A supervised baseline: all-MiniLM-L6-v2 fine-tuned with MultipleNegativesRankingLoss on ESCI relevance labels (not bag-derived), used as a comparison point for the bag-of-documents retrieval research.

SentenceTransformer based on sentence-transformers/all-MiniLM-L6-v2

This is a sentence-transformers model finetuned from sentence-transformers/all-MiniLM-L6-v2. It maps sentences & paragraphs to a 384-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: sentence-transformers/all-MiniLM-L6-v2
Maximum Sequence Length: 256 tokens
Output Dimensionality: 384 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': 256, 'do_lower_case': False, 'architecture': 'BertModel'})
  (1): Pooling({'word_embedding_dimension': 384, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, '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("sentence_transformers_model_id")
# Run inference
sentences = [
    'preservatives free eye drops',
    'Refresh Plus Lubricant Eye Drops, Preservative-Free, 0.01 Fl Oz Single-Use Containers, 30 Count (Pack of 1)',
    'Ocusoft Retaine MGD Ophthalmic Emulsion, Milky White Solution, 30 count Single Use Containers, 0.01 Fluid Ounce',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 384]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities)
# tensor([[1.0000, 0.7632, 0.4848],
#         [0.7632, 1.0000, 0.4949],
#         [0.4848, 0.4949, 1.0000]])

Training Details

Training Dataset

Unnamed Dataset

Size: 125,555 training samples
Columns: sentence_0, sentence_1, and sentence_2
Approximate statistics based on the first 1000 samples:
sentence_0 sentence_1 sentence_2
type string string string
details
min: 3 tokens
mean: 6.87 tokens
max: 36 tokens
min: 3 tokens
mean: 26.27 tokens
max: 91 tokens
min: 3 tokens
mean: 26.04 tokens
max: 96 tokens

Samples:

sentence_0	sentence_1	sentence_2
`e gift card`	`App Store & iTunes Gift Cards - Email Delivery`	`ASURION 4 Year Laptop Accident Protection Plan with Tech Support $600-699.99`
`mre's meals ready to eat military`	`Ozark Outdoorz, LLC 2020 Pack Date/2025 Best by Date - XMRE 1300XT (Meals Ready to Eat) - 4 Pack Random`	`5000mAh Weather Radio,Solar Hand Crank Emergency Radio,NOAA/AM/FM Shortwave Outdoor Survival Portable Radio, Power Bank USB Charger,Flashlight/Reading Lamp,Headphone Jack,SOS`
`nested magnesium glycinate chelate`	`EquiLife - Full Spectrum Magnesium, Magnesium Glycinate, Boosts Mood, Promotes Restful Sleep, Improves Bone Health, Enhances Energy & Focus, Helps Steady Blood Pressure, Non-GMO, Vegan (90 Servings)`	`Life Extension, Neuro-Mag, 90 Cápsulas vegetarianas`

Loss: MultipleNegativesRankingLoss with these parameters:
json
1{ 2 "scale": 20.0, 3 "similarity_fct": "cos_sim", 4 "gather_across_devices": false, 5 "directions": [ 6 "query_to_doc" 7 ], 8 "partition_mode": "joint", 9 "hardness_mode": null, 10 "hardness_strength": 0.0 11}

Training Hyperparameters

Non-Default Hyperparameters

per_device_train_batch_size: 32
num_train_epochs: 10
per_device_eval_batch_size: 32
multi_dataset_batch_sampler: round_robin

