yyzheng00/snomed_triplet_1M

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

This is a sentence-transformers model finetuned from sentence-transformers/all-MiniLM-L6-v2 on the parquet dataset. 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
• Training Dataset:
  • parquet

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}) with Transformer model: 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.

from sentence_transformers import SentenceTransformer

# Download from the 🤗 Hub
model = SentenceTransformer("yyzheng00/snomed_triplet_1M")
# Run inference
sentences = [
    '|Estradiol and/or estradiol derivative (substance)| + |Steroid hormone (substance)| + |Substance with estrogen receptor agonist mechanism of action (substance)| : |Has disposition (attribute)| = |Estrogen receptor agonist (disposition)|, ',
    '17-Beta oestradiol (substance)',
    "Rupture of Descemet's membrane of right eye (disorder)",
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 384]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [3, 3]

Evaluation

Metrics

Triplet

• Dataset: snomed_triplet_1M_3_4_3-dev
• Evaluated with TripletEvaluator

Metric	Value
cosine_accuracy	0.9793

Triplet

• Dataset: snomed_triplet_1M_3_4_3-dev
• Evaluated with TripletEvaluator

Metric	Value
cosine_accuracy	0.978

Training Details

Training Dataset

parquet

• Dataset: parquet
• Size: 1,000,000 training samples
• Columns: anchor, positive, and negative
• Approximate statistics based on the first 1000 samples:

  	anchor	positive	negative
  type	string	string	string
  details	min: 7 tokens mean: 50.47 tokens max: 256 tokens	min: 6 tokens mean: 14.36 tokens max: 42 tokens	min: 6 tokens mean: 22.41 tokens max: 256 tokens

• Samples:

  anchor	positive	negative
  Anas versicolor (organism)	Silver teal (organism)	Cryotherapy of gastric lesion (procedure)
  	Vitamin B2 and/or vitamin B2 derivative (substance)	:
  	Aplasia of distal phalanx of fifth toe (disorder)	+

• Loss: TripletLoss with these parameters:

  {
      "distance_metric": "TripletDistanceMetric.COSINE",
      "triplet_margin": 0.2
  }
  

Evaluation Dataset

parquet

• Dataset: parquet
• Size: 1,000,000 evaluation samples
• Columns: anchor, positive, and negative
• Approximate statistics based on the first 1000 samples:

  	anchor	positive	negative
  type	string	string	string
  details	min: 6 tokens mean: 48.58 tokens max: 256 tokens	min: 6 tokens mean: 14.51 tokens max: 45 tokens	min: 6 tokens mean: 20.96 tokens max: 256 tokens

• Samples:

  anchor	positive	negative
  	Genus Roseateles (organism)
  	Partial urinary cystectomy (procedure)	+
  	Product containing integrase strand transfer inhibitor (product)	+

• Loss: TripletLoss with these parameters:

  {
      "distance_metric": "TripletDistanceMetric.COSINE",
      "triplet_margin": 0.2
  }
  

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 16
• per_device_eval_batch_size: 16
• num_train_epochs: 1
• warmup_ratio: 0.1
• fp16: True
• batch_sampler: no_duplicates

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: 16
• 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: 1.0
• num_train_epochs: 1
• max_steps: -1
• lr_scheduler_type: linear
• lr_scheduler_kwargs: {}
• warmup_ratio: 0.1
• 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: 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: 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: no_duplicates
• multi_dataset_batch_sampler: proportional

