SentenceTransformer based on BAAI/bge-m3

This is a sentence-transformers model finetuned from BAAI/bge-m3. 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: BAAI/bge-m3
Maximum Sequence Length: 8192 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': 8192, 'do_lower_case': False, 'architecture': 'XLMRobertaModel'})
  (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("sentence_transformers_model_id")
# Run inference
sentences = [
    '{"patient":"{"age": "58.0", "age_cod": "yr", "gndr_cod": "f", "wt": "101.15", "wt_cod": "kg"}","treatment":"{"drugname": "rinvoq", "route": "oral", "dose": "frequency : daily;?"}","indi_pt":"rheumatoid arthritis"}',
    '{"pt": "pain; headache; dizziness; swelling face; therapy interrupted", "uni_code": "NA"}',
    '{"pt": "alanine aminotransferase increased; aspartate aminotransferase increased; bacteria urine identified; blood pressure increased; blood sodium decreased; blood urea increased; blood urine; cerebral ischaemia; chills; coma; haematocrit decreased; haemoglobin decreased; haemorrhage intracranial; musculoskeletal discomfort; pallor; platelet count decreased; protein urine present; pyrexia; refusal of treatment by patient; sinus pain; urinary casts; urinary sediment present; white blood cell count decreased; white blood cell count increased; white blood cells urine positive", "uni_code": "DE"}',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 1024]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities)
# tensor([[ 1.0000,  0.5472, -0.0551],
#         [ 0.5472,  1.0000, -0.1079],
#         [-0.0551, -0.1079,  1.0000]])

Evaluation

Metrics

Information Retrieval

Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.0021
cosine_accuracy@3	0.0053
cosine_accuracy@5	0.008
cosine_accuracy@10	0.013
cosine_precision@1	0.0021
cosine_precision@3	0.0018
cosine_precision@5	0.0016
cosine_precision@10	0.0013
cosine_recall@1	0.0021
cosine_recall@3	0.0053
cosine_recall@5	0.008
cosine_recall@10	0.013
cosine_ndcg@10	0.0067
cosine_mrr@10	0.0048
cosine_map@100	0.0063

Training Details

Training Dataset

Unnamed Dataset

Size: 400,000 training samples
Columns: sentence_0 and sentence_1
Approximate statistics based on the first 1000 samples:
sentence_0 sentence_1
type string string
details
min: 105 tokens
mean: 299.0 tokens
max: 2505 tokens

min: 20 tokens
mean: 40.49 tokens
max: 282 tokens

	sentence_0	sentence_1
type	string	string
details	min: 105 tokens mean: 299.0 tokens max: 2505 tokens	min: 20 tokens mean: 40.49 tokens max: 282 tokens

Samples:

sentence_0	sentence_1
`{"patient":"{"age": "15.0", "age_cod": "yr", "gndr_cod": "f", "wt": "60.0", "wt_cod": "kg"}","treatment":"{"drugname": "rizatriptan.", "route": "oral", "dose": "qd, tablet"}; {"drugname": "amitriptyline", "route": "oral", "dose": "qd, tablet"}; {"drugname": "mirtazapine.", "route": "oral", "dose": "qd, tablet"}; {"drugname": "sumatriptan.", "route": "oral", "dose": "qd, tablet"}","indi_pt":"suicidal ideation; suicidal ideation; suicidal ideation"}`	`{"pt": "acute respiratory failure; intentional overdose; somnolence; sopor; toxicity to various agents", "uni_code": "HO"}`
`{"patient":"{"age": "38.0", "age_cod": "yr", "gndr_cod": "m", "wt": "86.17", "wt_cod": "kg"}","treatment":"{"drugname": "albuterol sulfate.", "route": "respiratory (inhalation)", "dose": "180 mcg, prn"}; {"drugname": "albuterol sulfate.", "route": "respiratory (inhalation)", "dose": "180 mcg, prn"}","indi_pt":"asthma"}`	`{"pt": "chest discomfort; device delivery system issue", "uni_code": "NA"}`
{"patient":"{"age": "62.0", "age_cod": "yr", "gndr_cod": "f", "wt": "90.7", "wt_cod": "kg"}","treatment":"{"drugname": "effexor xr", "route": "nan", "dose": "nan"}; {"drugname": "effexor xr", "route": "nan", "dose": "nan"}; {"drugname": "effexor xr", "route": "nan", "dose": "nan"}; {"drugname": "venlafaxine hcl", "route": "nan", "dose": "nan"}; {"drugname": "venlafaxine hcl", "route": "nan", "dose": "nan"}; {"drugname": "venlafaxine hcl", "route": "nan", "dose": "nan"}; {"drugname": "venlafaxine hydrochloride", "route": "nan", "dose": "nan"}; {"drugname": "venlafaxine hydrochloride", "route": "nan", "dose": "nan"}; {"drugname": "venlafaxine hydrochloride", "route": "nan", "dose": "nan"}; {"drugname": "venlafaxine hydrochloride", "route": "nan", "dose": "nan"}; {"drugname": "venlafaxine hydrochloride", "route": "nan", "dose": "nan"}; {"drugname": "venlafaxine hydrochloride", "route": "nan", "dose": "nan"}; {"drugname": "wellbutrin", "route": "nan", "dose": "100 mg, daily"}","indi_pt":"m...	`{"pt": "arthropathy; drug ineffective", "uni_code": "NA"}`

