README.md

metadata

base_model: aubmindlab/bert-base-arabertv02
library_name: sentence-transformers
metrics:
  - pearson_cosine
  - spearman_cosine
  - pearson_manhattan
  - spearman_manhattan
  - pearson_euclidean
  - spearman_euclidean
  - pearson_dot
  - spearman_dot
  - pearson_max
  - spearman_max
pipeline_tag: sentence-similarity
tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - generated_from_trainer
  - dataset_size:2279719
  - loss:MatryoshkaLoss
  - loss:MultipleNegativesRankingLoss
widget:
  - source_sentence: ما هو علاج الفطريات الجلدية؟
    sentences:
      - >-
        كيف سيؤثر ذلك على الطلاب الهنود الذين يدرسون أو يعملون في الولايات
        المتحدة إذا أصبح ترامب رئيساً؟
      - كيف يمكنك معالجة الأكزيما بشكل طبيعي؟
      - كيف تعالج الفطريات الجلدية؟
  - source_sentence: >-
      So Eric had an initial design idea for a robot, but we didn't have all the
      parts figured out, so we did what anybody would do in our situation: we
      asked the Internet for help.
    sentences:
      - >-
        وهكذا أول شيء فعلناه هو , بمجرد أن التسلسل خرج من الماكينات , نشرناه على
        الإنترنت .
      - >-
        وكانت لدى "إريك" فكرة مبدئية لصناعة روبوت، ولكن لم يكن لدينا فكرة عن
        القطع التي نحتاجها لذلك قمنا بما يمكن أن يقوم به أي شخص بوضعنا قمنا بطلب
        المساعدة عبر الإنترنت
      - >-
        ما هي مواقع الويب التي يجب اتباعها لتوصيات الأسهم خلال اليوم في سوق
        الأسهم الهندية؟
  - source_sentence: Well, guess what? In England, it's seven per 100,000.
    sentences:
      - عندما نكون أطفالًا، نتعلم الضحك، ونتعلم الضحك بشكل أساسي في اللعب.
      - هذا ليس 10000 دولارا، إنه بالعملة المحلية .
      - خمنوا ماذا؟ في إنكلترا، النسبة سبع في كل 000 100.
  - source_sentence: ما هي العوامل الحيوية وغير الحيوية؟ كيف تختلف عن بعضها البعض؟
    sentences:
      - ما هي بعض النصائح لتعلم لغة بايثون؟
      - كما تم تسجيل نتائج إيجابية لثلاثة أيام متتالية.
      - كيف تقارن العوامل الحيوية والعوامل غير الحيوية وتتناقض؟
  - source_sentence: >-
      And the piece of art he bought at the yard sale is hanging in his
      classroom; he's a teacher now.
    sentences:
      - هل الرياضيات لغة أخرى؟
      - تدريجيا، أصبحت هذه العصافير بمثابة معلمين له.
      - >-
        أما اللوحات التي أشتراها منّي فهي معلّقة الآن في غرفة الصف خاصّته؛ فقد
        أصبح مدرّساً.
model-index:
  - name: SentenceTransformer based on aubmindlab/bert-base-arabertv02
    results:
      - task:
          type: semantic-similarity
          name: Semantic Similarity
        dataset:
          name: sts dev 768
          type: sts-dev-768
        metrics:
          - type: pearson_cosine
            value: 0.8410341962006318
            name: Pearson Cosine
          - type: spearman_cosine
            value: 0.8422963798504417
            name: Spearman Cosine
          - type: pearson_manhattan
            value: 0.8119358373898954
            name: Pearson Manhattan
          - type: spearman_manhattan
            value: 0.8260328397910858
            name: Spearman Manhattan
          - type: pearson_euclidean
            value: 0.8138598024349573
            name: Pearson Euclidean
          - type: spearman_euclidean
            value: 0.831707795171752
            name: Spearman Euclidean
          - type: pearson_dot
            value: 0.8371709698109359
            name: Pearson Dot
          - type: spearman_dot
            value: 0.8389681969788781
            name: Spearman Dot
          - type: pearson_max
            value: 0.8410341962006318
            name: Pearson Max
          - type: spearman_max
            value: 0.8422963798504417
            name: Spearman Max
      - task:
          type: semantic-similarity
          name: Semantic Similarity
        dataset:
          name: sts dev 512
          type: sts-dev-512
        metrics:
          - type: pearson_cosine
            value: 0.8408199016320912
            name: Pearson Cosine
          - type: spearman_cosine
            value: 0.8415754271206667
            name: Spearman Cosine
          - type: pearson_manhattan
            value: 0.8114852653680014
            name: Pearson Manhattan
          - type: spearman_manhattan
            value: 0.8231951698466913
            name: Spearman Manhattan
          - type: pearson_euclidean
            value: 0.8125911836775428
            name: Pearson Euclidean
          - type: spearman_euclidean
            value: 0.8267107276111355
            name: Spearman Euclidean
          - type: pearson_dot
            value: 0.8357223021732401
            name: Pearson Dot
          - type: spearman_dot
            value: 0.8377004761329118
            name: Spearman Dot
          - type: pearson_max
            value: 0.8408199016320912
            name: Pearson Max
          - type: spearman_max
            value: 0.8415754271206667
            name: Spearman Max

SentenceTransformer based on aubmindlab/bert-base-arabertv02

This is a sentence-transformers model finetuned from aubmindlab/bert-base-arabertv02. It maps sentences & paragraphs to a 768-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: aubmindlab/bert-base-arabertv02
• Maximum Sequence Length: 512 tokens
• Output Dimensionality: 768 tokens
• 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': 768, '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})
)

