This is a PyLate model finetuned from sentence-transformers/all-MiniLM-L6-v2. It maps sentences & paragraphs to sequences of 128-dimensional dense vectors and can be used for semantic textual similarity using the MaxSim operator.

Model Details

Model Description

Model Type: PyLate model
Base model: sentence-transformers/all-MiniLM-L6-v2
Document Length: 512 tokens
Query Length: 128 tokens
Output Dimensionality: 128 tokens
Similarity Function: MaxSim

Model Sources

Documentation: PyLate Documentation
Repository: PyLate on GitHub
Hugging Face: PyLate models on Hugging Face

Usage

First install the PyLate library:

pip install -U pylate

Retrieval

PyLate provides a streamlined interface to index and retrieve documents using ColBERT models. The index leverages the Voyager HNSW index to efficiently handle document embeddings and enable fast retrieval.

Indexing documents

First, load the ColBERT model and initialize the Voyager index, then encode and index your documents:

python

from pylate import indexes, models, retrieve

# Step 1: Load the ColBERT model
model = models.ColBERT(
    model_name_or_path=pylate_model_id,
)

# Step 2: Initialize the Voyager index
index = indexes.Voyager(
    index_folder="pylate-index",
    index_name="index",
    override=True,  # This overwrites the existing index if any
)

# Step 3: Encode the documents
documents_ids = ["1", "2", "3"]
documents = ["document 1 text", "document 2 text", "document 3 text"]

documents_embeddings = model.encode(
    documents,
    batch_size=32,
    is_query=False,  # Ensure that it is set to False to indicate that these are documents, not queries
    show_progress_bar=True,
)

# Step 4: Add document embeddings to the index by providing embeddings and corresponding ids
index.add_documents(
    documents_ids=documents_ids,
    documents_embeddings=documents_embeddings,
)

Note that you do not have to recreate the index and encode the documents every time. Once you have created an index and added the documents, you can re-use the index later by loading it:

python

# To load an index, simply instantiate it with the correct folder/name and without overriding it
index = indexes.Voyager(
    index_folder="pylate-index",
    index_name="index",
)

Retrieving top-k documents for queries

Once the documents are indexed, you can retrieve the top-k most relevant documents for a given set of queries. To do so, initialize the ColBERT retriever with the index you want to search in, encode the queries and then retrieve the top-k documents to get the top matches ids and relevance scores:

python

# Step 1: Initialize the ColBERT retriever
retriever = retrieve.ColBERT(index=index)

# Step 2: Encode the queries
queries_embeddings = model.encode(
    ["query for document 3", "query for document 1"],
    batch_size=32,
    is_query=True,  #  # Ensure that it is set to False to indicate that these are queries
    show_progress_bar=True,
)

# Step 3: Retrieve top-k documents
scores = retriever.retrieve(
    queries_embeddings=queries_embeddings,
    k=10,  # Retrieve the top 10 matches for each query
)

Reranking

If you only want to use the ColBERT model to perform reranking on top of your first-stage retrieval pipeline without building an index, you can simply use rank function and pass the queries and documents to rerank:

python

from pylate import rank, models

queries = [
    "query A",
    "query B",
]

documents = [
    ["document A", "document B"],
    ["document 1", "document C", "document B"],
]

documents_ids = [
    [1, 2],
    [1, 3, 2],
]

model = models.ColBERT(
    model_name_or_path=pylate_model_id,
)

queries_embeddings = model.encode(
    queries,
    is_query=True,
)

documents_embeddings = model.encode(
    documents,
    is_query=False,
)

reranked_documents = rank.rerank(
    documents_ids=documents_ids,
    queries_embeddings=queries_embeddings,
    documents_embeddings=documents_embeddings,
)

Evaluation

Metrics

Col BERTTriplet

Evaluated with pylate.evaluation.colbert_triplet.ColBERTTripletEvaluator

Metric	Value
accuracy	0.9513

Training Details

Training Dataset

Unnamed Dataset

Size: 310,935 training samples
Columns: query, positive, and negative
Approximate statistics based on the first 1000 samples:
query positive negative
type string string string
details
min: 38 tokens
mean: 97.92 tokens
max: 32 tokens
min: 84 tokens
mean: 128 tokens
max: 128 tokens
min: 84 tokens
mean: 127.7 tokens
max: 128 tokens

Samples:

