	3.3904	8.1507	2.8109	8.5607	21.0805	1.9734	7.0553	1.3466	0.3250	0.5686	0.6850	0.3520	0.6630	0.4044	0.3258	0.4424	0.8666	0.2684	0.4038	0.5769	0.5725	0.4965
0.3373	13250	3.7786	2.1146	9.1181	7.7333	8.2758	13.4782	3.3906	8.2021	2.8320	8.3097	21.1471	1.8529	7.3608	1.2242	0.3282	0.5649	0.6997	0.3761	0.6680	0.4097	0.3242	0.4143	0.8873	0.2784	0.3958	0.5956	0.5745	0.5013
0.3436	13500	3.8654	2.2053	9.0632	7.6973	8.4055	13.2312	3.3747	8.1627	2.8245	8.4075	20.2899	1.7553	7.1383	1.1577	0.3316	0.5672	0.6693	0.3901	0.6695	0.4017	0.3213	0.4138	0.8953	0.2703	0.4023	0.5856	0.5821	0.5000
0.3500	13750	3.7545	2.1424	9.0522	7.6998	8.3319	13.5322	3.3625	8.1303	2.8320	8.1860	20.4538	1.7997	7.0770	1.2512	0.3385	0.5654	0.6937	0.3856	0.6659	0.4025	0.3225	0.4156	0.8994	0.2675	0.4113	0.5886	0.5770	0.5026
0.3564	14000	3.715	2.1826	8.9396	7.6755	8.3111	13.3344	3.3331	8.1973	2.8760	8.8218	20.6306	1.9014	7.3386	1.2366	0.3256	0.5675	0.6903	0.3846	0.6706	0.4032	0.3300	0.4426	0.8876	0.2647	0.4082	0.5874	0.5770	0.5030
0.3627	14250	3.6348	2.1393	9.0032	7.7843	8.3942	13.2654	3.3368	8.1124	2.8874	8.5999	20.8261	1.8395	7.5311	1.1380	0.3339	0.5642	0.7175	0.3780	0.6600	0.3965	0.3229	0.4305	0.8915	0.2714	0.3914	0.5816	0.5732	0.5009
0.3691	14500	3.604	2.1709	8.9648	7.7166	8.4281	13.5507	3.3138	8.1362	2.8989	8.3940	20.6827	1.9298	7.3468	1.2705	0.3416	0.5594	0.7198	0.3776	0.6605	0.3965	0.3295	0.4341	0.8838	0.2632	0.3995	0.5884	0.5693	0.5018
0.3754	14750	3.5398	2.1451	9.0195	7.7407	8.4766	13.6169	3.2839	8.1386	2.9178	8.2585	21.1674	1.9484	7.5143	1.2747	0.3418	0.5592	0.6971	0.3763	0.6539	0.4060	0.3264	0.4214	0.8805	0.2524	0.3855	0.5872	0.5735	0.4970
0.3818	15000	3.7153	2.1038	8.9847	7.7279	8.2667	13.4033	3.2862	8.1232	2.8999	8.3224	20.9361	1.9569	7.3440	1.3036	0.3407	0.5687	0.7066	0.3497	0.6532	0.4043	0.3317	0.4300	0.8830	0.2564	0.3884	0.6013	0.5752	0.4992
0.3882	15250	3.752	2.0765	9.0293	7.7275	8.3168	13.2782	3.3105	8.0286	2.8404	8.2745	20.9582	1.7465	7.5436	1.2770	0.3313	0.5710	0.6945	0.3742	0.6699	0.4022	0.3327	0.4277	0.8903	0.2640	0.3872	0.5821	0.5662	0.4995
0.3945	15500	3.7794	2.0409	9.0264	7.7501	8.3903	13.3169	3.3058	8.0128	2.8463	8.5307	21.5663	1.7179	7.7448	1.1827	0.3286	0.5684	0.6961	0.3692	0.6743	0.4031	0.3267	0.4267	0.8927	0.2549	0.3777	0.5810	0.5686	0.4975
0.4009	15750	3.7444	2.0122	9.0629	7.7541	8.0134	13.2091	3.2956	8.0839	2.8528	8.1722	20.9378	1.7628	7.8655	1.2965	0.3282	0.5744	0.6839	0.3807	0.6644	0.4032	0.3277	0.4451	0.8896	0.2706	0.3916	0.5851	0.5685	0.5010
0.4073	16000	3.7817	2.0448	9.1787	7.7705	8.0529	13.1694	3.3128	8.1419	2.8104	8.2099	21.0454	1.7436	7.2934	1.2463	0.3137	0.5640	0.6884	0.3692	0.6600	0.3978	0.3215	0.4314	0.8937	0.2719	0.4094	0.6031	0.5697	0.4995
0.4136	16250	3.7293	2.0586	9.2379	7.7514	8.1877	13.1981	3.2983	8.0763	2.8564	8.6500	20.9279	1.8403	7.2051	1.1732	0.3405	0.5648	0.6944	0.3620	0.6614	0.3953	0.3286	0.4245	0.8921	0.2698	0.3946	0.5915	0.5770	0.4997
0.4200	16500	3.6243	2.0477	9.1718	7.6943	7.9493	13.3019	3.2908	8.0963	2.8306	8.7436	20.6790	1.8745	7.4356	1.1781	0.3272	0.5657	0.6944	0.3726	0.6848	0.3974	0.3318	0.4274	0.8939	0.2636	0.3948	0.5950	0.5746	0.5018
