This is a sentence-transformers model finetuned from Snowflake/snowflake-arctic-embed-l. It maps sentences & paragraphs to a 1024-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more.

Model Details

Model Description

Model Type: Sentence Transformer
Base model: Snowflake/snowflake-arctic-embed-l
Maximum Sequence Length: 512 tokens
Output Dimensionality: 1024 dimensions
Similarity Function: Cosine Similarity

Model Sources

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

Full Model Architecture

SentenceTransformer(
  (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
  (1): Pooling({'word_embedding_dimension': 1024, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
  (2): Normalize()
)

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("martintgc/finetuned_arctic_naive_ft-legal-ft-v0")
# Run inference
sentences = [
    'Who did the captaincy of the New Zealand cricket team transition from, and how was this change perceived?',
    "to be captain, but didn’t get it. Then, he kind of just took the captaincy from Ross Taylor in a bloodless coup. Although some people might say it wasn't particularly bloodless.And then New Zealand went to South Africa, and they had just the worst time. It was really, really bad. So all the players got together and asked themselves hard questions. “What are we doing? Are we going to become freelance players in franchise cricket? Are we going to take New Zealand cricket seriously?”Arising from that, they went all-in on a method that I suppose we would now call Bazball. But it did take a little while to come together. There were lots of different aspects that needed to fall in place. McCullum had to eventually give up the gloves, and most",
    "were so good in this period is that they had two quality spinners, and they had a bunch of people who could play spin.By the time they got to white-ball sides also officially adopting Bazball, their ability to bowl spin and play spin had almost completely disappeared. They've tried to disrupt as much as possible, and they stole a Test in India doing it. But, we've also seen the limitations of that, and Pakistan embarrassed them recently. Now, they're back trying to play spin again, and we can see that they don't have it.The genius of Bazball and Brendon McCullum has always been to highlight your strengths and ignore your weaknesses. How do you do that when just playing spin or bowling spin is one of them? That's a hard thing to overcome",
]
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]

Evaluation

Metrics

Information Retrieval

Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.8571
cosine_accuracy@3	0.9643
cosine_accuracy@5	1.0
cosine_accuracy@10	1.0
cosine_precision@1	0.8571
cosine_precision@3	0.3214
cosine_precision@5	0.2
cosine_precision@10	0.1
cosine_recall@1	0.8571
cosine_recall@3	0.9643
cosine_recall@5	1.0
cosine_recall@10	1.0
cosine_ndcg@10	0.9354
cosine_mrr@10	0.9137
cosine_map@100	0.9137

Training Details

Training Dataset

Unnamed Dataset

Size: 56 training samples
Columns: sentence_0 and sentence_1
Approximate statistics based on the first 56 samples:
sentence_0 sentence_1
type string string
details
min: 12 tokens
mean: 21.68 tokens
max: 32 tokens
min: 16 tokens
mean: 161.11 tokens
max: 213 tokens

Samples:

sentence_0	sentence_1
`What is the main focus of Jarrod Kimber's "Bazball's mission to Moscow"?`	`Bazball's mission to Moscow - by Jarrod Kimber - Good Areas`
`How does the concept of "Good Areas" relate to Bazball in the context of the article?`	`Bazball's mission to Moscow - by Jarrod Kimber - Good Areas`
`What is the significance of Brendon McCullum and England’s Test side in relation to Bazball?`	SubscribeSign inShare this postGood AreasBazball's mission to MoscowCopy linkFacebookEmailNotesMoreBazball's mission to MoscowIt created a revolution when Brendon McCullum and England’s Test side adopted it. But we’ve seen this before.Jarrod KimberFeb 13, 202513Share this postGood AreasBazball's mission to MoscowCopy linkFacebookEmailNotesMore1SharePre-order The Art of BattingSubscribeI've been thinking about Bazball a lot recently because Bazball’s back. Alright. We had Test Bazball, and now we have limited-overs Bazball. Which, I suppose, is Test Bazball on a slightly different frequency. But Bazball is much weirder when you think about it. I don’t know which version of Bazball we’re on, maybe it’s version three. Or maybe it’s version

Loss: MatryoshkaLoss with these parameters:
json
1{ 2 "loss": "MultipleNegativesRankingLoss", 3 "matryoshka_dims": [ 4 768, 5 512, 6 256, 7 128, 8 64 9 ], 10 "matryoshka_weights": [ 11 1, 12 1, 13 1, 14 1, 15 1 16 ], 17 "n_dims_per_step": -1 18}

Training Hyperparameters

Non-Default Hyperparameters

eval_strategy: steps
per_device_train_batch_size: 10
per_device_eval_batch_size: 10
num_train_epochs: 10
multi_dataset_batch_sampler: round_robin

All Hyperparameters

Click to expand

overwrite_output_dir: False
do_predict: False
eval_strategy: steps
prediction_loss_only: True
per_device_train_batch_size: 10
per_device_eval_batch_size: 10
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: 5e-05
weight_decay: 0.0
adam_beta1: 0.9
adam_beta2: 0.999
adam_epsilon: 1e-08
max_grad_norm: 1
num_train_epochs: 10
max_steps: -1
lr_scheduler_type: linear
lr_scheduler_kwargs: {}
warmup_ratio: 0.0
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: False
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: 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: round_robin

Training Logs

Epoch	Step	cosine_ndcg@10
1.0	6	0.9232
2.0	12	0.9354
3.0	18	0.9486
4.0	24	0.9643
5.0	30	0.9643
6.0	36	0.9643
7.0	42	0.9643
8.0	48	0.9643
8.3333	50	0.9643
9.0	54	0.9643
10.0	60	0.9643
1.0	6	0.9643
2.0	12	0.9643
3.0	18	0.9379
4.0	24	0.9354
5.0	30	0.9354
6.0	36	0.9354
7.0	42	0.9339
8.0	48	0.9339
8.3333	50	0.9339
9.0	54	0.9339
10.0	60	0.9339
1.0	6	0.9486
2.0	12	0.9486
3.0	18	0.9486
4.0	24	0.9486
5.0	30	0.9486
6.0	36	0.9486
7.0	42	0.9354
8.0	48	0.9354
8.3333	50	0.9354
9.0	54	0.9354
10.0	60	0.9354

Framework Versions

Python: 3.9.21
Sentence Transformers: 3.4.1
Transformers: 4.49.0
PyTorch: 2.6.0
Accelerate: 1.4.0
Datasets: 3.3.2
Tokenizers: 0.21.0

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}
}

	sentence_0	sentence_1
type	string	string
details	min: 12 tokens mean: 21.68 tokens max: 32 tokens	min: 16 tokens mean: 161.11 tokens max: 213 tokens