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("silma-ai/silma-embeddding-matryoshka-0.1")
# Run inference
sentences = [
    "And the piece of art he bought at the yard sale is hanging in his classroom; he's a teacher now.",
    'أما اللوحات التي أشتراها منّي فهي معلّقة الآن في غرفة الصف خاصّته؛ فقد أصبح مدرّساً.',
    'تدريجيا، أصبحت هذه العصافير بمثابة معلمين له.',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 768]

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

Evaluation

Metrics

Semantic Similarity

• Dataset: sts-dev-768
• Evaluated with EmbeddingSimilarityEvaluator

Metric	Value
pearson_cosine	0.841
spearman_cosine	0.8423
pearson_manhattan	0.8119
spearman_manhattan	0.826
pearson_euclidean	0.8139
spearman_euclidean	0.8317
pearson_dot	0.8372
spearman_dot	0.839
pearson_max	0.841
spearman_max	0.8423

Semantic Similarity

• Dataset: sts-dev-512
• Evaluated with EmbeddingSimilarityEvaluator

Metric	Value
pearson_cosine	0.8408
spearman_cosine	0.8416
pearson_manhattan	0.8115
spearman_manhattan	0.8232
pearson_euclidean	0.8126
spearman_euclidean	0.8267
pearson_dot	0.8357
spearman_dot	0.8377
pearson_max	0.8408
spearman_max	0.8416

Training Details

Training Dataset

Unnamed Dataset

• Size: 2,279,719 training samples
• Columns: anchor, positive, and negative
• Approximate statistics based on the first 1000 samples:

  	anchor	positive	negative
  type	string	string	string
  details	min: 4 tokens mean: 19.51 tokens max: 139 tokens	min: 4 tokens mean: 12.47 tokens max: 59 tokens	min: 4 tokens mean: 12.13 tokens max: 72 tokens

• Samples:

  anchor	positive	negative
  كيف أصنع صاروخاً؟	كيف أصنع صاروخاً صناعياً؟	كيف أصنع أول روبوت لي؟
  فتاة شابة تجلس على طاولة مع وعاء على رأسها	فتاة صغيرة لديها وعاء على رأسها	رجل يأكل الحبوب في سيارته
  كيف يمكنني الانضمام إلى الجيش الهندي بعد البكالوريوس؟	كيف تنضم للجيش الهندي بعد الهندسة؟	كيف لي أن أعرف ماذا أريد أن أفعل في حياتي؟

• Loss: MatryoshkaLoss with these parameters:

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

Evaluation Dataset

Unnamed Dataset

• Size: 600 evaluation samples
• Columns: anchor, positive, and negative
• Approximate statistics based on the first 600 samples:

  	anchor	positive	negative
  type	string	string	string
  details	min: 4 tokens mean: 19.5 tokens max: 146 tokens	min: 4 tokens mean: 12.67 tokens max: 43 tokens	min: 4 tokens mean: 12.15 tokens max: 41 tokens

• Samples:

  anchor	positive	negative
  And this explanation represents great progress.	وهذا التفسير يمثل تقدماً عظيماً	وأظهرت هذا الإتجاه المذهل.
  ثلاثة رجال يلعبون كرة السلة	ثلاثة رجال يلعبون لعبة كرة السلة	رجلين يرتديان ملابس غريبة يقفزان على ملعب كرة السلة
  الرجل جالس	رجل يرتدي قميصاً أحمر يعزف الطبول.	رجل في قميص رمادي يقف.

• Loss: MatryoshkaLoss with these parameters:

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

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 50
• per_device_eval_batch_size: 10
• learning_rate: 1e-05
• bf16: 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: 50
• 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: 1e-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: 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: True
• 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: True
• 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: False
• hub_always_push: False
• gradient_checkpointing: False
• gradient_checkpointing_kwargs: None
• include_inputs_for_metrics: False
• 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
• batch_sampler: no_duplicates
• multi_dataset_batch_sampler: proportional