query	positive	negative
`The primary objective of enacting a inheritance tax is to mitigate economic inequality and redistribute wealth among the poorer sections of society, although various empirical studies have demonstrated a lack of correlation between the two.`	`The principal goal of establishing estate duties as a form of taxation is not solely to address the problem of economic disparity, but more importantly, to redistribute wealth in an equitable manner so as to reduce the vast gap between the rich and the relatively poor segments of the population.`	`In a bid to abide by international agreements and world peaceful coexistence standards, most European nations have set up strict fiscal policies ensuring a strong relationship with neighboring countries, including strategic partnerships to promote tourism, as much as quotas to restrict immigration and asylum seekers.`
Usability Evaluation Report for the New Web Application[object Object]Introduction[object Object]This usability evaluation was conducted to identify issues related to user experience and provide recommendations for improving the overall usability of the new web application. The evaluation focused on the login and registration process, navigation, and search functionality.[object Object]Methodology[object Object]The evaluation consisted of user testing and heuristic evaluation. A total of five participants were recruited to participate in the user testing, and each participant was asked to complete several tasks using the web application. The participants' interactions with the application were observed and recorded. Heuristic evaluation was conducted based on a set of well-established usability principles to identify potential usability issues in the application's design and functionality.[object Object]Results[object Object]During the user testing, several usability issues were identified. These included difficulties in locating the login and registration features, p...	Design Document: Home and Landing Page Redesign for New Web Application[object Object]Executive Summary[object Object]As part of an ongoing effort to improve the user experience and engagement for the new web application, this project focuses on the redesign of the home and landing page. The new design will address usability issues identified in a previous evaluation, make the application more appealing to users, and help drive sales and conversions. The following report includes the design requirements, a full design specification, and guidance for implementation.[object Object]Goals and Objectives[object Object]The main goals of this project include: to redesign the home and landing pages to give users an improved first impression of the application; to improve task completion times and create a seamless user experience; to increase conversion rates by reducing bounce rates and making it easier for users to find the information they need.[object Object]Scope of Work[object Object]The redesign of the home and landing pages includes: creating a clear visual hierarchy ...	Designing Effective User Interfaces for Virtual Reality ApplicationsIntroductionVirtual reality (VR) technology has been rapidly advancing in recent years, with applications in various fields such as gaming, education, and healthcare. As VR continues to grow in popularity, the need for effective user interfaces has become increasingly important. A well-designed user interface can enhance the overall VR experience, while a poorly designed one can lead to frustration and disorientation.Principles of Effective VR User Interface Design1. Intuitive Interaction: The primary goal of a VR user interface is to provide an intuitive and natural way for users to interact with the virtual environment. This can be achieved through the use of gestures, voice commands, or other innovative methods.2. Visual Feedback: Visual feedback is crucial in VR, as it helps users understand the consequences of their actions. This can be in the form of animations, particles, or other visual effects that provide a c...
`The manager of the local conservation society recently explained measures for sustainable wildlife preservation.`	`The conservation society's manager recently explained measures for preserving wildlife sustainably.`	`After explaining university education requirements, the career counsellor also talked about wildlife preservation jobs.`

Loss: pylate.losses.contrastive.Contrastive

Evaluation Dataset

Unnamed Dataset

Size: 34,549 evaluation samples
Columns: query, positive, and negative
Approximate statistics based on the first 1000 samples:
query positive negative
type string string string
details
min: 38 tokens
mean: 98.32 tokens
max: 32 tokens
min: 84 tokens
mean: 127.37 tokens
max: 128 tokens
min: 84 tokens
mean: 127.12 tokens
max: 128 tokens

Samples:

query	positive	negative
`In a magical forest, there lived a group of animals that loved to dance under the stars. They danced to the rhythm of the crickets and felt the magic of the night.`	`In a magical forest, there lived a group of animals that loved to dance under the stars on a lovely night. They danced to the rhythm of the crickets.`	`The forest was a wonderful place where animals could sing and dance to the sounds of nature. Some liked the rustling of leaves, while others liked the buzzing of bees. But they all loved the music of a babbling brook.`
`Given this reasoning-intensive query, find relevant documents that could help answer the question.`	`food_percent/2063AApplicationsLeontiefModels_149.txt`	The use of matrix equations in computer graphics is gaining significant attention in recent years. In computer-aided design (CAD), matrix equations play a crucial role in transforming 2D and 3D objects. For instance, when designing a car model, the CAD software uses matrix equations to rotate, translate, and scale the object. The transformation matrix is a 4x4 matrix that stores the coordinates of the object and performs the required operations. Similarly, in computer gaming, matrix equations are used to animate characters and objects in 3D space. The game developers use transformation matrices to create realistic movements and interactions between objects. However, the complexity of these transformations leads to a high computational cost, making it difficult to achieve real-time rendering. To address this challenge, researchers are exploring the use of machine learning algorithms to optimize the transformation process. For example, a research paper titled 'Matrix Equation-Based 6-DoF...
`A study found that the use of virtual reality in therapy sessions can have a positive effect on mental health by reducing stress and anxiety.`	`A therapy session using virtual reality can significantly reduce patient stress and anxiety.`	`Research on artificial intelligence in mental health has also led to the innovation of virtual robots for therapy.`

Loss: pylate.losses.contrastive.Contrastive

Training Hyperparameters

Non-Default Hyperparameters

eval_strategy: steps
per_device_train_batch_size: 16
per_device_eval_batch_size: 32
gradient_accumulation_steps: 2
learning_rate: 2e-05
weight_decay: 0.01
num_train_epochs: 10
warmup_steps: 100
fp16: True
remove_unused_columns: False