0.4263	16750	3.5071	2.0483	9.2054	7.7004	8.1887	13.3662	3.2727	7.9229	2.8256	8.2771	21.1584	1.8469	7.6476	1.2131	0.3318	0.5660	0.7000	0.3801	0.6902	0.3938	0.3224	0.4191	0.8969	0.2643	0.4052	0.6106	0.5684	0.5038
0.4327	17000	3.6337	2.0383	9.1228	7.7337	8.2262	13.2250	3.2714	7.9983	2.7662	8.4949	20.8407	1.8184	7.7876	1.2807	0.3251	0.5595	0.6957	0.3897	0.6697	0.3933	0.3123	0.4334	0.8934	0.2663	0.3935	0.5931	0.5700	0.4996
0.4391	17250	3.5075	2.0327	8.9777	7.7194	8.2392	13.4002	3.2678	7.9239	2.7551	8.2470	21.1674	1.7744	7.9402	1.3115	0.3418	0.5595	0.7123	0.3790	0.6625	0.3978	0.3174	0.4232	0.8866	0.2683	0.3926	0.5876	0.5764	0.5004
0.4454	17500	3.6595	2.0419	8.9509	7.6536	8.3099	13.3537	3.2766	7.9939	2.7604	8.3880	20.8993	1.8358	7.6156	1.2238	0.3311	0.5643	0.7049	0.3564	0.6686	0.3932	0.3099	0.4350	0.8872	0.2687	0.3826	0.5854	0.5665	0.4964
0.4518	17750	3.5743	2.0049	9.0019	7.6693	8.3489	13.3261	3.2758	8.0051	2.7881	8.4247	20.8115	1.8714	7.7491	1.1884	0.3341	0.5614	0.6972	0.3571	0.6625	0.3893	0.3036	0.4368	0.8911	0.2625	0.3861	0.5734	0.5681	0.4941
0.4582	18000	3.6038	1.9810	9.0106	7.7162	8.3584	13.1195	3.2674	8.0266	2.8184	8.4383	20.5199	1.8854	7.8663	1.1762	0.3318	0.5583	0.7086	0.3591	0.6509	0.3878	0.3137	0.4247	0.8831	0.2601	0.3921	0.5978	0.5648	0.4948
0.4645	18250	3.6903	2.0005	9.0187	7.6743	8.4280	13.0108	3.2733	7.8845	2.7810	8.3511	20.1457	1.7802	8.0015	1.0885	0.3407	0.5657	0.7033	0.3626	0.6644	0.3841	0.3299	0.4358	0.8844	0.2642	0.3904	0.5871	0.5632	0.4981
0.4709	18500	3.7208	1.9972	9.1020	7.6472	8.1589	13.0717	3.2601	8.0039	2.7673	8.3361	20.0231	1.8054	7.7381	1.1832	0.3328	0.5652	0.7043	0.3473	0.6693	0.3810	0.3313	0.4293	0.8770	0.2633	0.3946	0.5914	0.5634	0.4962
0.4773	18750	3.6357	2.0069	9.1473	7.6843	8.2110	13.1578	3.2540	7.9856	2.7390	8.6913	20.3263	1.8252	7.9545	1.0354	0.3285	0.5631	0.7093	0.3648	0.6685	0.3842	0.3285	0.4361	0.8918	0.2744	0.4065	0.5814	0.5610	0.4998
0.4836	19000	3.5737	1.9755	9.1397	7.6784	8.2604	13.3462	3.2391	7.9876	2.7643	8.4540	20.2047	1.7528	7.6572	1.0906	0.3284	0.5631	0.7083	0.3657	0.6660	0.3795	0.3186	0.4393	0.8876	0.2613	0.4056	0.5835	0.5637	0.4977
0.4900	19250	3.5325	2.0232	9.1987	7.6517	8.2727	13.1941	3.2276	7.9841	2.7238	8.4698	19.9076	1.8015	7.2144	1.1134	0.3337	0.5635	0.7076	0.3570	0.6572	0.3891	0.3297	0.4360	0.8864	0.2689	0.3997	0.5813	0.5608	0.4978
0.4963	19500	3.4782	2.0033	9.1235	7.7026	8.3383	13.1817	3.2308	8.0122	2.6934	8.4448	19.7121	1.7192	6.9417	0.9934	0.3222	0.5656	0.7009	0.3562	0.6654	0.3792	0.3292	0.4544	0.8803	0.2703	0.4027	0.5892	0.5611	0.4982
0.5027	19750	3.7141	1.9830	9.2166	7.6724	8.3188	13.1514	3.2465	8.0979	2.7005	8.1956	19.8289	1.6910	7.1995	1.0668	0.3323	0.5632	0.7209	0.3442	0.6745	0.3864	0.3287	0.4255	0.8828	0.2751	0.4097	0.5785	0.5597	0.4986
0.5091	20000	3.7058	1.9625	9.1729	7.6554	8.3323	13.0269	3.2368	8.0075	2.7159	8.6974	20.1650	1.6862	7.3387	1.1706	0.3414	0.5637	0.7214	0.3498	0.6745	0.3817	0.3309	0.4290	0.8861	0.2711	0.3861	0.5854	0.5757	0.4998
0.5154	20250	3.502	1.9983	9.1304	7.6354	8.3832	13.2376	3.2262	7.9229	2.7119	8.7638	20.0759	1.6633	6.9686	1.1490	0.3296	0.5703	0.7116	0.3409	0.6717	0.3847	0.3294	0.4198	0.8859	0.2655	0.3924	0.5836	0.5647	0.4962
0.5218	20500	3.6424	1.9867	9.0935	7.6470	8.4280	13.0866	3.2338	7.9608	2.7225	8.4571	20.1605	1.6184	6.8877	1.1519	0.3377	0.5682	0.7128	0.3479	0.6646	0.3900	0.3255	0.4408	0.8863	0.2655	0.4184	0.5862	0.5627	0.5005
