SentenceTransformer based on Snowflake/snowflake-arctic-embed-l

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.

from sentence_transformers import SentenceTransformer

# Download from the 🤗 Hub
model = SentenceTransformer("shivXy/ot-midterm-v0")
# Run inference
sentences = [
    '1. What was the focus of the pilot study mentioned regarding tendinosis of the Achilles tendon?',
    'tendinosis of the Achilles tendon: a pilot study. AJR Am J Roentgenol . \n2007;189:W215–W220 . \n74. van Leeuwen WF , Janssen SJ , Ring D , Chen N . Incidental magnetic resonance \nimaging signal changes in the extensor carpi radialis brevis origin are more \ncommon with age. J Shoulder Elbow Surg . 2016;25:1175–1181 . \n75. Rabago D , Lee KS , Ryan M , et al. Hypertonic dextrose and morrhuate sodium \ninjections (prolotherapy) for lateral epicondylosis (tennis elbow): results of a \nsingle-blind, pilot-level, randomized controlled trial. Am J Phys Med Rehabil . \n2013;92:587–596 . \n76. Scarpone M , Rabago DP , Zgierska A , Arbogast G , Snell E . The eﬃcacy \nof prolotherapy for lateral epicondylosis: a pilot study. Clin J Sport Med .',
    '179. Dick FD , Graveling RA , Munro W , Walker-Bone K . Workplace management of \nupper limb disorders: a systematic review. Occup Med (Lond) . 2011;61:19–25 . \n180. Buchanan H , Van Niekerk L , Grimmer K . Work transition after hand injury: a \nscoping review. J Hand Ther . 2020 . \n181. Rost KA , Alvero AM . Participatory approaches to workplace safety man- \nagement: bridging the gap between behavioral safety and participatory er- \ngonomics. Int J Occup Saf Ergon . 2020;26:194–203 . \n182. Bernardes JM , Ruiz-Frutos C , Moro ARP , Dias A . A low-cost and eﬃcient par- \nticipatory ergonomic intervention to reduce the burden of work-related mus- \nculoskeletal disorders in an industrially developing country: an experience re-',
]
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	0.9545
cosine_accuracy@3	1.0
cosine_accuracy@5	1.0
cosine_accuracy@10	1.0
cosine_precision@1	0.9545
cosine_precision@3	0.3333
cosine_precision@5	0.2
cosine_precision@10	0.1
cosine_recall@1	0.9545
cosine_recall@3	1.0
cosine_recall@5	1.0
cosine_recall@10	1.0
cosine_ndcg@10	0.9832
cosine_mrr@10	0.9773
cosine_map@100	0.9773

Training Details

Training Dataset

Unnamed Dataset

Size: 812 training samples
Columns: sentence_0 and sentence_1
Approximate statistics based on the first 812 samples:
sentence_0 sentence_1
type string string
details
min: 12 tokens
mean: 24.85 tokens
max: 60 tokens

min: 8 tokens
mean: 158.41 tokens
max: 320 tokens

	sentence_0	sentence_1
type	string	string
details	min: 12 tokens mean: 24.85 tokens max: 60 tokens	min: 8 tokens mean: 158.41 tokens max: 320 tokens

Samples:

sentence_0	sentence_1
`2. What type of intervention is being compared to strength training in the study protocol by Sundstrup E and colleagues?`	ment for work-related lateral epicondylitis. Work . 2010;37:81–86 . 161. Parimalam P , Premalatha MR , Padmini DS , Ganguli AK . Participatory er- gonomics in redesigning a dyeing tub for fabric dyers. Work . 2012;43:453–458 . 162. Harari D , Casarotto RA . Effectiveness of a multifaceted intervention to manage musculoskeletal disorders in workers of a medium-sized company. Int J Occup Saf Ergon . 2021;27:247–257 . 163. Sundstrup E , Jakobsen MD , Andersen CH , et al. Participatory ergonomic inter- vention versus strength training on chronic pain and work disability in slaugh- terhouse workers: study protocol for a single-blind, randomized controlled trial. BMC Musculoskelet Disord . 2013;14:67 .
`2. What does the increased signal intensity in the proximal portion of the lateral collateral ligament suggest about the patient's condition?`	266 C.W. Stegink-Jansen, J.G. Bynum, A.L. Lambropoulos et al. / Journal of Hand Therapy 34 (2021) 263–297 Fig. 3. Pathology. A 60-year-old female with right elbow pain for 5 weeks. (A) Coronal fat-suppressed FSE T2-weighted image showing mild thickening of the proximal portion of the common extensor tendon with increased signal intensity (arrow), suggesting mild injury. Irregular thickening with increased signal intensity in the proximal portion of the lateral collateral ligament (arrowhead) is also noted, suggesting mild injury. (B) and (C) Coronal PD FSE image and oblique radiograph showing cortical
`2. What factors are assessed in relation to lower extremity (LE) issues according to the systematic review?`	Manufacturing: • Electronics • Auto parts • Windows • Cabinets • Medical equipment • Fitness equipment Healthcare (excluding direct patient care): • Hospitals • Health research Worker: structured interviews, physical examinations Environment: workplace walk through Hazards in work: individual assessments of biomechanical and psychosocial factors LE related to frequency of forceful exertions or forearm supination and forceful lifting; increased odds of LE related to being age 36-50, female, or a smoker; high social support appeared protective against LE van Rijn et al. 2009 Assess relationship between work-related physical factors, psychosocial factors, and LE Systematic review 13 studies: • 9 cross-sectional

Loss: MatryoshkaLoss with these parameters:

{
    "loss": "MultipleNegativesRankingLoss",
    "matryoshka_dims": [
        768,
        512,
        256,
        128,
        64
    ],
    "matryoshka_weights": [
        1,
        1,
        1,
        1,
        1
    ],
    "n_dims_per_step": -1
}

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	Training Loss	cosine_ndcg@10
0.6098	50	-	1.0
1.0	82	-	0.9773
1.2195	100	-	0.9664
1.8293	150	-	1.0
2.0	164	-	0.9832
2.4390	200	-	0.9832
3.0	246	-	0.9832
3.0488	250	-	0.9832
3.6585	300	-	0.9832
4.0	328	-	0.9832
4.2683	350	-	0.9832
4.8780	400	-	0.9832
5.0	410	-	0.9832
5.4878	450	-	0.9832
6.0	492	-	0.9832
6.0976	500	0.6578	0.9832
6.7073	550	-	0.9664
7.0	574	-	0.9664
7.3171	600	-	0.9664
7.9268	650	-	0.9664
8.0	656	-	0.9664
8.5366	700	-	0.9832
9.0	738	-	0.9832
9.1463	750	-	0.9832
9.7561	800	-	0.9832
10.0	820	-	0.9832

Framework Versions

Python: 3.13.2
Sentence Transformers: 3.4.1
Transformers: 4.49.0
PyTorch: 2.6.0+cpu
Accelerate: 1.4.0
Datasets: 3.3.2
Tokenizers: 0.21.0

Citation

BibTeX

Sentence Transformers

@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

@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

@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}
}

shivXy
/

ot-midterm-v0