Training Logs

Click to expand

Epoch	Step	Training Loss	Validation Loss	sts-dev-768_spearman_cosine	sts-dev-512_spearman_cosine
0.0044	50	-	0.7749	0.7784	0.7748
0.0088	100	-	0.6231	0.7854	0.7809
0.0132	150	-	0.5326	0.8028	0.7992
0.0175	200	-	0.4880	0.8103	0.8047
0.0219	250	1.1802	0.4398	0.8084	0.8043
0.0263	300	-	0.4203	0.8108	0.8058
0.0307	350	-	0.3880	0.8134	0.8075
0.0351	400	-	0.3998	0.8180	0.8145
0.0395	450	-	0.3840	0.8154	0.8114
0.0439	500	0.7483	0.3804	0.8105	0.8056
0.0483	550	-	0.3695	0.8147	0.8103
0.0526	600	-	0.3649	0.8145	0.8101
0.0570	650	-	0.3494	0.8192	0.8157
0.0614	700	-	0.3437	0.8159	0.8106
0.0658	750	0.6561	0.3302	0.8158	0.8104
0.0702	800	-	0.3359	0.8204	0.8174
0.0746	850	-	0.3446	0.8119	0.8094
0.0790	900	-	0.3419	0.8265	0.8252
0.0833	950	-	0.3197	0.8177	0.8141
0.0877	1000	0.6178	0.3250	0.8213	0.8185
0.0921	1050	-	0.3017	0.8161	0.8127
0.0965	1100	-	0.3058	0.8232	0.8180
0.1009	1150	-	0.3066	0.8236	0.8193
0.1053	1200	-	0.2924	0.8275	0.8237
0.1097	1250	0.5633	0.3096	0.8206	0.8173
0.1141	1300	-	0.3009	0.8299	0.8277
0.1184	1350	-	0.3067	0.8158	0.8111
0.1228	1400	-	0.2898	0.8215	0.8180
0.1272	1450	-	0.2810	0.8272	0.8261
0.1316	1500	0.5337	0.2810	0.8228	0.8187
0.1360	1550	-	0.2772	0.8167	0.8139
0.1404	1600	-	0.2772	0.8228	0.8194
0.1448	1650	-	0.2751	0.8193	0.8153
0.1491	1700	-	0.2579	0.8182	0.8147
0.1535	1750	0.5154	0.2542	0.8199	0.8166
0.1579	1800	-	0.2607	0.8243	0.8224
0.1623	1850	-	0.2595	0.8280	0.8254
0.1667	1900	-	0.2612	0.8272	0.8255
0.1711	1950	-	0.2644	0.8273	0.8242
0.1755	2000	0.4838	0.2618	0.8276	0.8246
0.1799	2050	-	0.2553	0.8219	0.8200
0.1842	2100	-	0.2581	0.8232	0.8217
0.1886	2150	-	0.2620	0.8254	0.8232
0.1930	2200	-	0.2627	0.8235	0.8193
0.1974	2250	0.486	0.2597	0.8170	0.8142
0.2018	2300	-	0.2605	0.8261	0.8231
0.2062	2350	-	0.2584	0.8252	0.8222
0.2106	2400	-	0.2663	0.8247	0.8228
0.2149	2450	-	0.2527	0.8285	0.8280
0.2193	2500	0.4523	0.2487	0.8291	0.8270
0.2237	2550	-	0.2524	0.8257	0.8244
0.2281	2600	-	0.2513	0.8228	0.8210
0.2325	2650	-	0.2531	0.8287	0.8265
0.2369	2700	-	0.2510	0.8224	0.8198
0.2413	2750	0.4522	0.2523	0.8275	0.8260
0.2457	2800	-	0.2563	0.8301	0.8278
0.2500	2850	-	0.2531	0.8242	0.8242
0.2544	2900	-	0.2527	0.8268	0.8268
0.2588	2950	-	0.2465	0.8228	0.8223
0.2632	3000	0.4472	0.2422	0.8263	0.8237
0.2676	3050	-	0.2484	0.8223	0.8195
0.2720	3100	-	0.2469	0.8209	0.8206
0.2764	3150	-	0.2419	0.8283	0.8281
0.2808	3200	-	0.2370	0.8303	0.8286
0.2851	3250	0.4499	0.2374	0.8293	0.8275
0.2895	3300	-	0.2340	0.8255	0.8255
0.2939	3350	-	0.2461	0.8277	0.8292
0.2983	3400	-	0.2421	0.8320	0.8307
0.3027	3450	-	0.2366	0.8286	0.8281
0.3071	3500	0.4305	0.2389	0.8312	0.8293
0.3115	3550	-	0.2360	0.8305	0.8310
0.3158	3600	-	0.2313	0.8271	0.8256
0.3202	3650	-	0.2182	0.8231	0.8197
0.3246	3700	-	0.2220	0.8274	0.8246
0.3290	3750	0.4221	0.2305	0.8301	0.8292
0.3334	3800	-	0.2244	0.8285	0.8265
0.3378	3850	-	0.2355	0.8349	0.8331
0.3422	3900	-	0.2256	0.8355	0.8330
0.3466	3950	-	0.2273	0.8330	0.8299
0.3509	4000	0.4203	0.2334	0.8304	0.8275
0.3553	4050	-	0.2223	0.8323	0.8305
0.3597	4100	-	0.2314	0.8323	0.8299
0.3641	4150	-	0.2196	0.8272	0.8244
0.3685	4200	-	0.2275	0.8342	0.8353
0.3729	4250	0.4039	0.2209	0.8348	0.8333
0.3773	4300	-	0.2152	0.8314	0.8307
0.3816	4350	-	0.2115	0.8353	0.8325
0.3860	4400	-	0.2195	0.8347	0.8310