0.5282	20750	3.6085	1.9589	9.1559	7.6282	8.3675	13.1289	3.2393	8.0050	2.7199	8.3705	20.4945	1.6721	6.8649	1.2063	0.3328	0.5719	0.7054	0.3526	0.6755	0.3974	0.3280	0.4204	0.8926	0.2686	0.4145	0.5783	0.5553	0.4995
0.5345	21000	3.5763	1.9392	9.1139	7.6917	8.2745	13.3345	3.2333	7.9848	2.7010	8.4815	20.5463	1.7049	7.0751	1.1276	0.3251	0.5710	0.7128	0.3561	0.6642	0.4022	0.3225	0.4394	0.8997	0.2737	0.4106	0.5732	0.5592	0.5008
0.5409	21250	3.5401	1.9744	9.0583	7.5955	8.2868	13.2877	3.2309	7.9686	2.6641	8.2829	20.3974	1.6843	6.9287	1.0128	0.3322	0.5692	0.7117	0.3348	0.6715	0.3834	0.3269	0.4372	0.8869	0.2662	0.4097	0.5797	0.5587	0.4975
0.5473	21500	3.489	1.9417	8.9543	7.6468	8.3612	13.3847	3.2314	7.9631	2.6635	8.4673	20.5367	1.7459	6.7037	1.0989	0.3328	0.5741	0.7097	0.3508	0.6660	0.3916	0.3277	0.4391	0.8860	0.2632	0.4100	0.5910	0.5578	0.5000
0.5536	21750	3.555	1.9533	8.9916	7.6360	8.3586	13.3764	3.2284	7.9575	2.7214	8.3429	20.5891	1.7569	6.6890	1.1157	0.3339	0.5744	0.7114	0.3392	0.6636	0.3964	0.3259	0.4504	0.8925	0.2651	0.4031	0.5792	0.5646	0.5000
0.5600	22000	3.586	1.9301	8.9806	7.6738	8.3535	13.2453	3.2316	7.9946	2.6907	8.2869	20.5606	1.6573	6.8912	1.1068	0.3239	0.5788	0.7121	0.3537	0.6589	0.4023	0.3243	0.4499	0.8891	0.2644	0.4083	0.5802	0.5616	0.5006
0.5663	22250	3.5084	1.9406	9.0040	7.6810	8.3815	13.1792	3.2167	7.9700	2.7316	8.6409	20.6425	1.6145	6.8099	1.1404	0.3174	0.5772	0.7177	0.3519	0.6613	0.3950	0.3324	0.4362	0.8894	0.2648	0.4173	0.5857	0.5626	0.5007
0.5727	22500	3.5095	1.9202	8.9619	7.6717	8.3825	13.1359	3.2147	7.9776	2.7114	8.2541	20.5107	1.6855	6.9227	1.1935	0.3335	0.5726	0.7134	0.3574	0.6578	0.3959	0.3262	0.4401	0.8968	0.2604	0.4105	0.5826	0.5620	0.5007
0.5791	22750	3.5059	1.9225	8.9956	7.6775	8.3968	13.0329	3.2114	7.9836	2.7107	8.4843	20.6793	1.6821	7.0987	1.0579	0.3273	0.5703	0.7118	0.3689	0.6490	0.3850	0.3257	0.4320	0.8824	0.2622	0.4119	0.5898	0.5605	0.4982
0.5854	23000	3.4047	1.9716	9.0017	7.6435	8.4379	13.0467	3.1957	7.9843	2.7017	8.5995	20.8783	1.5818	7.1997	1.0067	0.3272	0.5718	0.7132	0.3728	0.6544	0.3913	0.3255	0.4377	0.8869	0.2583	0.4107	0.5916	0.5539	0.4996
0.5918	23250	3.4732	1.9518	8.9827	7.6143	8.3925	13.2800	3.1877	7.9616	2.7122	8.5013	21.0376	1.6291	7.0831	1.0816	0.3316	0.5689	0.7129	0.3705	0.6536	0.3847	0.3211	0.4495	0.8844	0.2622	0.4141	0.5916	0.5551	0.5000
0.5982	23500	3.4271	1.9688	9.0092	7.5830	8.3763	13.2765	3.1857	7.9625	2.6794	8.4899	20.8080	1.6519	7.1604	1.1423	0.3346	0.5716	0.7054	0.3637	0.6562	0.3844	0.3249	0.4346	0.8912	0.2577	0.4009	0.5872	0.5697	0.4986
0.6045	23750	3.4701	2.0238	9.0036	7.5173	8.4058	13.2881	3.1791	7.9275	2.7149	8.8465	20.6630	1.7025	7.2286	1.1973	0.3308	0.5670	0.7040	0.3754	0.6609	0.3936	0.3252	0.4475	0.8948	0.2542	0.4062	0.5877	0.5717	0.5015
0.6109	24000	3.6199	2.0084	8.9869	7.5146	8.3790	13.1350	3.1771	7.9137	2.7032	8.5424	20.5441	1.7652	6.8017	1.1617	0.3224	0.5656	0.7045	0.3643	0.6643	0.3864	0.3174	0.4577	0.8873	0.2571	0.3735	0.5939	0.5611	0.4966
0.6173	24250	3.408	2.0137	9.0355	7.5305	8.3597	13.1604	3.1735	7.9201	2.7078	9.0787	20.5226	1.7341	6.7525	1.0237	0.3310	0.5676	0.7061	0.3643	0.6602	0.3995	0.3240	0.4588	0.8936	0.2572	0.3868	0.5928	0.5718	0.5011