Loss: MultipleNegativesRankingLoss with these parameters:

{
    "scale": 20.0,
    "similarity_fct": "cos_sim"
}

Training Hyperparameters

Non-Default Hyperparameters

eval_strategy: steps
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: 8
per_device_eval_batch_size: 8
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: 3
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
hub_revision: None
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
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
liger_kernel_config: None
eval_use_gather_object: False
average_tokens_across_devices: False
prompts: None
batch_sampler: batch_sampler
multi_dataset_batch_sampler: round_robin
router_mapping: {}
learning_rate_mapping: {}

Training Logs

Click to expand

Epoch	Step	Training Loss	cosine_ndcg@10
0.01	500	2.064	-
0.02	1000	1.9201	-
0.03	1500	1.8076	-
0.04	2000	1.7084	-
0.05	2500	1.6477	-
0.06	3000	1.5749	-
0.07	3500	1.5356	-
0.08	4000	1.4819	-
0.09	4500	1.422	-
0.1	5000	1.3966	-
0.11	5500	1.3675	-
0.12	6000	1.3818	-
0.13	6500	1.3584	-
0.14	7000	1.3121	-
0.15	7500	1.312	-
0.16	8000	1.2551	-
0.17	8500	1.2927	-
0.18	9000	1.2369	-
0.19	9500	1.252	-
0.2	10000	1.2314	-
0.21	10500	1.2033	-
0.22	11000	1.2347	-
0.23	11500	1.1843	-
0.24	12000	1.1809	-
0.25	12500	1.1919	-
0.26	13000	1.1592	-
0.27	13500	1.1628	-
0.28	14000	1.1466	-
0.29	14500	1.1374	-
0.3	15000	1.1423	-
0.31	15500	1.1371	-
0.32	16000	1.1454	-
0.33	16500	1.1266	-
0.34	17000	1.1363	-
0.35	17500	1.162	-
0.36	18000	1.125	-
0.37	18500	1.0757	-
0.38	19000	1.0912	-
0.39	19500	1.0784	-
0.4	20000	1.0761	-
0.41	20500	1.1171	-
0.42	21000	1.0636	-
0.43	21500	1.0759	-
0.44	22000	1.0536	-
0.45	22500	1.0589	-
0.46	23000	1.0116	-
0.47	23500	1.0576	-
0.48	24000	1.0583	-
0.49	24500	1.0401	-
0.5	25000	1.0324	-
0.51	25500	1.0211	-
0.52	26000	1.0055	-
0.53	26500	1.0083	-
0.54	27000	1.0241	-
0.55	27500	1.0217	-
0.56	28000	1.0202	-
0.57	28500	0.9915	-
0.58	29000	0.9601	-
0.59	29500	1.0089	-
0.6	30000	0.9765	-
0.61	30500	0.9635	-
0.62	31000	0.9602	-
0.63	31500	0.9607	-
0.64	32000	0.9732	-
0.65	32500	0.9554	-
0.66	33000	0.9314	-
0.67	33500	0.9533	-
0.68	34000	0.9554	-
0.69	34500	0.9422	-
0.7	35000	0.9356	-
0.71	35500	0.9478	-
0.72	36000	0.9426	-
0.73	36500	0.9425	-
0.74	37000	0.9447	-
0.75	37500	0.9251	-
0.76	38000	0.88	-
0.77	38500	0.9306	-
0.78	39000	0.9269	-
0.79	39500	0.8957	-
0.8	40000	0.9576	-
0.81	40500	0.921	-
0.82	41000	0.9027	-
0.83	41500	0.8977	-
0.84	42000	0.9079	-
0.85	42500	0.911	-
0.86	43000	0.9039	-
0.87	43500	0.9062	-
0.88	44000	0.9055	-
0.89	44500	0.8933	-
0.9	45000	0.8942	-
0.91	45500	0.9035	-
0.92	46000	0.8864	-
0.93	46500	0.8937	-
0.94	47000	0.8727	-
0.95	47500	0.8645	-
0.96	48000	0.8426	-
0.97	48500	0.8701	-
0.98	49000	0.8517	-
0.99	49500	0.8635	-
1.0	50000	0.8637	0.0043
1.01	50500	0.7842	-
1.02	51000	0.8217	-
1.03	51500	0.8421	-
1.04	52000	0.7827	-
1.05	52500	0.8445	-
1.06	53000	0.8121	-
1.07	53500	0.8057	-
1.08	54000	0.8027	-
1.09	54500	0.7986	-
1.1	55000	0.8065	-
1.11	55500	0.8096	-
1.12	56000	0.8062	-
1.13	56500	0.7977	-
1.1400	57000	0.7805	-
1.15	57500	0.7898	-
1.16	58000	0.8121	-
1.17	58500	0.777	-
1.18	59000	0.7883	-
1.19	59500	0.7868	-
1.2	60000	0.7805	-
1.21	60500	0.8078	-
1.22	61000	0.7893	-
1.23	61500	0.8214	-
1.24	62000	0.7962	-
1.25	62500	0.7769	-
1.26	63000	0.7775	-
1.27	63500	0.7735	-
1.28	64000	0.8054	-
1.29	64500	0.7613	-
1.3	65000	0.7894	-
1.31	65500	0.7791	-
1.32	66000	0.7628	-
1.33	66500	0.7737	-
1.34	67000	0.7474	-