Training Logs

Click to expand

Epoch	Step	Training Loss	Validation Loss	snomed_triplet_1M_3_4_3-dev_cosine_accuracy
0.0027	100	0.0553	0.0405	0.9199
0.0053	200	0.0412	0.0316	0.9369
0.008	300	0.0277	0.0296	0.9405
0.0107	400	0.0303	0.0282	0.9433
0.0133	500	0.0262	0.0275	0.9450
0.016	600	0.0293	0.0266	0.9466
0.0187	700	0.0301	0.0257	0.9480
0.0213	800	0.0262	0.0249	0.9506
0.024	900	0.0258	0.0240	0.9527
0.0267	1000	0.0286	0.0235	0.9537
0.0293	1100	0.0239	0.0229	0.9547
0.032	1200	0.0211	0.0231	0.9548
0.0347	1300	0.0235	0.0228	0.9555
0.0373	1400	0.0257	0.0225	0.9559
0.04	1500	0.025	0.0217	0.9572
0.0427	1600	0.0216	0.0214	0.9581
0.0453	1700	0.0247	0.0214	0.9580
0.048	1800	0.0229	0.0212	0.9588
0.0507	1900	0.0207	0.0211	0.9585
0.0533	2000	0.0224	0.0214	0.9585
0.056	2100	0.0237	0.0209	0.9587
0.0587	2200	0.0205	0.0205	0.9591
0.0613	2300	0.0218	0.0208	0.9590
0.064	2400	0.0209	0.0204	0.9601
0.0667	2500	0.0225	0.0207	0.9591
0.0693	2600	0.021	0.0206	0.9604
0.072	2700	0.0222	0.0197	0.9622
0.0747	2800	0.0214	0.0198	0.9615
0.0773	2900	0.0204	0.0200	0.9611
0.08	3000	0.026	0.0197	0.9622
0.0827	3100	0.0181	0.0197	0.9617
0.0853	3200	0.023	0.0195	0.9612
0.088	3300	0.0198	0.0195	0.9620
0.0907	3400	0.0205	0.0198	0.9611
0.0933	3500	0.0208	0.0194	0.9622
0.096	3600	0.0205	0.0205	0.9592
0.0987	3700	0.0242	0.0196	0.9619
0.1013	3800	0.0178	0.0191	0.9634
0.104	3900	0.0189	0.0189	0.9629
0.1067	4000	0.0249	0.0188	0.9637
0.1093	4100	0.0201	0.0186	0.9634
0.112	4200	0.0198	0.0185	0.9636
0.1147	4300	0.0208	0.0186	0.9639
0.1173	4400	0.019	0.0185	0.9639
0.12	4500	0.0203	0.0188	0.9638
0.1227	4600	0.0205	0.0191	0.9633
0.1253	4700	0.0183	0.0194	0.9623
0.128	4800	0.022	0.0183	0.9643
0.1307	4900	0.0193	0.0182	0.9649
0.1333	5000	0.0192	0.0178	0.9659
0.136	5100	0.0212	0.0185	0.9650
0.1387	5200	0.0181	0.0183	0.9639
0.1413	5300	0.0189	0.0177	0.9656
0.144	5400	0.0209	0.0179	0.9658
0.1467	5500	0.0216	0.0175	0.9665
0.1493	5600	0.0178	0.0176	0.9665
0.152	5700	0.019	0.0178	0.9658
0.1547	5800	0.0215	0.0180	0.9655
0.1573	5900	0.0194	0.0176	0.9663
0.16	6000	0.0182	0.0181	0.9651
0.1627	6100	0.0186	0.0185	0.9640
0.1653	6200	0.019	0.0178	0.9650
0.168	6300	0.019	0.0172	0.9667
0.1707	6400	0.0186	0.0178	0.9654
0.1733	6500	0.0192	0.0172	0.9669
0.176	6600	0.0185	0.0171	0.9670
0.1787	6700	0.019	0.0169	0.9674
0.1813	6800	0.0183	0.0170	0.9671
0.184	6900	0.0199	0.0168	0.9675
0.1867	7000	0.0186	0.0169	0.9673
0.1893	7100	0.016	0.0169	0.9676
0.192	7200	0.0158	0.0174	0.9663
0.1947	7300	0.0205	0.0169	0.9681
0.1973	7400	0.0189	0.0169	0.9669
0.2	7500	0.0188	0.0170	0.9672
0.2027	7600	0.0193	0.0168	0.9674
0.2053	7700	0.0202	0.0168	0.9673