0.6236	24500	3.4651	1.9564	8.9636	7.5635	8.3731	13.3261	3.1737	7.9529	2.6638	8.5231	20.4787	1.7792	6.7317	1.0516	0.3291	0.5670	0.7060	0.3572	0.6594	0.3982	0.3253	0.4400	0.8917	0.2577	0.3886	0.5951	0.5662	0.4986
0.6300	24750	3.6988	1.9668	8.9583	7.5875	8.4316	12.9644	3.1769	7.9801	2.6750	8.5780	20.3883	1.7045	6.9010	1.0783	0.3300	0.5668	0.7040	0.3653	0.6550	0.3957	0.3293	0.4570	0.8878	0.2621	0.3816	0.5932	0.5665	0.4996
0.6363	25000	3.4365	1.9782	8.9306	7.5829	8.4123	13.0937	3.1640	7.9870	2.6806	8.6281	20.2045	1.7244	6.9685	1.0365	0.3289	0.5651	0.6956	0.3703	0.6494	0.4025	0.3294	0.4511	0.8841	0.2634	0.3894	0.5970	0.5605	0.4990
0.6427	25250	3.6097	1.9653	8.9386	7.5722	8.4185	13.0207	3.1629	7.9910	2.6867	8.7326	20.2092	1.7321	6.8303	1.0569	0.3273	0.5643	0.7024	0.3587	0.6534	0.4018	0.3282	0.4401	0.8872	0.2614	0.3860	0.5966	0.5703	0.4983
0.6491	25500	3.5379	1.9518	8.9189	7.5241	8.4156	13.0924	3.1557	7.9747	2.6468	8.6005	20.3848	1.7474	6.9006	1.0193	0.3186	0.5692	0.7032	0.3581	0.6429	0.4068	0.3288	0.4426	0.8937	0.2584	0.4055	0.5963	0.5667	0.4993
0.6554	25750	3.6223	1.9609	8.9459	7.5554	8.4227	12.8817	3.1483	7.9813	2.6559	8.8423	20.3294	1.6381	6.8535	1.0140	0.3252	0.5722	0.7104	0.3689	0.6554	0.4087	0.3291	0.4293	0.8894	0.2632	0.3940	0.5995	0.5702	0.5012
0.6618	26000	3.402	1.9602	8.9366	7.5236	8.2409	12.9294	3.1424	7.9619	2.6323	8.6640	20.2959	1.6627	6.7331	1.0115	0.3276	0.5676	0.7031	0.3652	0.6493	0.4030	0.3263	0.4298	0.8830	0.2655	0.4030	0.6017	0.5724	0.4998
0.6682	26250	3.4876	1.9929	8.9621	7.4947	8.1977	12.8814	3.1350	7.8663	2.6554	8.5629	20.3156	1.7019	6.7602	0.9582	0.3253	0.5639	0.7127	0.3621	0.6544	0.4146	0.3278	0.4350	0.8895	0.2613	0.4056	0.5997	0.5635	0.5012
0.6745	26500	3.4301	1.9752	8.9474	7.5246	8.1379	12.7941	3.1369	7.8999	2.6256	8.5223	20.4417	1.6627	6.7999	0.9626	0.3232	0.5661	0.7094	0.3719	0.6474	0.4077	0.3299	0.4533	0.8849	0.2620	0.3993	0.6008	0.5665	0.5017
0.6809	26750	3.482	1.9646	8.9353	7.5199	8.2066	12.6512	3.1284	7.9080	2.6464	8.6207	20.4063	1.6852	6.7490	1.0089	0.3172	0.5689	0.7121	0.3734	0.6461	0.4089	0.3287	0.4516	0.8854	0.2601	0.3965	0.5993	0.5634	0.5009
0.6873	27000	3.5073	1.9431	8.9478	7.5188	8.0549	12.6486	3.1310	7.9241	2.6508	8.5172	20.4459	1.6855	6.8390	1.0012	0.3222	0.5705	0.6936	0.3675	0.6511	0.3970	0.3286	0.4467	0.8875	0.2587	0.3986	0.6177	0.5690	0.5007
0.6936	27250	3.5565	1.9438	8.9610	7.5044	8.0383	12.4879	3.1277	7.9219	2.6321	8.4003	20.6229	1.7421	6.7256	1.0533	0.3327	0.5737	0.6955	0.3653	0.6486	0.4067	0.3277	0.4368	0.8901	0.2589	0.3877	0.6135	0.5696	0.5005
0.7000	27500	3.4506	1.9639	8.9673	7.4644	8.1455	12.4523	3.1171	7.9159	2.6772	8.5339	20.7734	1.7690	6.6677	1.0708	0.3260	0.5730	0.6960	0.3562	0.6419	0.4053	0.3310	0.4395	0.8871	0.2594	0.4010	0.6155	0.5679	0.5000
0.7063	27750	3.4875	1.9177	8.9276	7.4754	8.1375	12.5163	3.1315	7.8697	2.6247	8.4426	20.4950	1.7047	6.6303	1.0030	0.3244	0.5690	0.6887	0.3499	0.6471	0.4057	0.3309	0.4464	0.8893	0.2570	0.4038	0.6038	0.5701	0.4989
0.7127	28000	3.5298	1.8939	8.9022	7.4889	8.1322	12.5461	3.1409	7.8621	2.5963	8.4173	20.4948	1.7078	6.4516	0.9874	0.3380	0.5721	0.6934	0.3631	0.6439	0.4046	0.3272	0.4486	0.8893	0.2593	0.4020	0.6017	0.5687	0.5009