1.35	67500	0.7773	-
1.3600	68000	0.7615	-
1.37	68500	0.7731	-
1.38	69000	0.7501	-
1.3900	69500	0.7538	-
1.4	70000	0.7782	-
1.41	70500	0.7416	-
1.42	71000	0.7777	-
1.43	71500	0.7485	-
1.44	72000	0.7726	-
1.45	72500	0.7316	-
1.46	73000	0.7297	-
1.47	73500	0.7597	-
1.48	74000	0.7556	-
1.49	74500	0.7501	-
1.5	75000	0.7687	-
1.51	75500	0.7382	-
1.52	76000	0.7349	-
1.53	76500	0.7469	-
1.54	77000	0.7523	-
1.55	77500	0.7543	-
1.56	78000	0.7593	-
1.5700	78500	0.7208	-
1.58	79000	0.7623	-
1.5900	79500	0.7429	-
1.6	80000	0.7274	-
1.6100	80500	0.766	-
1.62	81000	0.7573	-
1.63	81500	0.7422	-
1.6400	82000	0.7288	-
1.65	82500	0.7088	-
1.6600	83000	0.7682	-
1.67	83500	0.7426	-
1.6800	84000	0.7325	-
1.69	84500	0.7128	-
1.7	85000	0.7092	-
1.71	85500	0.7194	-
1.72	86000	0.735	-
1.73	86500	0.7461	-
1.74	87000	0.7232	-
1.75	87500	0.7491	-
1.76	88000	0.7272	-
1.77	88500	0.7214	-
1.78	89000	0.7082	-
1.79	89500	0.692	-
1.8	90000	0.7201	-
1.81	90500	0.6991	-
1.8200	91000	0.7233	-
1.83	91500	0.7144	-
1.8400	92000	0.7128	-
1.85	92500	0.6971	-
1.8600	93000	0.7096	-
1.87	93500	0.7091	-
1.88	94000	0.7256	-
1.8900	94500	0.7154	-
1.9	95000	0.7237	-
1.9100	95500	0.725	-
1.92	96000	0.6913	-
1.9300	96500	0.6932	-
1.94	97000	0.7294	-
1.95	97500	0.7016	-
1.96	98000	0.7033	-
1.97	98500	0.7057	-
1.98	99000	0.7374	-
1.99	99500	0.7036	-
2.0	100000	0.7072	0.0057
2.01	100500	0.6372	-
2.02	101000	0.6088	-
2.03	101500	0.5999	-
2.04	102000	0.6285	-
2.05	102500	0.6135	-
2.06	103000	0.6229	-
2.07	103500	0.6098	-
2.08	104000	0.6228	-
2.09	104500	0.6536	-
2.1	105000	0.6269	-
2.11	105500	0.6015	-
2.12	106000	0.6176	-
2.13	106500	0.6207	-
2.14	107000	0.6466	-
2.15	107500	0.5975	-
2.16	108000	0.5948	-
2.17	108500	0.6076	-
2.18	109000	0.6264	-
2.19	109500	0.6043	-
2.2	110000	0.6118	-
2.21	110500	0.6211	-
2.22	111000	0.5927	-
2.23	111500	0.6021	-
2.24	112000	0.6353	-
2.25	112500	0.6215	-
2.26	113000	0.6167	-
2.27	113500	0.6013	-
2.2800	114000	0.6208	-
2.29	114500	0.6026	-
2.3	115000	0.5851	-
2.31	115500	0.6124	-
2.32	116000	0.6307	-
2.33	116500	0.6032	-
2.34	117000	0.598	-
2.35	117500	0.6496	-
2.36	118000	0.6151	-
2.37	118500	0.5772	-
2.38	119000	0.5902	-
2.39	119500	0.5716	-
2.4	120000	0.6081	-
2.41	120500	0.566	-
2.42	121000	0.5792	-
2.43	121500	0.6168	-
2.44	122000	0.5764	-
2.45	122500	0.6209	-
2.46	123000	0.605	-
2.4700	123500	0.6045	-
2.48	124000	0.5886	-
2.49	124500	0.6054	-
2.5	125000	0.5711	-
2.51	125500	0.589	-
2.52	126000	0.5732	-
2.5300	126500	0.6116	-
2.54	127000	0.5793	-
2.55	127500	0.5502	-
2.56	128000	0.5678	-
2.57	128500	0.5892	-
2.58	129000	0.58	-
2.59	129500	0.5951	-
2.6	130000	0.5732	-
2.61	130500	0.5799	-
2.62	131000	0.5987	-
2.63	131500	0.5716	-
2.64	132000	0.5691	-
2.65	132500	0.5789	-
2.66	133000	0.6035	-
2.67	133500	0.5751	-
2.68	134000	0.5772	-
2.69	134500	0.575	-
2.7	135000	0.6058	-
2.71	135500	0.5622	-
2.7200	136000	0.5709	-
2.73	136500	0.5533	-
2.74	137000	0.5791	-
2.75	137500	0.5647	-
2.76	138000	0.5541	-
2.77	138500	0.5763	-
2.7800	139000	0.5614	-
2.79	139500	0.5589	-
2.8	140000	0.589	-
2.81	140500	0.5852	-
2.82	141000	0.5819	-
2.83	141500	0.5351	-
2.84	142000	0.5831	-
2.85	142500	0.5697	-
2.86	143000	0.564	-
2.87	143500	0.572	-
2.88	144000	0.5882	-
2.89	144500	0.5732	-
2.9	145000	0.5597	-
2.91	145500	0.5541	-
2.92	146000	0.5771	-
2.93	146500	0.5561	-
2.94	147000	0.5855	-
2.95	147500	0.5779	-
2.96	148000	0.5432	-
2.9700	148500	0.5533	-
2.98	149000	0.5591	-
2.99	149500	0.5605	-
3.0	150000	0.5704	0.0067