0.208	7800	0.0184	0.0165	0.9676
0.2107	7900	0.0196	0.0162	0.9687
0.2133	8000	0.0186	0.0161	0.9688
0.216	8100	0.0174	0.0166	0.9670
0.2187	8200	0.0178	0.0166	0.9676
0.2213	8300	0.0187	0.0172	0.9664
0.224	8400	0.0175	0.0162	0.9685
0.2267	8500	0.0165	0.0163	0.9674
0.2293	8600	0.018	0.0164	0.9678
0.232	8700	0.0192	0.0165	0.9680
0.2347	8800	0.0182	0.0164	0.9680
0.2373	8900	0.0191	0.0162	0.9689
0.24	9000	0.0173	0.0161	0.9683
0.2427	9100	0.022	0.0159	0.9685
0.2453	9200	0.0182	0.0161	0.9685
0.248	9300	0.0174	0.0165	0.9684
0.2507	9400	0.0181	0.0168	0.9667
0.2533	9500	0.0159	0.0163	0.9684
0.256	9600	0.0176	0.0162	0.9685
0.2587	9700	0.0155	0.0170	0.9668
0.2613	9800	0.0183	0.0162	0.9679
0.264	9900	0.0183	0.0156	0.9693
0.2667	10000	0.019	0.0156	0.9695
0.2693	10100	0.0167	0.0162	0.9683
0.272	10200	0.0202	0.0156	0.9695
0.2747	10300	0.0174	0.0157	0.9694
0.2773	10400	0.0165	0.0155	0.9694
0.28	10500	0.0176	0.0155	0.9700
0.2827	10600	0.0181	0.0153	0.9699
0.2853	10700	0.0184	0.0154	0.9697
0.288	10800	0.0172	0.0155	0.9692
0.2907	10900	0.0153	0.0156	0.9694
0.2933	11000	0.0169	0.0154	0.9700
0.296	11100	0.0181	0.0153	0.9698
0.2987	11200	0.0164	0.0154	0.9700
0.3013	11300	0.0177	0.0158	0.9691
0.304	11400	0.0154	0.0153	0.9700
0.3067	11500	0.0159	0.0153	0.9700
0.3093	11600	0.0162	0.0152	0.9699
0.312	11700	0.0172	0.0150	0.9710
0.3147	11800	0.0151	0.0153	0.9696
0.3173	11900	0.0157	0.0153	0.9697
0.32	12000	0.0145	0.0150	0.9705
0.3227	12100	0.0184	0.0153	0.9701
0.3253	12200	0.0173	0.0151	0.9706
0.328	12300	0.0158	0.0151	0.971
0.3307	12400	0.0154	0.0154	0.9697
0.3333	12500	0.0126	0.0153	0.9697
0.336	12600	0.0151	0.0150	0.9704
0.3387	12700	0.0152	0.0152	0.9698
0.3413	12800	0.0176	0.0150	0.9707
0.344	12900	0.0172	0.0149	0.9705
0.3467	13000	0.0149	0.0151	0.9704
0.3493	13100	0.0154	0.0151	0.9701
0.352	13200	0.0138	0.0148	0.9705
0.3547	13300	0.0195	0.0149	0.9705
0.3573	13400	0.0162	0.0151	0.9707
0.36	13500	0.0137	0.0150	0.9708
0.3627	13600	0.0153	0.0151	0.9704
0.3653	13700	0.0143	0.0150	0.9705
0.368	13800	0.0161	0.0149	0.9709
0.3707	13900	0.0136	0.0149	0.9712
0.3733	14000	0.0161	0.0150	0.9709
0.376	14100	0.0171	0.0148	0.9718
0.3787	14200	0.0168	0.0147	0.9717
0.3813	14300	0.0159	0.0147	0.9718
0.384	14400	0.0167	0.0145	0.9721
0.3867	14500	0.0158	0.0147	0.9715
0.3893	14600	0.0153	0.0146	0.9713
0.392	14700	0.0131	0.0145	0.9717
0.3947	14800	0.0166	0.0144	0.9722
0.3973	14900	0.0164	0.0142	0.9720
0.4	15000	0.0166	0.0143	0.9720
0.4027	15100	0.0168	0.0143	0.9726
0.4053	15200	0.0145	0.0143	0.9723
0.408	15300	0.0149	0.0144	0.9717
0.4107	15400	0.0152	0.0141	0.9729
0.4133	15500	0.0147	0.0140	0.9734
0.416	15600	0.0141	0.0140	0.9731
