--- language: - multilingual license: mit tags: - sentence-transformers - sentence-similarity - feature-extraction - dense - generated_from_trainer - dataset_size:74864 - loss:CoSENTLoss base_model: intfloat/multilingual-e5-small widget: - source_sentence: Légumes mijotés Jardinière et haricots blancs sentences: - AMSCAN GOLD PLSTC FORKS | PARTY SUPPLY | 240 CT. - 辣椒酱 - Pizza de verduras brasadas - source_sentence: VTech Crazy Legs Learning Bugs, Pink sentences: - LEGO Creator Expert Garagem de Canto 10264 Kit de Construção, Novo 2019 (2569 Peças), Embalagem Sem Frustrações - Silver Glitter Hanging Fans (4 ct) - VTech Aspirateur Pop et Compte - source_sentence: Pacon Tru-Ray Construction Paper, 18-Inches by 24-Inches, 50-Count, Red (103094) sentences: - Funko POP Televisione Westworld Bernard Lowe Action figure - Carta da costruzione Tru-Ray pesante, colori assortiti caldi, 12" x 18", 50 fogli - Max Factory Kizuna Ai Figma Action Figure - source_sentence: Zesty Cilantro Salsa, Medium sentences: - Melange de fruits - Salsa de Texas - T.S. Shure Rubber Band Powered Rescue Flier Model Plane Kit - source_sentence: Fun World Angelic Maiden Child Costume sentences: - Melissa & Doug Personalized Pattern Blocks & Boards Classic Toy - Winter sprats gerookt - Rubie's Costume Co - Girls Gypsy Costume pipeline_tag: sentence-similarity library_name: sentence-transformers metrics: - cosine_accuracy@1 - cosine_accuracy@3 - cosine_accuracy@5 - cosine_accuracy@10 - cosine_precision@1 - cosine_precision@3 - cosine_precision@5 - cosine_precision@10 - cosine_recall@1 - cosine_recall@3 - cosine_recall@5 - cosine_recall@10 - cosine_ndcg@10 - cosine_mrr@10 - cosine_map@1 - cosine_map@3 - cosine_map@5 - cosine_map@10 model-index: - name: multilingual-e5-small embeddings (CoSENTLoss on graded listwise pairs) results: - task: type: information-retrieval name: Information Retrieval dataset: name: ir eval type: ir_eval metrics: - type: cosine_accuracy@1 value: 0.91015625 name: Cosine Accuracy@1 - type: cosine_accuracy@3 value: 0.95703125 name: Cosine Accuracy@3 - type: cosine_accuracy@5 value: 0.97265625 name: Cosine Accuracy@5 - type: cosine_accuracy@10 value: 1.0 name: Cosine Accuracy@10 - type: cosine_precision@1 value: 0.91015625 name: Cosine Precision@1 - type: cosine_precision@3 value: 0.5104166666666666 name: Cosine Precision@3 - type: cosine_precision@5 value: 0.40078125000000003 name: Cosine Precision@5 - type: cosine_precision@10 value: 0.296875 name: Cosine Precision@10 - type: cosine_recall@1 value: 0.13477527216379598 name: Cosine Recall@1 - type: cosine_recall@3 value: 0.1739842681808551 name: Cosine Recall@3 - type: cosine_recall@5 value: 0.1983227020362507 name: Cosine Recall@5 - type: cosine_recall@10 value: 0.2486998357621607 name: Cosine Recall@10 - type: cosine_ndcg@10 value: 0.4650339807377877 name: Cosine Ndcg@10 - type: cosine_mrr@10 value: 0.937943328373016 name: Cosine Mrr@10 - type: cosine_map@1 value: 0.91015625 name: Cosine Map@1 - type: cosine_map@3 value: 0.5282118055555556 name: Cosine Map@3 - type: cosine_map@5 value: 0.42098524305555557 name: Cosine Map@5 - type: cosine_map@10 value: 0.3311448220781368 name: Cosine Map@10 --- # multilingual-e5-small embeddings (CoSENTLoss on graded listwise pairs) This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [intfloat/multilingual-e5-small](https://huggingface.co/intfloat/multilingual-e5-small). 