Framework Versions

Python: 3.11.5
Sentence Transformers: 5.0.0
Transformers: 4.54.0
PyTorch: 2.7.1+cu126
Accelerate: 1.9.0
Datasets: 4.0.0
Tokenizers: 0.21.2

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",
}

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}
}

Downloads last month: 12

Safetensors

Model size

0.6B params

Tensor type

F32

Model tree for nexusmed-ai/adr_bge_m3_embedding

Base model

BAAI/bge-m3

Finetuned

(544)

this model

Papers for nexusmed-ai/adr_bge_m3_embedding

Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks

Paper • 1908.10084 • Published Aug 27, 2019 • 17

Efficient Natural Language Response Suggestion for Smart Reply

Paper • 1705.00652 • Published May 1, 2017

Evaluation results

Cosine Accuracy@1 on Unknown
self-reported

0.002
Cosine Accuracy@3 on Unknown
self-reported

0.005
Cosine Accuracy@5 on Unknown
self-reported

0.008
Cosine Accuracy@10 on Unknown
self-reported

0.013
Cosine Precision@1 on Unknown
self-reported

0.002
Cosine Precision@3 on Unknown
self-reported

0.002
Cosine Precision@5 on Unknown
self-reported

0.002
Cosine Precision@10 on Unknown
self-reported

0.001