0.4187	15700	0.0147	0.0140	0.9731
0.4213	15800	0.0158	0.0139	0.9734
0.424	15900	0.0177	0.0141	0.9728
0.4267	16000	0.0151	0.0137	0.9734
0.4293	16100	0.0148	0.0145	0.9724
0.432	16200	0.0135	0.0144	0.9721
0.4347	16300	0.0167	0.0138	0.9736
0.4373	16400	0.0153	0.0138	0.9739
0.44	16500	0.014	0.0139	0.9731
0.4427	16600	0.0168	0.0139	0.9734
0.4453	16700	0.0125	0.0139	0.9734
0.448	16800	0.0163	0.0139	0.9733
0.4507	16900	0.0179	0.0137	0.9742
0.4533	17000	0.0162	0.0136	0.9738
0.456	17100	0.0148	0.0137	0.9734
0.4587	17200	0.0154	0.0137	0.9737
0.4613	17300	0.0178	0.0139	0.9732
0.464	17400	0.0176	0.0138	0.9731
0.4667	17500	0.012	0.0135	0.9738
0.4693	17600	0.0136	0.0137	0.9731
0.472	17700	0.0156	0.0133	0.9740
0.4747	17800	0.0151	0.0136	0.9738
0.4773	17900	0.0145	0.0135	0.9741
0.48	18000	0.0176	0.0136	0.9735
0.4827	18100	0.0143	0.0133	0.9744
0.4853	18200	0.0144	0.0133	0.9742
0.488	18300	0.0139	0.0135	0.9738
0.4907	18400	0.0134	0.0134	0.9740
0.4933	18500	0.0135	0.0134	0.9738
0.496	18600	0.0144	0.0134	0.9738
0.4987	18700	0.0143	0.0135	0.9744
0.5013	18800	0.0165	0.0133	0.9748
0.504	18900	0.0147	0.0133	0.9742
0.5067	19000	0.0159	0.0133	0.9743
0.5093	19100	0.013	0.0132	0.9746
0.512	19200	0.0145	0.0133	0.9744
0.5147	19300	0.0147	0.0134	0.9743
0.5173	19400	0.0151	0.0131	0.9748
0.52	19500	0.0134	0.0132	0.9742
0.5227	19600	0.0148	0.0135	0.9740
0.5253	19700	0.0142	0.0134	0.9744
0.528	19800	0.0158	0.0132	0.9746
0.5307	19900	0.015	0.0134	0.9748
0.5333	20000	0.0146	0.0132	0.9745
0.536	20100	0.0136	0.0130	0.9752
0.5387	20200	0.0142	0.0131	0.9750
0.5413	20300	0.0137	0.0130	0.9749
0.544	20400	0.0118	0.0132	0.9741
0.5467	20500	0.0129	0.0131	0.9750
0.5493	20600	0.015	0.0131	0.9749
0.552	20700	0.0154	0.0132	0.9743
0.5547	20800	0.0165	0.0132	0.9747
0.5573	20900	0.0158	0.0131	0.9751
0.56	21000	0.014	0.0130	0.9746
0.5627	21100	0.0157	0.0129	0.9755
0.5653	21200	0.014	0.0129	0.9754
0.568	21300	0.0149	0.0129	0.9751
0.5707	21400	0.0114	0.0129	0.9754
0.5733	21500	0.0116	0.0128	0.9755
0.576	21600	0.0114	0.0132	0.9743
0.5787	21700	0.0164	0.0127	0.9759
0.5813	21800	0.0137	0.0127	0.9754
0.584	21900	0.0118	0.0129	0.9745
0.5867	22000	0.0126	0.0129	0.9752
0.5893	22100	0.0153	0.0126	0.9758
0.592	22200	0.0128	0.0126	0.9759
0.5947	22300	0.0161	0.0128	0.9755
0.5973	22400	0.0121	0.0128	0.9754
0.6	22500	0.0144	0.0126	0.9758
0.6027	22600	0.0138	0.0127	0.9754
0.6053	22700	0.0114	0.0125	0.9757
0.608	22800	0.0163	0.0126	0.9755
0.6107	22900	0.0127	0.0125	0.9757
0.6133	23000	0.0139	0.0126	0.9752
0.616	23100	0.015	0.0126	0.9754
0.6187	23200	0.0128	0.0124	0.9759
0.6213	23300	0.0127	0.0126	0.9758
0.624	23400	0.0137	0.0126	0.9755