All Hyperparameters

Click to expand

overwrite_output_dir: False
do_predict: False
eval_strategy: steps
prediction_loss_only: True
per_device_train_batch_size: 16
per_device_eval_batch_size: 32
per_gpu_train_batch_size: None
per_gpu_eval_batch_size: None
gradient_accumulation_steps: 2
eval_accumulation_steps: None
torch_empty_cache_steps: None
learning_rate: 2e-05
weight_decay: 0.01
adam_beta1: 0.9
adam_beta2: 0.999
adam_epsilon: 1e-08
max_grad_norm: 1.0
num_train_epochs: 10
max_steps: -1
lr_scheduler_type: linear
lr_scheduler_kwargs: {}
warmup_ratio: 0.0
warmup_steps: 100
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: False
fp16: True
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: False
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: None
hub_always_push: False
gradient_checkpointing: False
gradient_checkpointing_kwargs: None
include_inputs_for_metrics: False
include_for_metrics: []
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
average_tokens_across_devices: False
prompts: None
batch_sampler: batch_sampler
multi_dataset_batch_sampler: proportional

Training Logs

Click to expand

Epoch	Step	Training Loss	Validation Loss	accuracy
0.0051	50	4.8488	-	-
0.0103	100	2.2402	-	-
0.0154	150	1.8204	-	-
0.0206	200	1.7765	-	-
0.0257	250	1.7482	-	-
0	0	-	-	0.9227
0.0257	250	-	1.1625	-
0.0309	300	1.7821	-	-
0.0360	350	1.6761	-	-
0.0412	400	1.4887	-	-
0.0463	450	1.6001	-	-
0.0515	500	1.7426	-	-
0	0	-	-	0.9317
0.0515	500	-	1.1088	-
0.0566	550	1.5562	-	-
0.0617	600	1.6811	-	-
0.0669	650	1.5994	-	-
0.0720	700	1.5981	-	-
0.0772	750	1.5713	-	-
0	0	-	-	0.9369
0.0772	750	-	1.0817	-
0.0823	800	1.6516	-	-
0.0875	850	1.5768	-	-
0.0926	900	1.5902	-	-
0.0978	950	1.4613	-	-
0.1029	1000	1.6295	-	-
0	0	-	-	0.9374
0.1029	1000	-	1.0677	-
0.1081	1050	1.5301	-	-
0.1132	1100	1.6072	-	-
0.1183	1150	1.4644	-	-
0.1235	1200	1.6331	-	-
0.1286	1250	1.5464	-	-
0	0	-	-	0.9408
0.1286	1250	-	1.0547	-
0.1338	1300	1.5406	-	-
0.1389	1350	1.5471	-	-
0.1441	1400	1.6685	-	-
0.1492	1450	1.5644	-	-
0.1544	1500	1.6587	-	-
0	0	-	-	0.9420
0.1544	1500	-	1.0590	-
0.1595	1550	1.5793	-	-
0.1647	1600	1.4877	-	-
0.1698	1650	1.5781	-	-
0.1750	1700	1.5081	-	-
0.1801	1750	1.5434	-	-
0	0	-	-	0.9396
0.1801	1750	-	1.0564	-
0.1852	1800	1.4617	-	-
0.1904	1850	1.4531	-	-
0.1955	1900	1.5713	-	-
0.2007	1950	1.5166	-	-
0.2058	2000	1.4771	-	-
0	0	-	-	0.9431
0.2058	2000	-	1.0344	-
0.2110	2050	1.4706	-	-
0.2161	2100	1.5276	-	-
0.2213	2150	1.4002	-	-
0.2264	2200	1.5605	-	-
0.2316	2250	1.4871	-	-
0	0	-	-	0.9441
0.2316	2250	-	1.0355	-
0.2367	2300	1.56	-	-
0.2418	2350	1.4322	-	-
0.2470	2400	1.4682	-	-
0.2521	2450	1.4375	-	-
0.2573	2500	1.4499	-	-
0	0	-	-	0.9434
0.2573	2500	-	1.0306	-
0.2624	2550	1.5088	-	-
0.2676	2600	1.5577	-	-
0.2727	2650	1.4221	-	-
0.2779	2700	1.5105	-	-
0.2830	2750	1.4681	-	-
0	0	-	-	0.9453
0.2830	2750	-	1.0219	-
0.2882	2800	1.4354	-	-
0.2933	2850	1.4982	-	-
0.2984	2900	1.5374	-	-
0.3036	2950	1.4769	-	-
0.3087	3000	1.5767	-	-
0	0	-	-	0.9450
0.3087	3000	-	1.0168	-
0.3139	3050	1.3712	-	-
0.3190	3100	1.4979	-	-
0.3242	3150	1.4633	-	-
0.3293	3200	1.5025	-	-
0.3345	3250	1.5206	-	-
0	0	-	-	0.9457
0.3345	3250	-	1.0161	-
0.3396	3300	1.5119	-	-
0.3448	3350	1.6285	-	-
0.3499	3400	1.4421	-	-
0.3550	3450	1.4866	-	-
0.3602	3500	1.4651	-	-
0	0	-	-	0.9465
0.3602	3500	-	1.0085	-
0.3653	3550	1.3777	-	-
0.3705	3600	1.5256	-	-
0.3756	3650	1.358	-	-