0.3904	4450	-	0.2110	0.8293	0.8264
0.3948	4500	0.4065	0.2115	0.8321	0.8293
0.3992	4550	-	0.2139	0.8312	0.8286
0.4036	4600	-	0.2145	0.8319	0.8285
0.4080	4650	-	0.2127	0.8281	0.8255
0.4124	4700	-	0.2122	0.8292	0.8268
0.4167	4750	0.4019	0.2160	0.8354	0.8329
0.4211	4800	-	0.2069	0.8296	0.8258
0.4255	4850	-	0.2106	0.8362	0.8335
0.4299	4900	-	0.2130	0.8345	0.8321
0.4343	4950	-	0.2080	0.8307	0.8277
0.4387	5000	0.3941	0.2184	0.8394	0.8370
0.4431	5050	-	0.2061	0.8334	0.8325
0.4474	5100	-	0.2092	0.8318	0.8307
0.4518	5150	-	0.2108	0.8319	0.8289
0.4562	5200	-	0.2046	0.8359	0.8337
0.4606	5250	0.3873	0.1990	0.8327	0.8305
0.4650	5300	-	0.2007	0.8332	0.8305
0.4694	5350	-	0.1989	0.8284	0.8247
0.4738	5400	-	0.2117	0.8363	0.8346
0.4782	5450	-	0.2036	0.8329	0.8296
0.4825	5500	0.3808	0.1999	0.8341	0.8295
0.4869	5550	-	0.1998	0.8336	0.8300
0.4913	5600	-	0.2040	0.8348	0.8331
0.4957	5650	-	0.2068	0.8367	0.8346
0.5001	5700	-	0.1947	0.8333	0.8305
0.5045	5750	0.3779	0.1969	0.8352	0.8329
0.5089	5800	-	0.2028	0.8372	0.8369
0.5132	5850	-	0.2029	0.8336	0.8319
0.5176	5900	-	0.2029	0.8317	0.8309
0.5220	5950	-	0.2059	0.8270	0.8270
0.5264	6000	0.3704	0.1997	0.8263	0.8236
0.5308	6050	-	0.2001	0.8280	0.8252
0.5352	6100	-	0.1985	0.8275	0.8241
0.5396	6150	-	0.1976	0.8281	0.8281
0.5440	6200	-	0.1987	0.8270	0.8247
0.5483	6250	0.3722	0.2045	0.8320	0.8303
0.5527	6300	-	0.2013	0.8292	0.8278
0.5571	6350	-	0.2007	0.8302	0.8279
0.5615	6400	-	0.1949	0.8297	0.8274
0.5659	6450	-	0.2037	0.8335	0.8313
0.5703	6500	0.3638	0.2060	0.8316	0.8280
0.5747	6550	-	0.2030	0.8372	0.8348
0.5790	6600	-	0.1982	0.8317	0.8295
0.5834	6650	-	0.2075	0.8324	0.8325
0.5878	6700	-	0.2014	0.8306	0.8284
0.5922	6750	0.3581	0.1983	0.8360	0.8344
0.5966	6800	-	0.2007	0.8337	0.8313
0.6010	6850	-	0.2003	0.8349	0.8338
0.6054	6900	-	0.2018	0.8313	0.8305
0.6098	6950	-	0.1978	0.8323	0.8307
0.6141	7000	0.3596	0.1991	0.8370	0.8340
0.6185	7050	-	0.1963	0.8330	0.8302
0.6229	7100	-	0.1918	0.8334	0.8320
0.6273	7150	-	0.2008	0.8338	0.8327
0.6317	7200	-	0.1973	0.8320	0.8295
0.6361	7250	0.3614	0.1891	0.8339	0.8322
0.6405	7300	-	0.1961	0.8355	0.8332
0.6448	7350	-	0.1910	0.8322	0.8304
0.6492	7400	-	0.1926	0.8343	0.8331
0.6536	7450	-	0.1935	0.8310	0.8292
0.6580	7500	0.3513	0.1969	0.8337	0.8346
0.6624	7550	-	0.1891	0.8331	0.8311
0.6668	7600	-	0.1932	0.8369	0.8341
0.6712	7650	-	0.2041	0.8370	0.8357
0.6756	7700	-	0.1946	0.8335	0.8314
0.6799	7750	0.3426	0.1955	0.8364	0.8330
0.6843	7800	-	0.1940	0.8316	0.8307
0.6887	7850	-	0.1893	0.8323	0.8322
0.6931	7900	-	0.1839	0.8296	0.8286
0.6975	7950	-	0.1895	0.8321	0.8296
0.7019	8000	0.3406	0.1901	0.8277	0.8263
0.7063	8050	-	0.1835	0.8331	0.8284
0.7107	8100	-	0.1847	0.8359	0.8342
0.7150	8150	-	0.1892	0.8362	0.8348
0.7194	8200	-	0.1775	0.8339	0.8305
0.7238	8250	0.3357	0.1921	0.8359	0.8340
0.7282	8300	-	0.1881	0.8369	0.8344
0.7326	8350	-	0.1891	0.8371	0.8363
0.7370	8400	-	0.1880	0.8394	0.8364
0.7414	8450	-	0.1892	0.8348	0.8306
0.7457	8500	0.327	0.1868	0.8388	0.8353
0.7501	8550	-	0.1815	0.8378	0.8352
0.7545	8600	-	0.1877	0.8398	0.8370
0.7589	8650	-	0.1878	0.8392	0.8378
0.7633	8700	-	0.1778	0.8330	0.8304
0.7677	8750	0.3288	0.1791	0.8390	0.8360
0.7721	8800	-	0.1803	0.8298	0.8270
0.7765	8850	-	0.1803	0.8358	0.8323
0.7808	8900	-	0.1832	0.8330	0.8322