All Hyperparameters

Click to expand

per_device_train_batch_size: 32
num_train_epochs: 10
max_steps: -1
learning_rate: 5e-05
lr_scheduler_type: linear
lr_scheduler_kwargs: None
warmup_steps: 0
optim: adamw_torch_fused
optim_args: None
weight_decay: 0.0
adam_beta1: 0.9
adam_beta2: 0.999
adam_epsilon: 1e-08
optim_target_modules: None
gradient_accumulation_steps: 1
average_tokens_across_devices: True
max_grad_norm: 1
label_smoothing_factor: 0.0
bf16: False
fp16: False
bf16_full_eval: False
fp16_full_eval: False
tf32: None
gradient_checkpointing: False
gradient_checkpointing_kwargs: None
torch_compile: False
torch_compile_backend: None
torch_compile_mode: None
use_liger_kernel: False
liger_kernel_config: None
use_cache: False
neftune_noise_alpha: None
torch_empty_cache_steps: None
auto_find_batch_size: False
log_on_each_node: True
logging_nan_inf_filter: True
include_num_input_tokens_seen: no
log_level: passive
log_level_replica: warning
disable_tqdm: False
project: huggingface
trackio_space_id: trackio
eval_strategy: no
per_device_eval_batch_size: 32
prediction_loss_only: True
eval_on_start: False
eval_do_concat_batches: True
eval_use_gather_object: False
eval_accumulation_steps: None
include_for_metrics: []
batch_eval_metrics: False
save_only_model: False
save_on_each_node: False
enable_jit_checkpoint: False
push_to_hub: False
hub_private_repo: None
hub_model_id: None
hub_strategy: every_save
hub_always_push: False
hub_revision: None
load_best_model_at_end: False
ignore_data_skip: False
restore_callback_states_from_checkpoint: False
full_determinism: False
seed: 42
data_seed: None
use_cpu: False
accelerator_config: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
parallelism_config: None
dataloader_drop_last: False
dataloader_num_workers: 0
dataloader_pin_memory: True
dataloader_persistent_workers: False
dataloader_prefetch_factor: None
remove_unused_columns: True
label_names: None
train_sampling_strategy: random
length_column_name: length
ddp_find_unused_parameters: None
ddp_bucket_cap_mb: None
ddp_broadcast_buffers: False
ddp_backend: None
ddp_timeout: 1800
fsdp: []
fsdp_config: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}
deepspeed: None
debug: []
skip_memory_metrics: True
do_predict: False
resume_from_checkpoint: None
warmup_ratio: None
local_rank: -1
prompts: None
batch_sampler: batch_sampler
multi_dataset_batch_sampler: round_robin
router_mapping: {}
learning_rate_mapping: {}

Training Logs

Epoch	Step	Training Loss
0.1274	500	0.9736
0.2548	1000	0.8578
0.3823	1500	0.8133
0.5097	2000	0.7662
0.6371	2500	0.7642
0.7645	3000	0.7438
0.8919	3500	0.7035
1.0194	4000	0.6752
1.1468	4500	0.6039
1.2742	5000	0.5819
1.4016	5500	0.5710
1.5291	6000	0.5514
1.6565	6500	0.5315
1.7839	7000	0.5164
1.9113	7500	0.5121
2.0387	8000	0.4657
2.1662	8500	0.3968
2.2936	9000	0.3996
2.4210	9500	0.3826
2.5484	10000	0.3845
2.6758	10500	0.3783
2.8033	11000	0.3745
2.9307	11500	0.3708
3.0581	12000	0.3310
3.1855	12500	0.2975
3.3129	13000	0.2886
3.4404	13500	0.2926
3.5678	14000	0.2910
3.6952	14500	0.2800
3.8226	15000	0.2918
3.9501	15500	0.2828
4.0775	16000	0.2514
4.2049	16500	0.2344
4.3323	17000	0.2299
4.4597	17500	0.2315
4.5872	18000	0.2357
4.7146	18500	0.2350
4.8420	19000	0.2358
4.9694	19500	0.2350
5.0968	20000	0.2065
5.2243	20500	0.1891
5.3517	21000	0.1934
5.4791	21500	0.1922
5.6065	22000	0.1962
5.7339	22500	0.1927
5.8614	23000	0.1853
5.9888	23500	0.1886
6.1162	24000	0.1669
6.2436	24500	0.1593
6.3710	25000	0.1661
6.4985	25500	0.1674
6.6259	26000	0.1651
6.7533	26500	0.1707
6.8807	27000	0.1619
7.0082	27500	0.1639
7.1356	28000	0.1431
7.2630	28500	0.1525
7.3904	29000	0.1479
7.5178	29500	0.1450
7.6453	30000	0.1430
7.7727	30500	0.1424
7.9001	31000	0.1467
8.0275	31500	0.1416
8.1549	32000	0.1287
8.2824	32500	0.1345
8.4098	33000	0.1349
8.5372	33500	0.1372
8.6646	34000	0.1344
8.7920	34500	0.1294
8.9195	35000	0.1324
9.0469	35500	0.1312
9.1743	36000	0.1295
9.3017	36500	0.1259
9.4292	37000	0.1316
9.5566	37500	0.1228
9.6840	38000	0.1277
9.8114	38500	0.1279
9.9388	39000	0.1228

Framework Versions

Python: 3.14.4
Sentence Transformers: 5.3.0
Transformers: 5.4.0
PyTorch: 2.11.0
Accelerate: 1.13.0
Datasets: 4.8.4
Tokenizers: 0.22.2

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",
}

MultipleNegativesRankingLoss