0.3808	3700	1.4384	-	-
0.3859	3750	1.4847	-	-
0	0	-	-	0.9461
0.3859	3750	-	1.0093	-
0.3911	3800	1.327	-	-
0.3962	3850	1.4463	-	-
0.4014	3900	1.3179	-	-
0.4065	3950	1.4312	-	-
0.4116	4000	1.4179	-	-
0	0	-	-	0.9460
0.4116	4000	-	1.0145	-
0.4168	4050	1.4828	-	-
0.4219	4100	1.4568	-	-
0.4271	4150	1.4921	-	-
0.4322	4200	1.4485	-	-
0.4374	4250	1.4908	-	-
0	0	-	-	0.9478
0.4374	4250	-	1.0121	-
0.4425	4300	1.295	-	-
0.4477	4350	1.4687	-	-
0.4528	4400	1.3846	-	-
0.4580	4450	1.4704	-	-
0.4631	4500	1.3646	-	-
0	0	-	-	0.9480
0.4631	4500	-	1.0056	-
0.4683	4550	1.4779	-	-
0.4734	4600	1.4581	-	-
0.4785	4650	1.3786	-	-
0.4837	4700	1.56	-	-
0.4888	4750	1.4334	-	-
0	0	-	-	0.9475
0.4888	4750	-	1.0032	-
0.4940	4800	1.3877	-	-
0.4991	4850	1.3485	-	-
0.5043	4900	1.4509	-	-
0.5094	4950	1.3693	-	-
0.5146	5000	1.5226	-	-
0	0	-	-	0.9477
0.5146	5000	-	0.9976	-
0.5197	5050	1.4423	-	-
0.5249	5100	1.4191	-	-
0.5300	5150	1.5109	-	-
0.5351	5200	1.4509	-	-
0.5403	5250	1.4351	-	-
0	0	-	-	0.9486
0.5403	5250	-	1.0001	-
0.5454	5300	1.3868	-	-
0.5506	5350	1.4339	-	-
0.5557	5400	1.365	-	-
0.5609	5450	1.44	-	-
0.5660	5500	1.2895	-	-
0	0	-	-	0.9491
0.5660	5500	-	1.0065	-
0.5712	5550	1.4253	-	-
0.5763	5600	1.4438	-	-
0.5815	5650	1.3543	-	-
0.5866	5700	1.5587	-	-
0.5917	5750	1.342	-	-
0	0	-	-	0.9488
0.5917	5750	-	0.9927	-
0.5969	5800	1.4503	-	-
0.6020	5850	1.4045	-	-
0.6072	5900	1.4092	-	-
0.6123	5950	1.3318	-	-
0.6175	6000	1.416	-	-
0	0	-	-	0.9504
0.6175	6000	-	0.9910	-
0.6226	6050	1.5132	-	-
0.6278	6100	1.3275	-	-
0.6329	6150	1.4595	-	-
0.6381	6200	1.5112	-	-
0.6432	6250	1.4435	-	-
0	0	-	-	0.9515
0.6432	6250	-	0.9928	-
0.6483	6300	1.4268	-	-
0.6535	6350	1.5071	-	-
0.6586	6400	1.3817	-	-
0.6638	6450	1.5101	-	-
0.6689	6500	1.4014	-	-
0	0	-	-	0.9490
0.6689	6500	-	0.9954	-
0.6741	6550	1.2797	-	-
0.6792	6600	1.3829	-	-
0.6844	6650	1.4907	-	-
0.6895	6700	1.4098	-	-
0.6947	6750	1.482	-	-
0	0	-	-	0.9492
0.6947	6750	-	0.9937	-
0.6998	6800	1.3779	-	-
0.7050	6850	1.3791	-	-
0.7101	6900	1.5183	-	-
0.7152	6950	1.4022	-	-
0.7204	7000	1.544	-	-
0	0	-	-	0.9508
0.7204	7000	-	0.9935	-
0.7255	7050	1.4566	-	-
0.7307	7100	1.4641	-	-
0.7358	7150	1.4208	-	-
0.7410	7200	1.3391	-	-
0.7461	7250	1.5002	-	-
0	0	-	-	0.9497
0.7461	7250	-	0.9861	-
0.7513	7300	1.2985	-	-
0.7564	7350	1.5496	-	-
0.7616	7400	1.5046	-	-
0.7667	7450	1.3687	-	-
0.7718	7500	1.3841	-	-
0	0	-	-	0.9501
0.7718	7500	-	0.9868	-
0.7770	7550	1.3996	-	-
0.7821	7600	1.5112	-	-
0.7873	7650	1.4335	-	-
0.7924	7700	1.3867	-	-
0.7976	7750	1.3865	-	-
0	0	-	-	0.9511
0.7976	7750	-	0.9863	-
0.8027	7800	1.4039	-	-
0.8079	7850	1.379	-	-
0.8130	7900	1.3459	-	-
0.8182	7950	1.3996	-	-
0.8233	8000	1.4151	-	-
0	0	-	-	0.9511
0.8233	8000	-	0.9822	-
0.8284	8050	1.3745	-	-
0.8336	8100	1.4404	-	-
0.8387	8150	1.4776	-	-
0.8439	8200	1.398	-	-
0.8490	8250	1.4482	-	-
0	0	-	-	0.9506
0.8490	8250	-	0.9803	-
0.8542	8300	1.4551	-	-
0.8593	8350	1.46	-	-
0.8645	8400	1.5179	-	-
0.8696	8450	1.4067	-	-
0.8748	8500	1.4393	-	-
0	0	-	-	0.9504
0.8748	8500	-	0.9809	-
0.8799	8550	1.4995	-	-
0.8850	8600	1.4077	-	-
0.8902	8650	1.4088	-	-
0.8953	8700	1.3464	-	-
0.9005	8750	1.3455	-	-
0	0	-	-	0.9506
0.9005	8750	-	0.9797	-
0.9056	8800	1.5172	-	-
0.9108	8850	1.3922	-	-
0.9159	8900	1.3645	-	-
0.9211	8950	1.3627	-	-
0.9262	9000	1.3896	-	-
0	0	-	-	0.9506
0.9262	9000	-	0.9806	-
0.9314	9050	1.433	-	-
0.9365	9100	1.4678	-	-
0.9416	9150	1.3206	-	-