0.7852	8950	-	0.1767	0.8316	0.8286
0.7896	9000	0.329	0.1808	0.8283	0.8254
0.7940	9050	-	0.1842	0.8331	0.8293
0.7984	9100	-	0.1750	0.8304	0.8275
0.8028	9150	-	0.1779	0.8299	0.8270
0.8072	9200	-	0.1799	0.8332	0.8332
0.8115	9250	0.3283	0.1872	0.8399	0.8371
0.8159	9300	-	0.1842	0.8364	0.8352
0.8203	9350	-	0.1785	0.8415	0.8382
0.8247	9400	-	0.1822	0.8432	0.8407
0.8291	9450	-	0.1745	0.8380	0.8364
0.8335	9500	0.3271	0.1745	0.8374	0.8352
0.8379	9550	-	0.1746	0.8363	0.8332
0.8423	9600	-	0.1776	0.8391	0.8374
0.8466	9650	-	0.1760	0.8379	0.8353
0.8510	9700	-	0.1806	0.8360	0.8335
0.8554	9750	0.3309	0.1822	0.8368	0.8337
0.8598	9800	-	0.1765	0.8366	0.8336
0.8642	9850	-	0.1766	0.8353	0.8323
0.8686	9900	-	0.1698	0.8353	0.8315
0.8730	9950	-	0.1715	0.8378	0.8338
0.8773	10000	0.318	0.1782	0.8396	0.8357
0.8817	10050	-	0.1727	0.8382	0.8368
0.8861	10100	-	0.1740	0.8356	0.8330
0.8905	10150	-	0.1723	0.8347	0.8319
0.8949	10200	-	0.1656	0.8336	0.8314
0.8993	10250	0.3284	0.1742	0.8288	0.8264
0.9037	10300	-	0.1679	0.8315	0.8296
0.9081	10350	-	0.1694	0.8325	0.8296
0.9124	10400	-	0.1723	0.8319	0.8305
0.9168	10450	-	0.1638	0.8340	0.8310
0.9212	10500	0.313	0.1730	0.8371	0.8368
0.9256	10550	-	0.1639	0.8351	0.8327
0.9300	10600	-	0.1634	0.8379	0.8350
0.9344	10650	-	0.1745	0.8353	0.8340
0.9388	10700	-	0.1731	0.8349	0.8346
0.9431	10750	0.3145	0.1668	0.8333	0.8314
0.9475	10800	-	0.1653	0.8351	0.8338
0.9519	10850	-	0.1655	0.8401	0.8390
0.9563	10900	-	0.1708	0.8376	0.8360
0.9607	10950	-	0.1740	0.8382	0.8364
0.9651	11000	0.3002	0.1714	0.8401	0.8382
0.9695	11050	-	0.1647	0.8411	0.8393
0.9739	11100	-	0.1701	0.8418	0.8396
0.9782	11150	-	0.1665	0.8394	0.8379
0.9826	11200	-	0.1652	0.8377	0.8376
0.9870	11250	0.3094	0.1665	0.8408	0.8397
0.9914	11300	-	0.1689	0.8412	0.8393
0.9958	11350	-	0.1674	0.8400	0.8374
1.0002	11400	-	0.1694	0.8395	0.8376
1.0046	11450	-	0.1697	0.8434	0.8419
1.0089	11500	0.3004	0.1640	0.8399	0.8388
1.0133	11550	-	0.1731	0.8445	0.8426
1.0177	11600	-	0.1618	0.8430	0.8389
1.0221	11650	-	0.1646	0.8414	0.8377
1.0265	11700	-	0.1679	0.8435	0.8401
1.0309	11750	0.2984	0.1646	0.8413	0.8385
1.0353	11800	-	0.1797	0.8465	0.8432
1.0397	11850	-	0.1758	0.8393	0.8390
1.0440	11900	-	0.1690	0.8401	0.8379
1.0484	11950	-	0.1735	0.8423	0.8404
1.0528	12000	0.2896	0.1719	0.8384	0.8367
1.0572	12050	-	0.1759	0.8420	0.8403
1.0616	12100	-	0.1659	0.8360	0.8340
1.0660	12150	-	0.1645	0.8368	0.8362
1.0704	12200	-	0.1601	0.8380	0.8355
1.0747	12250	0.2954	0.1711	0.8406	0.8387
1.0791	12300	-	0.1691	0.8389	0.8370
1.0835	12350	-	0.1721	0.8397	0.8385
1.0879	12400	-	0.1689	0.8379	0.8351
1.0923	12450	-	0.1663	0.8424	0.8402
1.0967	12500	0.2864	0.1672	0.8418	0.8403
1.1011	12550	-	0.1689	0.8389	0.8386
1.1055	12600	-	0.1664	0.8410	0.8402
1.1098	12650	-	0.1685	0.8387	0.8376
1.1142	12700	-	0.1715	0.8419	0.8402
1.1186	12750	0.2745	0.1607	0.8373	0.8336
1.1230	12800	-	0.1620	0.8388	0.8379
1.1274	12850	-	0.1623	0.8417	0.8396
1.1318	12900	-	0.1589	0.8360	0.8342
1.1362	12950	-	0.1567	0.8300	0.8298
1.1406	13000	0.2768	0.1557	0.8406	0.8365
1.1449	13050	-	0.1581	0.8389	0.8363
1.1493	13100	-	0.1611	0.8399	0.8366
1.1537	13150	-	0.1583	0.8358	0.8348
1.1581	13200	-	0.1619	0.8405	0.8387
1.1625	13250	0.2737	0.1567	0.8373	0.8339
1.1669	13300	-	0.1642	0.8393	0.8374
1.1713	13350	-	0.1646	0.8404	0.8376