0.7191	28250	3.3329	1.8970	8.9172	7.5001	8.1758	12.5515	3.1333	7.9075	2.5928	8.5025	20.3447	1.6879	6.5795	0.9658	0.3247	0.5729	0.7000	0.3710	0.6468	0.4069	0.3336	0.4564	0.8933	0.2637	0.4105	0.6012	0.5749	0.5043
0.7254	28500	3.3897	1.8966	8.9038	7.5265	8.2199	12.6173	3.1244	7.8882	2.5974	8.3728	20.3628	1.6804	6.6887	0.9733	0.3278	0.5750	0.7001	0.3645	0.6528	0.4145	0.3310	0.4669	0.8964	0.2638	0.4032	0.6086	0.5713	0.5058
0.7318	28750	3.4588	1.8934	8.8948	7.5026	8.2374	12.4866	3.1287	7.8941	2.5811	8.3781	20.4334	1.7109	6.6195	0.9699	0.3251	0.5707	0.7083	0.3702	0.6411	0.4086	0.3252	0.4553	0.8935	0.2641	0.4068	0.6063	0.5592	0.5026
0.7382	29000	3.3675	1.8959	8.8925	7.5043	8.2628	12.6063	3.1174	7.9132	2.5908	8.2436	20.3771	1.6740	6.7151	0.9895	0.3262	0.5725	0.7069	0.3694	0.6495	0.4063	0.3169	0.4622	0.8962	0.2609	0.4124	0.6101	0.5635	0.5041
0.7445	29250	3.3886	1.8976	8.9215	7.4975	8.2401	12.4978	3.1127	7.8615	2.5814	8.2715	20.4379	1.6740	6.7705	0.9492	0.3269	0.5683	0.7061	0.3760	0.6554	0.4089	0.3209	0.4508	0.8982	0.2658	0.3982	0.6053	0.5668	0.5037
0.7509	29500	3.4826	1.8855	8.9320	7.5047	8.2470	12.5056	3.1168	7.8967	2.5742	8.3762	20.4917	1.7122	6.3650	0.9882	0.3380	0.5681	0.7061	0.3727	0.6521	0.4071	0.3222	0.4771	0.8963	0.2701	0.3943	0.6010	0.5624	0.5052
0.7572	29750	3.4268	1.8725	8.9332	7.5158	8.2481	12.4708	3.1146	7.9010	2.5617	8.2478	20.3435	1.6860	6.4246	0.9801	0.3331	0.5691	0.7061	0.3767	0.6523	0.4068	0.3195	0.4589	0.9097	0.2627	0.4113	0.6059	0.5641	0.5059
0.7636	30000	3.2621	1.8813	8.9125	7.5203	8.2736	12.5555	3.1094	7.9083	2.5488	8.3690	20.3392	1.7015	6.5219	0.9901	0.3366	0.5686	0.6941	0.3696	0.6465	0.4062	0.3228	0.4570	0.8967	0.2657	0.3975	0.6054	0.5702	0.5028
0.7700	30250	3.3289	1.8893	8.9211	7.5322	8.1753	12.4890	3.1063	7.9272	2.5320	8.4628	20.3169	1.6841	6.5986	0.9798	0.3302	0.5719	0.6944	0.3745	0.6471	0.4007	0.3172	0.4741	0.9046	0.2685	0.4126	0.6014	0.5666	0.5049
0.7763	30500	3.5363	1.8836	8.8909	7.5323	8.2197	12.4000	3.1064	7.9088	2.5494	8.3271	20.3110	1.7132	6.4502	0.9974	0.3298	0.5705	0.6995	0.3671	0.6511	0.4057	0.3204	0.4560	0.8949	0.2613	0.4153	0.6011	0.5738	0.5036
0.7827	30750	3.3557	1.8824	8.9002	7.5249	8.2047	12.4618	3.1055	7.9265	2.5501	8.2708	20.2254	1.7222	6.4927	0.9813	0.3333	0.5678	0.6926	0.3781	0.6575	0.4005	0.3225	0.4497	0.8991	0.2649	0.4018	0.6051	0.5698	0.5033
0.7891	31000	3.4481	1.8725	8.9077	7.5043	8.2095	12.5197	3.1095	7.9124	2.5216	8.1396	20.0618	1.6962	6.4808	0.9764	0.3321	0.5691	0.6941	0.3650	0.6464	0.4013	0.3239	0.4522	0.9035	0.2657	0.4058	0.6041	0.5639	0.5021
0.7954	31250	3.3888	1.8596	8.9249	7.5248	8.2429	12.4622	3.1054	7.9247	2.5351	8.3011	19.9880	1.7062	6.5598	0.9628	0.3428	0.5694	0.7008	0.3761	0.6564	0.4013	0.3223	0.4589	0.9024	0.2633	0.4111	0.6052	0.5670	0.5059
0.8018	31500	3.3989	1.8662	8.9299	7.5206	8.1957	12.4667	3.1043	7.9089	2.5465	8.2149	19.8897	1.6830	6.6131	0.9204	0.3452	0.5658	0.7002	0.3709	0.6570	0.4004	0.3233	0.4588	0.9015	0.2619	0.4044	0.6015	0.5654	0.5043
0.8082	31750	3.3603	1.8671	8.9158	7.5135	8.1945	12.5201	3.1009	7.9130	2.5539	8.3876	19.9818	1.6685	6.6359	0.9307	0.3516	0.5659	0.6928	0.3767	0.6534	0.4004	0.3245	0.4627	0.9021	0.2619	0.4252	0.6019	0.5699	0.5068