0.9468	9200	1.4589	-	-
0.9519	9250	1.3494	-	-
0	0	-	-	0.9509
0.9519	9250	-	0.9761	-
0.9571	9300	1.3768	-	-
0.9622	9350	1.4449	-	-
0.9674	9400	1.4187	-	-
0.9725	9450	1.3046	-	-
0.9777	9500	1.3586	-	-
0	0	-	-	0.9512
0.9777	9500	-	0.9817	-
0.9828	9550	1.4631	-	-
0.9880	9600	1.3113	-	-
0.9931	9650	1.2972	-	-
0.9983	9700	1.3793	-	-
1.0034	9750	1.1729	-	-
0	0	-	-	0.9509
1.0034	9750	-	0.9847	-
1.0085	9800	1.2009	-	-
1.0137	9850	1.2576	-	-
1.0188	9900	1.3483	-	-
1.0240	9950	1.2609	-	-
1.0291	10000	1.3099	-	-
0	0	-	-	0.9513
1.0291	10000	-	0.9895	-
1.0343	10050	1.2224	-	-
1.0394	10100	1.3552	-	-
1.0446	10150	1.3508	-	-
1.0497	10200	1.3242	-	-
1.0549	10250	1.2287	-	-
0	0	-	-	0.9512
1.0549	10250	-	0.9977	-
1.0600	10300	1.2863	-	-
1.0651	10350	1.2377	-	-
1.0703	10400	1.3058	-	-
1.0754	10450	1.3013	-	-
1.0806	10500	1.3233	-	-
0	0	-	-	0.9488
1.0806	10500	-	0.9948	-
1.0857	10550	1.334	-	-
1.0909	10600	1.246	-	-
1.0960	10650	1.2298	-	-
1.1012	10700	1.2016	-	-
1.1063	10750	1.3035	-	-
0	0	-	-	0.9506
1.1063	10750	-	0.9947	-
1.1115	10800	1.2457	-	-
1.1166	10850	1.2882	-	-
1.1217	10900	1.2365	-	-
1.1269	10950	1.19	-	-
1.1320	11000	1.2377	-	-
0	0	-	-	0.9511
1.1320	11000	-	0.9915	-
1.1372	11050	1.3028	-	-
1.1423	11100	1.319	-	-
1.1475	11150	1.3315	-	-
1.1526	11200	1.2161	-	-
1.1578	11250	1.3555	-	-
0	0	-	-	0.9511
1.1578	11250	-	0.9902	-
1.1629	11300	1.1874	-	-
1.1681	11350	1.2373	-	-
1.1732	11400	1.2474	-	-
1.1783	11450	1.2838	-	-
1.1835	11500	1.2242	-	-
0	0	-	-	0.9518
1.1835	11500	-	0.9927	-
1.1886	11550	1.3123	-	-
1.1938	11600	1.2874	-	-
1.1989	11650	1.2568	-	-
1.2041	11700	1.2526	-	-
1.2092	11750	1.347	-	-
0	0	-	-	0.9509
1.2092	11750	-	0.9883	-
1.2144	11800	1.3098	-	-
1.2195	11850	1.2541	-	-
1.2247	11900	1.2791	-	-
1.2298	11950	1.2333	-	-
1.2349	12000	1.3827	-	-
0	0	-	-	0.9507
1.2349	12000	-	0.9943	-
1.2401	12050	1.2732	-	-
1.2452	12100	1.2993	-	-
1.2504	12150	1.2947	-	-
1.2555	12200	1.3001	-	-
1.2607	12250	1.2957	-	-
0	0	-	-	0.9514
1.2607	12250	-	0.9865	-
1.2658	12300	1.1393	-	-
1.2710	12350	1.2996	-	-
1.2761	12400	1.3218	-	-
1.2813	12450	1.2138	-	-
1.2864	12500	1.1731	-	-
0	0	-	-	0.9510
1.2864	12500	-	0.9964	-
1.2916	12550	1.3326	-	-
1.2967	12600	1.3575	-	-
1.3018	12650	1.2948	-	-
1.3070	12700	1.2921	-	-
1.3121	12750	1.3052	-	-
0	0	-	-	0.9509
1.3121	12750	-	0.9840	-
1.3173	12800	1.3662	-	-
1.3224	12850	1.3673	-	-
1.3276	12900	1.3006	-	-
1.3327	12950	1.4217	-	-
1.3379	13000	1.1608	-	-
0	0	-	-	0.9520
1.3379	13000	-	0.9848	-
1.3430	13050	1.2066	-	-
1.3482	13100	1.408	-	-
1.3533	13150	1.3574	-	-
1.3584	13200	1.3171	-	-
1.3636	13250	1.3188	-	-
0	0	-	-	0.9502
1.3636	13250	-	0.9888	-
1.3687	13300	1.299	-	-
1.3739	13350	1.3015	-	-
1.3790	13400	1.3159	-	-
1.3842	13450	1.2139	-	-
1.3893	13500	1.2855	-	-
0	0	-	-	0.9514
1.3893	13500	-	0.9957	-
1.3945	13550	1.2705	-	-
1.3996	13600	1.3099	-	-
1.4048	13650	1.3144	-	-
1.4099	13700	1.2948	-	-
1.4150	13750	1.3313	-	-
0	0	-	-	0.9512
1.4150	13750	-	0.9910	-
1.4202	13800	1.3473	-	-
1.4253	13850	1.2037	-	-
1.4305	13900	1.3059	-	-
1.4356	13950	1.3763	-	-
1.4408	14000	1.2606	-	-
0	0	-	-	0.9523
1.4408	14000	-	0.9876	-
1.4459	14050	1.2394	-	-
1.4511	14100	1.219	-	-
1.4562	14150	1.3501	-	-
1.4614	14200	1.2664	-	-
1.4665	14250	1.2704	-	-
0	0	-	-	0.9513
1.4665	14250	-	0.9945	-
1.4716	14300	1.2332	-	-
1.4768	14350	1.2286	-	-
1.4819	14400	1.2123	-	-
1.4871	14450	1.2437	-	-
1.4922	14500	1.2292	-	-