0.6267	23500	0.0171	0.0125	0.9760
0.6293	23600	0.0154	0.0123	0.9761
0.632	23700	0.0133	0.0125	0.9757
0.6347	23800	0.0147	0.0122	0.9762
0.6373	23900	0.012	0.0123	0.9759
0.64	24000	0.0121	0.0124	0.9762
0.6427	24100	0.0156	0.0122	0.9768
0.6453	24200	0.0135	0.0122	0.9763
0.648	24300	0.0111	0.0123	0.9762
0.6507	24400	0.0131	0.0121	0.9766
0.6533	24500	0.0166	0.0120	0.9766
0.656	24600	0.0145	0.0121	0.9764
0.6587	24700	0.0138	0.0122	0.9763
0.6613	24800	0.0127	0.0120	0.9766
0.664	24900	0.0142	0.0120	0.9767
0.6667	25000	0.0119	0.0122	0.9764
0.6693	25100	0.0157	0.0120	0.9768
0.672	25200	0.0126	0.0119	0.9769
0.6747	25300	0.0113	0.0119	0.9772
0.6773	25400	0.0138	0.0121	0.9767
0.68	25500	0.0135	0.0124	0.9759
0.6827	25600	0.0147	0.0120	0.9765
0.6853	25700	0.0119	0.0120	0.9764
0.688	25800	0.0167	0.0120	0.9765
0.6907	25900	0.0132	0.0120	0.9767
0.6933	26000	0.0144	0.0118	0.9768
0.696	26100	0.0135	0.0118	0.9771
0.6987	26200	0.0156	0.0119	0.9769
0.7013	26300	0.0132	0.0119	0.9769
0.704	26400	0.0139	0.0120	0.9769
0.7067	26500	0.014	0.0118	0.9771
0.7093	26600	0.0133	0.0118	0.9770
0.712	26700	0.0142	0.0118	0.9773
0.7147	26800	0.0113	0.0117	0.977
0.7173	26900	0.0142	0.0117	0.977
0.72	27000	0.0112	0.0117	0.9771
0.7227	27100	0.012	0.0118	0.9768
0.7253	27200	0.0135	0.0117	0.9768
0.728	27300	0.0126	0.0116	0.9769
0.7307	27400	0.0136	0.0117	0.9767
0.7333	27500	0.013	0.0116	0.9770
0.736	27600	0.0131	0.0117	0.9767
0.7387	27700	0.0127	0.0116	0.9772
0.7413	27800	0.0124	0.0116	0.9770
0.744	27900	0.011	0.0116	0.9771
0.7467	28000	0.0159	0.0116	0.9770
0.7493	28100	0.0118	0.0116	0.9770
0.752	28200	0.0146	0.0115	0.9773
0.7547	28300	0.0112	0.0116	0.9772
0.7573	28400	0.0116	0.0115	0.9776
0.76	28500	0.0115	0.0115	0.9775
0.7627	28600	0.0137	0.0115	0.9779
0.7653	28700	0.0106	0.0115	0.9777
0.768	28800	0.011	0.0116	0.9774
0.7707	28900	0.0132	0.0115	0.9774
0.7733	29000	0.0119	0.0114	0.9776
0.776	29100	0.0121	0.0114	0.9779
0.7787	29200	0.0136	0.0113	0.9780
0.7813	29300	0.0114	0.0114	0.9779
0.784	29400	0.0122	0.0115	0.9778
0.7867	29500	0.0117	0.0114	0.9780
0.7893	29600	0.0119	0.0114	0.9778
0.792	29700	0.0145	0.0114	0.9778
0.7947	29800	0.0098	0.0113	0.9779
0.7973	29900	0.015	0.0114	0.9777
0.8	30000	0.0123	0.0113	0.9779
0.8027	30100	0.0111	0.0115	0.9774
0.8053	30200	0.0126	0.0114	0.9778
0.808	30300	0.0131	0.0113	0.9783
0.8107	30400	0.0131	0.0113	0.9784
0.8133	30500	0.0113	0.0113	0.9783
0.816	30600	0.0131	0.0113	0.9783
0.8187	30700	0.0137	0.0113	0.9782
0.8213	30800	0.0119	0.0112	0.9784
0.824	30900	0.0127	0.0113	0.9782
0.8267	31000	0.0114	0.0112	0.9787
0.8293	31100	0.0116	0.0111	0.9784