0.8145	32000	3.3783	1.8588	8.8993	7.4994	8.2058	12.5200	3.0980	7.8902	2.5448	8.3444	19.9960	1.6815	6.7156	0.9509	0.3358	0.5665	0.7012	0.3759	0.6551	0.4002	0.3259	0.4505	0.8960	0.2621	0.4113	0.6059	0.5716	0.5045
0.8209	32250	3.4379	1.8693	8.8896	7.4639	8.0939	12.5840	3.0894	7.8559	2.5590	8.4227	19.9565	1.6855	6.6835	0.9650	0.3433	0.5663	0.7004	0.3717	0.6527	0.4076	0.3255	0.4491	0.8990	0.2643	0.4127	0.5984	0.5741	0.5050
0.8272	32500	3.4962	1.8838	8.8634	7.4636	8.1354	12.5144	3.0866	7.8561	2.5410	8.4169	20.0794	1.6625	6.6897	0.9849	0.3461	0.5678	0.7024	0.3808	0.6564	0.4083	0.3200	0.4531	0.9015	0.2622	0.4116	0.6061	0.5685	0.5065
0.8336	32750	3.4927	1.8696	8.8655	7.4274	8.0888	12.5431	3.0874	7.8320	2.5324	8.3181	20.0095	1.6403	6.7182	0.9837	0.3439	0.5657	0.7078	0.3771	0.6580	0.4078	0.3210	0.4543	0.9087	0.2625	0.4184	0.6095	0.5691	0.5080
0.8400	33000	3.33	1.8743	8.8776	7.4438	8.1127	12.5571	3.0793	7.8474	2.5423	8.3195	20.1064	1.6546	6.7761	1.0118	0.3400	0.5684	0.6938	0.3760	0.6560	0.4077	0.3248	0.4586	0.8965	0.2613	0.4085	0.6094	0.5704	0.5055
0.8463	33250	3.3431	1.8687	8.8074	7.4577	8.1313	12.5392	3.0816	7.8629	2.5306	8.3183	20.1216	1.6653	6.8390	1.0199	0.3233	0.5685	0.7031	0.3770	0.6493	0.4075	0.3243	0.4586	0.9015	0.2600	0.4068	0.6050	0.5659	0.5039
0.8527	33500	3.4455	1.8618	8.8108	7.4587	8.1421	12.5672	3.0800	7.8533	2.5364	8.2589	20.0766	1.6537	6.8825	1.0161	0.3346	0.5647	0.7003	0.3761	0.6523	0.4075	0.3239	0.4598	0.9088	0.2641	0.4134	0.5976	0.5702	0.5056
0.8591	33750	3.3189	1.8578	8.8046	7.4684	8.1488	12.6076	3.0817	7.8637	2.5249	8.2008	20.1733	1.6486	6.9228	1.0004	0.3428	0.5684	0.7019	0.3782	0.6483	0.4049	0.3229	0.4525	0.9039	0.2641	0.4088	0.6114	0.5669	0.5058
0.8654	34000	3.3815	1.8587	8.7921	7.4675	8.0968	12.6359	3.0800	7.8680	2.5266	8.3492	20.1037	1.6352	6.8485	1.0061	0.3412	0.5655	0.6945	0.3806	0.6510	0.4061	0.3178	0.4583	0.9085	0.2640	0.4078	0.6105	0.5689	0.5058
0.8718	34250	3.3381	1.8586	8.7828	7.4672	8.1115	12.6341	3.0783	7.8743	2.5275	8.3363	20.0616	1.6445	6.8898	1.0146	0.3255	0.5645	0.6941	0.3784	0.6488	0.4027	0.3190	0.4518	0.9039	0.2637	0.4079	0.6115	0.5730	0.5034
0.8782	34500	3.3992	1.8597	8.7906	7.4658	8.1316	12.6647	3.0781	7.8643	2.5249	8.3280	20.0237	1.6348	6.8277	1.0203	0.3283	0.5645	0.6925	0.3744	0.6525	0.4027	0.3250	0.4593	0.9040	0.2625	0.4097	0.6121	0.5676	0.5042
0.8845	34750	3.2951	1.8608	8.8033	7.4729	8.1063	12.6896	3.0719	7.8671	2.5243	8.3368	20.0587	1.6455	6.7907	1.0106	0.3274	0.5651	0.6925	0.3738	0.6527	0.4045	0.3219	0.4592	0.9041	0.2619	0.3985	0.6120	0.5679	0.5032
0.8909	35000	3.262	1.8677	8.8024	7.4317	8.1170	12.7225	3.0671	7.8519	2.5303	8.4655	20.1156	1.6405	6.7997	1.0141	0.3288	0.5667	0.6935	0.3738	0.6554	0.4049	0.3196	0.4532	0.9041	0.2626	0.4116	0.6123	0.5653	0.5040
0.8972	35250	3.3218	1.8698	8.7992	7.4495	8.1390	12.7198	3.0627	7.8528	2.5399	8.4549	20.1652	1.6288	6.7857	1.0261	0.3328	0.5666	0.6951	0.3686	0.6538	0.4043	0.3248	0.4596	0.9042	0.2646	0.4097	0.6123	0.5725	0.5053
0.9036	35500	3.3529	1.8633	8.7964	7.4339	8.1347	12.7295	3.0657	7.8262	2.5256	8.4119	20.1363	1.6165	6.7987	1.0260	0.3268	0.5660	0.6951	0.3695	0.6543	0.4042	0.3232	0.4511	0.9068	0.2666	0.4047	0.6112	0.5701	0.5038