1.1756	13400	-	0.1601	0.8419	0.8402
1.1800	13450	-	0.1648	0.8412	0.8391
1.1844	13500	0.2627	0.1635	0.8403	0.8403
1.1888	13550	-	0.1662	0.8427	0.8407
1.1932	13600	-	0.1687	0.8381	0.8368
1.1976	13650	-	0.1693	0.8366	0.8365
1.2020	13700	-	0.1665	0.8410	0.8397
1.2064	13750	0.2738	0.1665	0.8373	0.8360
1.2107	13800	-	0.1667	0.8388	0.8389
1.2151	13850	-	0.1674	0.8455	0.8413
1.2195	13900	-	0.1704	0.8419	0.8382
1.2239	13950	-	0.1654	0.8417	0.8398
1.2283	14000	0.2563	0.1610	0.8414	0.8403
1.2327	14050	-	0.1625	0.8416	0.8380
1.2371	14100	-	0.1705	0.8411	0.8400
1.2414	14150	-	0.1628	0.8400	0.8384
1.2458	14200	-	0.1667	0.8448	0.8435
1.2502	14250	0.2693	0.1651	0.8406	0.8396
1.2546	14300	-	0.1673	0.8404	0.8388
1.2590	14350	-	0.1630	0.8392	0.8375
1.2634	14400	-	0.1633	0.8413	0.8403
1.2678	14450	-	0.1636	0.8412	0.8398
1.2722	14500	0.266	0.1613	0.8404	0.8379
1.2765	14550	-	0.1625	0.8392	0.8380
1.2809	14600	-	0.1634	0.8418	0.8397
1.2853	14650	-	0.1689	0.8426	0.8428
1.2897	14700	-	0.1617	0.8410	0.8405
1.2941	14750	0.2643	0.1661	0.8437	0.8417
1.2985	14800	-	0.1629	0.8409	0.8394
1.3029	14850	-	0.1584	0.8413	0.8387
1.3072	14900	-	0.1638	0.8446	0.8433
1.3116	14950	-	0.1644	0.8429	0.8426
1.3160	15000	0.2624	0.1570	0.8391	0.8386
1.3204	15050	-	0.1535	0.8367	0.8348
1.3248	15100	-	0.1591	0.8381	0.8367
1.3292	15150	-	0.1618	0.8421	0.8409
1.3336	15200	-	0.1554	0.8402	0.8381
1.3380	15250	0.2621	0.1595	0.8431	0.8427
1.3423	15300	-	0.1595	0.8447	0.8435
1.3467	15350	-	0.1585	0.8408	0.8394
1.3511	15400	-	0.1635	0.8403	0.8389
1.3555	15450	-	0.1569	0.8453	0.8444
1.3599	15500	0.2552	0.1605	0.8434	0.8412
1.3643	15550	-	0.1542	0.8420	0.8397
1.3687	15600	-	0.1622	0.8456	0.8451
1.3730	15650	-	0.1569	0.8466	0.8443
1.3774	15700	-	0.1550	0.8440	0.8416
1.3818	15750	0.2532	0.1569	0.8459	0.8445
1.3862	15800	-	0.1567	0.8462	0.8451
1.3906	15850	-	0.1504	0.8442	0.8422
1.3950	15900	-	0.1524	0.8437	0.8419
1.3994	15950	-	0.1491	0.8438	0.8413
1.4038	16000	0.265	0.1533	0.8428	0.8406
1.4081	16050	-	0.1492	0.8425	0.8399
1.4125	16100	-	0.1486	0.8410	0.8386
1.4169	16150	-	0.1530	0.8458	0.8433
1.4213	16200	-	0.1535	0.8437	0.8427
1.4257	16250	0.2512	0.1508	0.8453	0.8446
1.4301	16300	-	0.1540	0.8427	0.8411
1.4345	16350	-	0.1513	0.8414	0.8388
1.4388	16400	-	0.1553	0.8464	0.8461
1.4432	16450	-	0.1528	0.8434	0.8412
1.4476	16500	0.2545	0.1522	0.8419	0.8399
1.4520	16550	-	0.1521	0.8423	0.8416
1.4564	16600	-	0.1433	0.8427	0.8410
1.4608	16650	-	0.1500	0.8419	0.8401
1.4652	16700	-	0.1442	0.8425	0.8392
1.4696	16750	0.2549	0.1496	0.8397	0.8376
1.4739	16800	-	0.1556	0.8463	0.8435
1.4783	16850	-	0.1510	0.8458	0.8432
1.4827	16900	-	0.1469	0.8431	0.8423
1.4871	16950	-	0.1481	0.8456	0.8441
1.4915	17000	0.2522	0.1512	0.8456	0.8437
1.4959	17050	-	0.1471	0.8455	0.8430
1.5003	17100	-	0.1397	0.8409	0.8383
1.5046	17150	-	0.1414	0.8427	0.8404
1.5090	17200	-	0.1474	0.8432	0.8420
1.5134	17250	0.2489	0.1499	0.8414	0.8412
1.5178	17300	-	0.1442	0.8390	0.8376
1.5222	17350	-	0.1474	0.8373	0.8370
1.5266	17400	-	0.1435	0.8353	0.8352
1.5310	17450	-	0.1461	0.8380	0.8363
1.5354	17500	0.2493	0.1477	0.8362	0.8353
1.5397	17550	-	0.1503	0.8398	0.8385
1.5441	17600	-	0.1474	0.8372	0.8376
1.5485	17650	-	0.1499	0.8408	0.8390
1.5529	17700	-	0.1501	0.8386	0.8369
1.5573	17750	0.2499	0.1474	0.8367	0.8351