0.832	31200	0.0117	0.0112	0.9784
0.8347	31300	0.0128	0.0112	0.9782
0.8373	31400	0.0125	0.0112	0.9782
0.84	31500	0.0136	0.0111	0.9787
0.8427	31600	0.0121	0.0111	0.9785
0.8453	31700	0.0137	0.0112	0.9785
0.848	31800	0.0115	0.0111	0.9786
0.8507	31900	0.0111	0.0111	0.9784
0.8533	32000	0.012	0.0111	0.9786
0.856	32100	0.0115	0.0111	0.9787
0.8587	32200	0.0125	0.0111	0.9785
0.8613	32300	0.0111	0.0111	0.9788
0.864	32400	0.0127	0.0111	0.9788
0.8667	32500	0.0126	0.0110	0.9788
0.8693	32600	0.012	0.0111	0.9788
0.872	32700	0.0117	0.0111	0.9787
0.8747	32800	0.0136	0.0110	0.9787
0.8773	32900	0.0118	0.0110	0.9788
0.88	33000	0.015	0.0110	0.9789
0.8827	33100	0.0105	0.0110	0.9788
0.8853	33200	0.0135	0.0110	0.9786
0.888	33300	0.0099	0.0110	0.9790
0.8907	33400	0.013	0.0109	0.9787
0.8933	33500	0.0149	0.0109	0.9788
0.896	33600	0.012	0.0109	0.9789
0.8987	33700	0.01	0.0110	0.9788
0.9013	33800	0.0132	0.0110	0.9788
0.904	33900	0.0138	0.0109	0.9791
0.9067	34000	0.0107	0.0109	0.9789
0.9093	34100	0.0133	0.0109	0.9789
0.912	34200	0.0124	0.0109	0.9788
0.9147	34300	0.0119	0.0109	0.9788
0.9173	34400	0.0101	0.0109	0.9787
0.92	34500	0.0135	0.0109	0.9790
0.9227	34600	0.0116	0.0109	0.9789
0.9253	34700	0.0116	0.0109	0.9791
0.928	34800	0.0082	0.0108	0.9791
0.9307	34900	0.0129	0.0108	0.9791
0.9333	35000	0.0129	0.0108	0.9792
0.936	35100	0.0147	0.0108	0.9791
0.9387	35200	0.0112	0.0108	0.9790
0.9413	35300	0.0108	0.0108	0.9790
0.944	35400	0.0114	0.0108	0.9791
0.9467	35500	0.0096	0.0108	0.9792
0.9493	35600	0.0111	0.0108	0.9790
0.952	35700	0.0131	0.0108	0.9790
0.9547	35800	0.0147	0.0108	0.9792
0.9573	35900	0.0121	0.0108	0.9792
0.96	36000	0.0105	0.0108	0.9791
0.9627	36100	0.0081	0.0108	0.9791
0.9653	36200	0.013	0.0108	0.9791
0.968	36300	0.0121	0.0108	0.9792
0.9707	36400	0.0122	0.0108	0.9792
0.9733	36500	0.0121	0.0108	0.9792
0.976	36600	0.011	0.0108	0.9792
0.9787	36700	0.0109	0.0107	0.9792
0.9813	36800	0.0114	0.0107	0.9792
0.984	36900	0.0113	0.0107	0.9793
0.9867	37000	0.0111	0.0107	0.9794
0.9893	37100	0.0097	0.0107	0.9793
0.992	37200	0.0127	0.0107	0.9793
0.9947	37300	0.0143	0.0107	0.9794
0.9973	37400	0.0103	0.0107	0.9794
1.0	37500	0.014	0.0107	0.9780

Framework Versions

• Python: 3.11.1
• Sentence Transformers: 3.3.1
• Transformers: 4.47.0
• PyTorch: 2.1.1+cu121
• Accelerate: 1.2.0
• Datasets: 2.18.0
• 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",
}

TripletLoss

@misc{hermans2017defense,
    title={In Defense of the Triplet Loss for Person Re-Identification},
    author={Alexander Hermans and Lucas Beyer and Bastian Leibe},
    year={2017},
    eprint={1703.07737},
    archivePrefix={arXiv},
    primaryClass={cs.CV}
}