0.9100	35750	3.2205	1.8626	8.7751	7.4335	8.1346	12.7532	3.0654	7.8318	2.5077	8.3883	20.0738	1.5839	6.7819	1.0282	0.3291	0.5659	0.6951	0.3702	0.6530	0.4045	0.3241	0.4493	0.9067	0.2622	0.4192	0.6112	0.5710	0.5047
0.9163	36000	3.3671	1.8521	8.7693	7.4500	8.1155	12.8041	3.0671	7.8451	2.4994	8.3053	20.0666	1.5984	6.7696	1.0377	0.3383	0.5667	0.6946	0.3739	0.6536	0.4049	0.3262	0.4457	0.9065	0.2625	0.4143	0.6112	0.5719	0.5054
0.9227	36250	3.4074	1.8555	8.7683	7.4607	8.1465	12.7545	3.0621	7.8488	2.5080	8.3816	20.0420	1.6043	6.7991	1.0539	0.3367	0.5685	0.6935	0.3714	0.6523	0.4068	0.3237	0.4471	0.9065	0.2645	0.4176	0.6110	0.5692	0.5053
0.9291	36500	3.4806	1.8543	8.7833	7.4584	8.1070	12.7444	3.0605	7.8475	2.5166	8.3809	19.3340	1.6181	6.7663	1.0555	0.3271	0.5693	0.6935	0.3733	0.6560	0.3990	0.3229	0.4530	0.9065	0.2618	0.4105	0.6129	0.5654	0.5039
0.9354	36750	3.4848	1.8558	8.7967	7.4512	8.1175	12.6960	3.0616	7.8420	2.5137	8.4267	19.4008	1.6181	6.7376	1.0629	0.3284	0.5697	0.6922	0.3723	0.6530	0.4048	0.3234	0.4534	0.9065	0.2641	0.4088	0.6115	0.5635	0.5040
0.9418	37000	3.1947	1.8574	8.8026	7.4555	8.1222	12.7155	3.0591	7.8391	2.5180	8.4551	19.4094	1.6134	6.7200	1.0657	0.3318	0.5681	0.6922	0.3711	0.6550	0.4043	0.3234	0.4534	0.9037	0.2645	0.4090	0.6113	0.5685	0.5043
0.9482	37250	3.3557	1.8569	8.7972	7.4535	8.1200	12.7427	3.0585	7.8460	2.5167	8.5049	19.4395	1.6141	6.7080	1.0700	0.3308	0.5699	0.6930	0.3708	0.6513	0.4048	0.3247	0.4589	0.9039	0.2645	0.4087	0.6113	0.5677	0.5046
0.9545	37500	3.369	1.8580	8.7958	7.4558	8.1193	12.7407	3.0579	7.8509	2.5139	8.5231	19.4691	1.6139	6.7228	1.0696	0.3388	0.5683	0.6922	0.3696	0.6556	0.4050	0.3236	0.4460	0.8965	0.2624	0.4145	0.6113	0.5682	0.5040
0.9609	37750	3.398	1.8564	8.7955	7.4567	8.1174	12.7478	3.0589	7.8415	2.5086	8.5030	19.4719	1.5989	6.7117	1.0689	0.3381	0.5678	0.6922	0.3652	0.6544	0.4049	0.3234	0.4460	0.9039	0.2599	0.3998	0.6113	0.5658	0.5025
0.9672	38000	3.3699	1.8534	8.8011	7.4604	8.1135	12.7689	3.0574	7.8412	2.5097	8.4866	19.4973	1.5966	6.7309	1.0655	0.3372	0.5680	0.6922	0.3651	0.6553	0.4038	0.3224	0.4560	0.9025	0.2675	0.4061	0.6111	0.5712	0.5045
0.9736	38250	3.4483	1.8540	8.8045	7.4506	8.0942	12.7725	3.0569	7.8379	2.5135	8.4817	19.5038	1.5995	6.7289	1.0693	0.3378	0.5682	0.6922	0.3724	0.6550	0.4038	0.3243	0.4534	0.9025	0.2645	0.4058	0.6111	0.5710	0.5048
0.9800	38500	3.254	1.8546	8.8026	7.4530	8.1004	12.7796	3.0559	7.8378	2.5086	8.4538	19.5061	1.5997	6.7390	1.0724	0.3381	0.5681	0.6922	0.3647	0.6547	0.4041	0.3242	0.4534	0.8951	0.2642	0.4064	0.6111	0.5677	0.5034
0.9863	38750	3.2759	1.8549	8.8024	7.4545	8.0996	12.7875	3.0562	7.8400	2.5054	8.4401	19.5192	1.6018	6.7433	1.0737	0.3382	0.5681	0.6922	0.3647	0.6547	0.4041	0.3241	0.4534	0.8951	0.2642	0.4078	0.6111	0.5689	0.5036
0.9927	39000	3.3273	1.8548	8.8017	7.4536	8.0908	12.7885	3.0557	7.8396	2.5052	8.4335	19.5169	1.6002	6.7439	1.0716	0.3382	0.5681	0.6922	0.3647	0.6547	0.4041	0.3242	0.4534	0.8951	0.2641	0.4078	0.6111	0.5703	0.5037
0.9991	39250	3.3902	1.8540	8.8010	7.4543	8.0872	12.7890	3.0561	7.8412	2.5040	8.3945	19.5165	1.5997	6.7462	1.0718	0.3309	0.5681	0.6922	0.3651	0.6547	0.4041	0.3242	0.4534	0.8951	0.2643	0.4078	0.6111	0.5703	0.5032
1.0	39287	-	-	-	-	-	-	-	-	-	-	-	-	-	-	0.3309	0.5681	0.6922	0.3651	0.6547	0.4041	0.3242	0.4534	0.8951	0.2643	0.4078	0.6111	0.5703	0.5032