0	0	-	-	0.9502
1.4922	14500	-	0.9886	-
1.4974	14550	1.3007	-	-
1.5025	14600	1.308	-	-
1.5077	14650	1.174	-	-
1.5128	14700	1.2648	-	-
1.5180	14750	1.2533	-	-
0	0	-	-	0.9517
1.5180	14750	-	0.9885	-
1.5231	14800	1.2576	-	-
1.5282	14850	1.3659	-	-
1.5334	14900	1.298	-	-
1.5385	14950	1.2723	-	-
1.5437	15000	1.3099	-	-
0	0	-	-	0.9518
1.5437	15000	-	0.9875	-
1.5488	15050	1.2984	-	-
1.5540	15100	1.2128	-	-
1.5591	15150	1.2689	-	-
1.5643	15200	1.2516	-	-
1.5694	15250	1.3028	-	-
0	0	-	-	0.9523
1.5694	15250	-	0.9856	-
1.5746	15300	1.3619	-	-
1.5797	15350	1.3524	-	-
1.5849	15400	1.1749	-	-
1.5900	15450	1.205	-	-
1.5951	15500	1.297	-	-
0	0	-	-	0.9513
1.5951	15500	-	0.9780	-
1.6003	15550	1.2469	-	-
1.6054	15600	1.2285	-	-
1.6106	15650	1.2963	-	-
1.6157	15700	1.2406	-	-
1.6209	15750	1.3049	-	-
0	0	-	-	0.9512
1.6209	15750	-	0.9873	-
1.6260	15800	1.2174	-	-
1.6312	15850	1.2789	-	-
1.6363	15900	1.289	-	-
1.6415	15950	1.3242	-	-
1.6466	16000	1.2974	-	-
0	0	-	-	0.9522
1.6466	16000	-	0.9755	-
1.6517	16050	1.2741	-	-
1.6569	16100	1.1625	-	-
1.6620	16150	1.2795	-	-
1.6672	16200	1.2301	-	-
1.6723	16250	1.2348	-	-
0	0	-	-	0.9528
1.6723	16250	-	0.9801	-
1.6775	16300	1.2408	-	-
1.6826	16350	1.2477	-	-
1.6878	16400	1.3386	-	-
1.6929	16450	1.2346	-	-
1.6981	16500	1.2904	-	-
0	0	-	-	0.9520
1.6981	16500	-	0.9906	-
1.7032	16550	1.2947	-	-
1.7083	16600	1.2572	-	-
1.7135	16650	1.2738	-	-
1.7186	16700	1.2686	-	-
1.7238	16750	1.4041	-	-
0	0	-	-	0.9528
1.7238	16750	-	0.9791	-
1.7289	16800	1.2935	-	-
1.7341	16850	1.2501	-	-
1.7392	16900	1.3208	-	-
1.7444	16950	1.2486	-	-
1.7495	17000	1.2587	-	-
0	0	-	-	0.9520
1.7495	17000	-	0.9862	-
1.7547	17050	1.3325	-	-
1.7598	17100	1.3104	-	-
1.7649	17150	1.2504	-	-
1.7701	17200	1.3153	-	-
1.7752	17250	1.328	-	-
0	0	-	-	0.9530
1.7752	17250	-	0.9803	-
1.7804	17300	1.3417	-	-
1.7855	17350	1.2486	-	-
1.7907	17400	1.2869	-	-
1.7958	17450	1.3599	-	-
1.8010	17500	1.2822	-	-
0	0	-	-	0.9526
1.8010	17500	-	0.9847	-
1.8061	17550	1.3001	-	-
1.8113	17600	1.0848	-	-
1.8164	17650	1.3171	-	-
1.8215	17700	1.3387	-	-
1.8267	17750	1.2401	-	-
0	0	-	-	0.9528
1.8267	17750	-	0.9804	-
1.8318	17800	1.2979	-	-
1.8370	17850	1.2222	-	-
1.8421	17900	1.27	-	-
1.8473	17950	1.3109	-	-
1.8524	18000	1.2306	-	-
0	0	-	-	0.9537
1.8524	18000	-	0.9876	-
1.8576	18050	1.1878	-	-
1.8627	18100	1.2398	-	-
1.8679	18150	1.2576	-	-
1.8730	18200	1.1579	-	-
1.8782	18250	1.2889	-	-
0	0	-	-	0.9519
1.8782	18250	-	0.9859	-
1.8833	18300	1.3331	-	-
1.8884	18350	1.2957	-	-
1.8936	18400	1.2286	-	-
1.8987	18450	1.2513	-	-
1.9039	18500	1.1702	-	-
0	0	-	-	0.9541
1.9039	18500	-	0.9840	-
1.9090	18550	1.3181	-	-
1.9142	18600	1.1976	-	-
1.9193	18650	1.3623	-	-
1.9245	18700	1.2594	-	-
1.9296	18750	1.2902	-	-
0	0	-	-	0.9522
1.9296	18750	-	0.9844	-
1.9348	18800	1.3283	-	-
1.9399	18850	1.2987	-	-
1.9450	18900	1.1987	-	-
1.9502	18950	1.2385	-	-
1.9553	19000	1.2772	-	-
0	0	-	-	0.9533
1.9553	19000	-	0.9861	-
1.9605	19050	1.1906	-	-
1.9656	19100	1.3041	-	-
1.9708	19150	1.2345	-	-
1.9759	19200	1.2586	-	-
1.9811	19250	1.196	-	-
0	0	-	-	0.9522
1.9811	19250	-	0.9835	-
1.9862	19300	1.2872	-	-
1.9914	19350	1.2449	-	-
1.9965	19400	1.2435	-	-
2.0016	19450	1.3096	-	-
2.0068	19500	1.1697	-	-
0	0	-	-	0.9514
2.0068	19500	-	1.0036	-
2.0119	19550	1.0556	-	-
2.0171	19600	1.1592	-	-
2.0222	19650	1.1808	-	-
2.0274	19700	1.141	-	-
2.0325	19750	1.1139	-	-
0	0	-	-	0.9517