1.5617	17800	-	0.1406	0.8380	0.8362
1.5661	17850	-	0.1457	0.8399	0.8396
1.5705	17900	-	0.1486	0.8409	0.8399
1.5748	17950	-	0.1493	0.8407	0.8397
1.5792	18000	0.2419	0.1490	0.8400	0.8386
1.5836	18050	-	0.1496	0.8403	0.8388
1.5880	18100	-	0.1509	0.8422	0.8401
1.5924	18150	-	0.1513	0.8433	0.8420
1.5968	18200	-	0.1546	0.8420	0.8408
1.6012	18250	0.2458	0.1529	0.8414	0.8398
1.6055	18300	-	0.1580	0.8414	0.8391
1.6099	18350	-	0.1483	0.8389	0.8363
1.6143	18400	-	0.1501	0.8419	0.8405
1.6187	18450	-	0.1488	0.8413	0.8388
1.6231	18500	0.2532	0.1499	0.8418	0.8410
1.6275	18550	-	0.1520	0.8409	0.8408
1.6319	18600	-	0.1521	0.8407	0.8392
1.6363	18650	-	0.1459	0.8402	0.8382
1.6406	18700	-	0.1556	0.8433	0.8427
1.6450	18750	0.24	0.1501	0.8421	0.8410
1.6494	18800	-	0.1485	0.8439	0.8425
1.6538	18850	-	0.1526	0.8412	0.8406
1.6582	18900	-	0.1522	0.8422	0.8425
1.6626	18950	-	0.1456	0.8406	0.8390
1.6670	19000	0.2404	0.1483	0.8412	0.8408
1.6713	19050	-	0.1550	0.8424	0.8428
1.6757	19100	-	0.1493	0.8387	0.8384
1.6801	19150	-	0.1523	0.8391	0.8379
1.6845	19200	-	0.1512	0.8366	0.8343
1.6889	19250	0.2401	0.1506	0.8372	0.8348
1.6933	19300	-	0.1457	0.8375	0.8343
1.6977	19350	-	0.1500	0.8403	0.8379
1.7021	19400	-	0.1464	0.8380	0.8367
1.7064	19450	-	0.1485	0.8403	0.8397
1.7108	19500	0.2329	0.1469	0.8450	0.8417
1.7152	19550	-	0.1498	0.8418	0.8391
1.7196	19600	-	0.1427	0.8394	0.8384
1.7240	19650	-	0.1493	0.8399	0.8392
1.7284	19700	-	0.1487	0.8423	0.8406
1.7328	19750	0.2397	0.1464	0.8420	0.8398
1.7371	19800	-	0.1511	0.8433	0.8406
1.7415	19850	-	0.1502	0.8391	0.8365
1.7459	19900	-	0.1527	0.8404	0.8386
1.7503	19950	-	0.1498	0.8397	0.8390
1.7547	20000	0.2312	0.1505	0.8413	0.8389
1.7591	20050	-	0.1525	0.8411	0.8396
1.7635	20100	-	0.1491	0.8380	0.8370
1.7679	20150	-	0.1431	0.8395	0.8382
1.7722	20200	-	0.1451	0.8365	0.8352
1.7766	20250	0.2319	0.1485	0.8388	0.8366
1.7810	20300	-	0.1499	0.8376	0.8367
1.7854	20350	-	0.1448	0.8364	0.8349
1.7898	20400	-	0.1485	0.8346	0.8328
1.7942	20450	-	0.1470	0.8376	0.8364
1.7986	20500	0.2295	0.1471	0.8386	0.8363
1.8029	20550	-	0.1501	0.8351	0.8329
1.8073	20600	-	0.1494	0.8382	0.8364
1.8117	20650	-	0.1489	0.8405	0.8386
1.8161	20700	-	0.1465	0.8381	0.8372
1.8205	20750	0.2408	0.1435	0.8398	0.8390
1.8249	20800	-	0.1498	0.8449	0.8431
1.8293	20850	-	0.1487	0.8431	0.8416
1.8337	20900	-	0.1456	0.8419	0.8394
1.8380	20950	-	0.1437	0.8423	0.8408
1.8424	21000	0.2374	0.1408	0.8425	0.8414
1.8468	21050	-	0.1434	0.8434	0.8418
1.8512	21100	-	0.1486	0.8422	0.8403
1.8556	21150	-	0.1467	0.8429	0.8421
1.8600	21200	-	0.1458	0.8409	0.8402
1.8644	21250	0.2385	0.1449	0.8411	0.8395
1.8687	21300	-	0.1415	0.8401	0.8390
1.8731	21350	-	0.1462	0.8417	0.8403
1.8775	21400	-	0.1468	0.8423	0.8403
1.8819	21450	-	0.1459	0.8417	0.8394
1.8863	21500	0.2302	0.1466	0.8396	0.8372
1.8907	21550	-	0.1479	0.8391	0.8363
1.8951	21600	-	0.1407	0.8382	0.8365
1.8995	21650	-	0.1462	0.8377	0.8355
1.9038	21700	-	0.1438	0.8348	0.8343
1.9082	21750	0.2383	0.1451	0.8371	0.8363
1.9126	21800	-	0.1448	0.8375	0.8360
1.9170	21850	-	0.1389	0.8383	0.8377
1.9214	21900	-	0.1409	0.8379	0.8367
1.9258	21950	-	0.1397	0.8374	0.8352
1.9302	22000	0.2321	0.1408	0.8405	0.8385
1.9345	22050	-	0.1451	0.8381	0.8363
1.9389	22100	-	0.1467	0.8363	0.8353
1.9433	22150	-	0.1459	0.8352	0.8337