Environmental Impact

Carbon emissions were measured using CodeCarbon.

Energy Consumed: 2.611 kWh
Carbon Emitted: 1.015 kg of CO2
Hours Used: 17.883 hours

Training Hardware

On Cloud: No
GPU Model: 1 x NVIDIA GeForce RTX 3090
CPU Model: 13th Gen Intel(R) Core(TM) i7-13700K
RAM Size: 31.78 GB

Framework Versions

Python: 3.11.6
Sentence Transformers: 3.3.0.dev0
Transformers: 4.45.2
PyTorch: 2.5.0.dev20240807+cu121
Accelerate: 1.0.0
Datasets: 2.20.0
Tokenizers: 0.20.1-dev.0

Citation

BibTeX

Sentence Transformers

bibtex

@inproceedings{reimers-2019-sentence-bert,
    title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks",
    author = "Reimers, Nils and Gurevych, Iryna",
    booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing",
    month = "11",
    year = "2019",
    publisher = "Association for Computational Linguistics",
    url = "https://arxiv.org/abs/1908.10084",
}

MatryoshkaLoss

bibtex

@misc{kusupati2024matryoshka,
    title={Matryoshka Representation Learning},
    author={Aditya Kusupati and Gantavya Bhatt and Aniket Rege and Matthew Wallingford and Aditya Sinha and Vivek Ramanujan and William Howard-Snyder and Kaifeng Chen and Sham Kakade and Prateek Jain and Ali Farhadi},
    year={2024},
    eprint={2205.13147},
    archivePrefix={arXiv},
    primaryClass={cs.LG}
}

MultipleNegativesRankingLoss

bibtex

@misc{henderson2017efficient,
    title={Efficient Natural Language Response Suggestion for Smart Reply},
    author={Matthew Henderson and Rami Al-Rfou and Brian Strope and Yun-hsuan Sung and Laszlo Lukacs and Ruiqi Guo and Sanjiv Kumar and Balint Miklos and Ray Kurzweil},
    year={2017},
    eprint={1705.00652},
    archivePrefix={arXiv},
    primaryClass={cs.CL}
}