2.0325	19750	-	1.0205	-
2.0377	19800	1.1959	-	-
2.0428	19850	1.0762	-	-
2.0480	19900	1.3522	-	-
2.0531	19950	1.1175	-	-
2.0582	20000	1.178	-	-
0	0	-	-	0.9512
2.0582	20000	-	1.0184	-
2.0634	20050	1.1416	-	-
2.0685	20100	1.1523	-	-
2.0737	20150	1.2561	-	-
2.0788	20200	1.119	-	-
2.0840	20250	1.095	-	-
0	0	-	-	0.9504
2.0840	20250	-	1.0155	-
2.0891	20300	1.1432	-	-
2.0943	20350	1.1455	-	-
2.0994	20400	1.0913	-	-
2.1046	20450	1.1671	-	-
2.1097	20500	1.2776	-	-
0	0	-	-	0.9514
2.1097	20500	-	1.0334	-
2.1149	20550	1.3092	-	-
2.1200	20600	1.1981	-	-
2.1251	20650	1.1399	-	-
2.1303	20700	1.0976	-	-
2.1354	20750	1.1335	-	-
0	0	-	-	0.9518
2.1354	20750	-	1.0136	-
2.1406	20800	1.1567	-	-
2.1457	20850	1.2536	-	-
2.1509	20900	1.1717	-	-
2.1560	20950	1.1433	-	-
2.1612	21000	1.1885	-	-
0	0	-	-	0.9512
2.1612	21000	-	1.0185	-
2.1663	21050	1.0543	-	-
2.1715	21100	1.1122	-	-
2.1766	21150	1.17	-	-
2.1817	21200	1.0757	-	-
2.1869	21250	1.3008	-	-
0	0	-	-	0.9506
2.1869	21250	-	1.0161	-
2.1920	21300	1.1723	-	-
2.1972	21350	1.2517	-	-
2.2023	21400	1.1834	-	-
2.2075	21450	1.1284	-	-
2.2126	21500	1.28	-	-
0	0	-	-	0.9507
2.2126	21500	-	1.0217	-
2.2178	21550	1.2478	-	-
2.2229	21600	1.1798	-	-
2.2281	21650	1.1218	-	-
2.2332	21700	1.2787	-	-
2.2383	21750	1.1254	-	-
0	0	-	-	0.9508
2.2383	21750	-	1.0312	-
2.2435	21800	1.2375	-	-
2.2486	21850	1.1074	-	-
2.2538	21900	1.0927	-	-
2.2589	21950	1.1691	-	-
2.2641	22000	1.1703	-	-
0	0	-	-	0.9499
2.2641	22000	-	1.0275	-
2.2692	22050	1.2158	-	-
2.2744	22100	1.1026	-	-
2.2795	22150	1.0644	-	-
2.2847	22200	1.1092	-	-
2.2898	22250	1.1686	-	-
0	0	-	-	0.9512
2.2898	22250	-	1.0343	-
2.2949	22300	1.2711	-	-
2.3001	22350	1.2942	-	-
2.3052	22400	1.2073	-	-
2.3104	22450	1.2131	-	-
2.3155	22500	1.1445	-	-
0	0	-	-	0.9517
2.3155	22500	-	1.0128	-
2.3207	22550	1.1553	-	-
2.3258	22600	1.1512	-	-
2.3310	22650	1.2069	-	-
2.3361	22700	1.1345	-	-
2.3413	22750	1.1681	-	-
0	0	-	-	0.9509
2.3413	22750	-	1.0101	-
2.3464	22800	1.1372	-	-
2.3515	22850	1.1393	-	-
2.3567	22900	1.1327	-	-
2.3618	22950	1.0903	-	-
2.3670	23000	1.1354	-	-
0	0	-	-	0.9513
2.3670	23000	-	1.0173	-
2.3721	23050	1.2517	-	-
2.3773	23100	1.0634	-	-
2.3824	23150	1.2095	-	-
2.3876	23200	1.1686	-	-
2.3927	23250	1.1063	-	-
0	0	-	-	0.9517
2.3927	23250	-	1.0243	-
2.3979	23300	1.1309	-	-
2.4030	23350	1.1869	-	-
2.4082	23400	1.1743	-	-
2.4133	23450	1.1001	-	-
2.4184	23500	1.1696	-	-
0	0	-	-	0.9525
2.4184	23500	-	1.0315	-
2.4236	23550	1.1493	-	-
2.4287	23600	1.1486	-	-
2.4339	23650	1.2302	-	-
2.4390	23700	1.1427	-	-
2.4442	23750	1.2123	-	-
0	0	-	-	0.9510
2.4442	23750	-	1.0297	-
2.4493	23800	1.1169	-	-
2.4545	23850	1.1688	-	-
2.4596	23900	1.0506	-	-
2.4648	23950	1.1965	-	-
2.4699	24000	1.1253	-	-
0	0	-	-	0.9508
2.4699	24000	-	1.0238	-
2.4750	24050	1.1957	-	-
2.4802	24100	1.1395	-	-
2.4853	24150	1.1238	-	-
2.4905	24200	1.1342	-	-
2.4956	24250	1.1703	-	-
0	0	-	-	0.9506
2.4956	24250	-	1.0219	-
2.5008	24300	1.0947	-	-
2.5059	24350	1.1281	-	-
2.5111	24400	1.1029	-	-
2.5162	24450	1.1784	-	-
2.5214	24500	1.101	-	-
0	0	-	-	0.9528
2.5214	24500	-	1.0267	-
2.5265	24550	1.1231	-	-
2.5316	24600	1.1364	-	-
2.5368	24650	1.1778	-	-
2.5419	24700	1.1089	-	-
2.5471	24750	1.1626	-	-
0	0	-	-	0.9508
2.5471	24750	-	1.0254	-
2.5522	24800	1.2019	-	-
2.5574	24850	1.1503	-	-
2.5625	24900	1.1697	-	-
2.5677	24950	1.0921	-	-
2.5728	25000	1.3136	-	-
0	0	-	-	0.9513
2.5728	25000	-	1.0222	-