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:** [intfloat/multilingual-e5-small](https://huggingface.co/intfloat/multilingual-e5-small) - **Maximum Sequence Length:** 256 tokens - **Output Dimensionality:** 384 dimensions - **Similarity Function:** Cosine Similarity - **Language:** multilingual - **License:** mit ### Model Sources - **Documentation:** [Sentence Transformers Documentation](https://sbert.net) - **Repository:** [Sentence Transformers on GitHub](https://github.com/UKPLab/sentence-transformers) - **Hugging Face:** [Sentence Transformers on Hugging Face](https://huggingface.co/models?library=sentence-transformers) ### 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: ```bash 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("Antix5/product-embed-multi-e5-small") # Run inference sentences = [ 'Fun World Angelic Maiden Child Costume', "Rubie's Costume Co - Girls Gypsy Costume", 'Melissa & Doug Personalized Pattern Blocks & Boards Classic Toy', ] 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, 0.7135, 0.6875], # [0.7135, 1.0000, 0.6791], # [0.6875, 0.6791, 1.0000]]) ``` ## Evaluation ### Metrics #### Information Retrieval * Dataset: `ir_eval` * Evaluated with [InformationRetrievalEvaluator](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.InformationRetrievalEvaluator) | Metric | Value | |:--------------------|:----------| | cosine_accuracy@1 | 0.9102 | | cosine_accuracy@3 | 0.957 | | cosine_accuracy@5 | 0.9727 | | cosine_accuracy@10 | 1.0 | | cosine_precision@1 | 0.9102 | | cosine_precision@3 | 0.5104 | | cosine_precision@5 | 0.4008 | | cosine_precision@10 | 0.2969 | | cosine_recall@1 | 0.1348 | | cosine_recall@3 | 0.174 | | cosine_recall@5 | 0.1983 | | cosine_recall@10 | 0.2487 | | **cosine_ndcg@10** | **0.465** | | cosine_mrr@10 | 0.9379 | | cosine_map@1 | 0.9102 | | cosine_map@3 | 0.5282 | | cosine_map@5 | 0.421 | | cosine_map@10 | 0.3311 | ## Training Details ### Training Dataset #### Unnamed Dataset * Size: 74,864 training samples * Columns: text1, text2, and label * Approximate statistics based on the first 1000 samples: | | text1 | text2 | label | |:--------|:----------------------------------------------------------------------------------|:----------------------------------------------------------------------------------|:---------------------------------------------------------------| | type | string | string | float | | details | | | | * Samples: | text1 | text2 | label | |:-----------------------------------------------------------------|:------------------------------------------------------------------------------|:-----------------| | Premier 26764 Car Spinner, Santa, 25 by 19-1/2-Inch | Premier 26764 Tourbillon pour voiture, Santa, 25 x 19-1/2 pouces | 1.0 | | Premier 26764 Car Spinner, Santa, 25 by 19-1/2-Inch | BNTS, ЧИПСЫ ИЗ ФАСОЛИ NV И МОРСКАЯ СОЛЬ | 0.0 | | Premier 26764 Car Spinner, Santa, 25 by 19-1/2-Inch | Beanitos, Чипс из фасоли navy, Сыр на чо | 0.0 | * Loss: [CoSENTLoss](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#cosentloss) with these parameters: ```json { "scale": 20.0, "similarity_fct": "pairwise_cos_sim" } ``` ### Training Hyperparameters #### Non-Default Hyperparameters - `eval_strategy`: steps - `per_device_train_batch_size`: 32 - `per_device_eval_batch_size`: 256 - `learning_rate`: 2e-05 - `num_train_epochs`: 2 - `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`: 32 - `per_device_eval_batch_size`: 256 - `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`: 2e-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`: 2 - `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} - `parallelism_config`: None - `deepspeed`: None - `label_smoothing_factor`: 0.0 - `optim`: adamw_torch_fused - `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`: no_duplicates - `multi_dataset_batch_sampler`: proportional - `router_mapping`: {} - `learning_rate_mapping`: {}
### Training Logs | Epoch | Step | Training Loss | ir_eval_cosine_ndcg@10 | |:------:|:----:|:-------------:|:----------------------:| | 0.0004 | 1 | 5.9178 | - | | 0.0427 | 100 | 5.7854 | - | | 0.0855 | 200 | 5.7118 | - | | 0.1282 | 300 | 5.6765 | - | | 0.1709 | 400 | 5.647 | - | | 0.2137 | 500 | 5.6046 | - | | 0.2564 | 600 | 5.5859 | - | | 0.2991 | 700 | 5.5586 | - | | 0.3419 | 800 | 5.5319 | - | | 0.3846 | 900 | 5.564 | - | | 0.4274 | 1000 | 5.577 | 0.4854 | | 0.4701 | 1100 | 5.5229 | - | | 0.5128 | 1200 | 5.5294 | - | | 0.5556 | 1300 | 5.4836 | - | | 0.5983 | 1400 | 5.4851 | - | | 0.6410 | 1500 | 5.4646 | - | | 0.6838 | 1600 | 5.4784 | - | | 0.7265 | 1700 | 5.481 | - | | 0.7692 | 1800 | 5.4923 | - | | 0.8120 | 1900 | 5.4696 | - | | 0.8547 | 2000 | 5.4932 | 0.4749 | | 0.8974 | 2100 | 5.4752 | - | | 0.9402 | 2200 | 5.459 | - | | 0.9829 | 2300 | 5.4371 | - | | 1.0256 | 2400 | 5.3701 | - | | 1.0684 | 2500 | 5.3562 | - | | 1.1111 | 2600 | 5.4101 | - | | 1.1538 | 2700 | 5.3829 | - | | 1.1966 | 2800 | 5.3687 | - | | 1.2393 | 2900 | 5.36 | - | | 1.2821 | 3000 | 5.3446 | 0.4725 | | 1.3248 | 3100 | 5.3757 | - | | 1.3675 | 3200 | 5.3821 | - | | 1.4103 | 3300 | 5.3918 | - | | 1.4530 | 3400 | 5.3083 | - | | 1.4957 | 3500 | 5.3389 | - | | 1.5385 | 3600 | 5.3037 | - | | 1.5812 | 3700 | 5.3424 | - | | 1.6239 | 3800 | 5.3383 | - | | 1.6667 | 3900 | 5.3252 | - | | 1.7094 | 4000 | 5.3358 | 0.4676 | | 1.7521 | 4100 | 5.2704 | - | | 1.7949 | 4200 | 5.3415 | - | | 1.8376 | 4300 | 5.361 | - | | 1.8803 | 4400 | 5.3654 | - | | 1.9231 | 4500 | 5.3386 | - | | 1.9658 | 4600 | 5.3392 | - | | -1 | -1 | - | 0.4650 | ### Framework Versions - Python: 3.12.11 - Sentence Transformers: 5.1.1 - Transformers: 4.56.2 - PyTorch: 2.8.0+cu126 - Accelerate: 1.10.1 - Datasets: 2.20.0 - Tokenizers: 0.22.1 ## 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", } ``` #### CoSENTLoss ```bibtex @article{10531646, author={Huang, Xiang and Peng, Hao and Zou, Dongcheng and Liu, Zhiwei and Li, Jianxin and Liu, Kay and Wu, Jia and Su, Jianlin and Yu, Philip S.}, journal={IEEE/ACM Transactions on Audio, Speech, and Language Processing}, title={CoSENT: Consistent Sentence Embedding via Similarity Ranking}, year={2024}, doi={10.1109/TASLP.2024.3402087} } ```