1.9477	22200	-	0.1431	0.8382	0.8355
1.9521	22250	0.2282	0.1457	0.8385	0.8371
1.9565	22300	-	0.1475	0.8364	0.8359
1.9609	22350	-	0.1483	0.8370	0.8336
1.9653	22400	-	0.1469	0.8406	0.8373
1.9696	22450	-	0.1430	0.8415	0.8391
1.9740	22500	0.2294	0.1471	0.8417	0.8399
1.9784	22550	-	0.1467	0.8414	0.8413
1.9828	22600	-	0.1464	0.8423	0.8410
1.9872	22650	-	0.1475	0.8431	0.8432
1.9916	22700	-	0.1476	0.8450	0.8442
1.9960	22750	0.2242	0.1463	0.8443	0.8418
2.0004	22800	-	0.1472	0.8422	0.8412
2.0047	22850	-	0.1506	0.8452	0.8435
2.0091	22900	-	0.1478	0.8463	0.8432
2.0135	22950	-	0.1536	0.8479	0.8454
2.0179	23000	0.2249	0.1487	0.8453	0.8422
2.0223	23050	-	0.1484	0.8430	0.8410
2.0267	23100	-	0.1524	0.8454	0.8440
2.0311	23150	-	0.1475	0.8450	0.8422
2.0354	23200	-	0.1533	0.8460	0.8435
2.0398	23250	0.2165	0.1551	0.8428	0.8410
2.0442	23300	-	0.1507	0.8425	0.8400
2.0486	23350	-	0.1517	0.8427	0.8410
2.0530	23400	-	0.1524	0.8404	0.8391
2.0574	23450	-	0.1515	0.8415	0.8408
2.0618	23500	0.2258	0.1500	0.8392	0.8384
2.0662	23550	-	0.1461	0.8387	0.8362
2.0705	23600	-	0.1429	0.8408	0.8378
2.0749	23650	-	0.1473	0.8410	0.8398
2.0793	23700	-	0.1474	0.8415	0.8402
2.0837	23750	0.2309	0.1479	0.8425	0.8408
2.0881	23800	-	0.1493	0.8427	0.8390
2.0925	23850	-	0.1469	0.8419	0.8394
2.0969	23900	-	0.1460	0.8426	0.8406
2.1012	23950	-	0.1502	0.8433	0.8418
2.1056	24000	0.2113	0.1462	0.8423	0.8406
2.1100	24050	-	0.1463	0.8429	0.8398
2.1144	24100	-	0.1459	0.8431	0.8400
2.1188	24150	-	0.1417	0.8403	0.8381
2.1232	24200	-	0.1396	0.8376	0.8371
2.1276	24250	0.2132	0.1419	0.8382	0.8380
2.1320	24300	-	0.1444	0.8378	0.8377
2.1363	24350	-	0.1399	0.8334	0.8342
2.1407	24400	-	0.1363	0.8382	0.8361
2.1451	24450	-	0.1379	0.8381	0.8369
2.1495	24500	0.2124	0.1421	0.8403	0.8391
2.1539	24550	-	0.1445	0.8399	0.8391
2.1583	24600	-	0.1452	0.8416	0.8401
2.1627	24650	-	0.1426	0.8411	0.8385
2.1670	24700	-	0.1447	0.8424	0.8407
2.1714	24750	0.2058	0.1460	0.8422	0.8413
2.1758	24800	-	0.1434	0.8422	0.8418
2.1802	24850	-	0.1443	0.8438	0.8416
2.1846	24900	-	0.1414	0.8422	0.8405
2.1890	24950	-	0.1437	0.8424	0.8407
2.1934	25000	0.2111	0.1466	0.8401	0.8394
2.1978	25050	-	0.1437	0.8390	0.8377
2.2021	25100	-	0.1446	0.8402	0.8394
2.2065	25150	-	0.1457	0.8394	0.8380
2.2109	25200	-	0.1432	0.8406	0.8380
2.2153	25250	0.2013	0.1464	0.8412	0.8397
2.2197	25300	-	0.1499	0.8419	0.8388
2.2241	25350	-	0.1466	0.8425	0.8402
2.2285	25400	-	0.1429	0.8424	0.8397
2.2328	25450	-	0.1433	0.8430	0.8404
2.2372	25500	0.2064	0.1472	0.8410	0.8404
2.2416	25550	-	0.1451	0.8406	0.8386
2.2460	25600	-	0.1480	0.8427	0.8419
2.2504	25650	-	0.1507	0.8409	0.8412
2.2548	25700	-	0.1488	0.8407	0.8398
2.2592	25750	0.2084	0.1476	0.8401	0.8392
2.2636	25800	-	0.1478	0.8403	0.8388
2.2679	25850	-	0.1509	0.8420	0.8417
2.2723	25900	-	0.1464	0.8417	0.8396
2.2767	25950	-	0.1469	0.8406	0.8388
2.2811	26000	0.2113	0.1470	0.8422	0.8404
2.2855	26050	-	0.1479	0.8414	0.8411
2.2899	26100	-	0.1488	0.8424	0.8418
2.2943	26150	-	0.1508	0.8429	0.8428
2.2986	26200	-	0.1507	0.8425	0.8422
2.3030	26250	0.2045	0.1496	0.8423	0.8416

Framework Versions

• Python: 3.10.14
• Sentence Transformers: 3.2.0
• Transformers: 4.45.2
• PyTorch: 2.3.1
• Accelerate: 1.0.1
• Datasets: 3.0.1
• Tokenizers: 0.20.1

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}
}