Framework Versions

Python: 3.12.4
Sentence Transformers: 4.0.2
PyLate: 1.2.0
Transformers: 4.48.2
PyTorch: 2.6.0+cu124
Accelerate: 1.7.0
Datasets: 3.6.0
Tokenizers: 0.21.1

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"
}

PyLate

bibtex

@misc{PyLate,
title={PyLate: Flexible Training and Retrieval for Late Interaction Models},
author={Chaffin, Antoine and Sourty, Raphaël},
url={https://github.com/lightonai/pylate},
year={2024}
}

	query	positive	negative
type	string	string	string
details	min: 38 tokens mean: 97.92 tokens max: 32 tokens	min: 84 tokens mean: 128 tokens max: 128 tokens	min: 84 tokens mean: 127.7 tokens max: 128 tokens

souvickdascmsa019/all-MiniLM-L6-v2-reason_ir

PyLate model based on sentence-transformers/all-MiniLM-L6-v2

Model Details

Model Description

Model Sources

Usage

Retrieval

Indexing documents

Retrieving top-k documents for queries

Reranking

Evaluation

Metrics

Col BERTTriplet

Training Details

Training Dataset

Unnamed Dataset

Evaluation Dataset

Unnamed Dataset

Training Hyperparameters

Non-Default Hyperparameters

All Hyperparameters

Training Logs

Framework Versions

Citation

BibTeX

Sentence Transformers

PyLate

souvickdascmsa019/
all-MiniLM-L6-v2-reason_ir