--- language: - en license: apache-2.0 tags: - sentence-transformers - sentence-similarity - feature-extraction - generated_from_trainer - dataset_size:6300 - loss:MatryoshkaLoss - loss:MultipleNegativesRankingLoss base_model: BAAI/bge-base-en-v1.5 widget: - source_sentence: As of the end of 2023, Hilton's development pipeline included projects in 118 countries and territories. sentences: - What was the total net income attributed to AT&T common stockholders in 2023? - How many countries and territories did Hilton's development pipeline encompass as of the end of 2023? - What caused the increase in Medicare receivables in 2023 compared to 2022? - source_sentence: Alex G. Balazs was appointed as the Executive Vice President and Chief Technology Officer effective September 5, 2023. sentences: - What page of IBM's Form 10-K contains the Financial Statement Schedule? - When was Alex G. Balazs appointed as the Executive Vice President and Chief Technology Officer? - How much were the valuation allowances provided for deferred tax assets related to loss carryforwards as of December 31, 2023? - source_sentence: 'HP''s global wellness program emphasizes five pillars of wellness: physical, financial, emotional, life balance, and social/community.' sentences: - What are the five pillars of wellness emphasized in HP's global wellness program? - What was the fair value of money market mutual funds measured at as of January 31, 2023 and how was it categorized in the fair value hierarchy? - What amount was authorized for future share repurchases by the company as of October 31, 2023? - source_sentence: Item 3, titled 'Legal Proceedings' in a 10-K filing, directs to Note 16 where specific information is further detailed in Item 8 of Part II. sentences: - What was the grant date fair value of options vested for HP in fiscal years 2023, 2022, and 2021? - What is the balance at the end of the year for Comcast's Total Equity in 2023? - What is indicated by Item 3, 'Legal Proceedings', in a 10-K filing? - source_sentence: During 2023, we received approximately $220 of cash collateral, on a net basis. sentences: - How much cash collateral did AT&T receive on a net basis during 2023? - What percentage of FedEx Corporation's consolidated revenues did jet fuel costs represent in 2023? - What measures has Bank of America taken to streamline its organizational structure? 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@100 model-index: - name: BGE base Financial Matryoshka results: - task: type: information-retrieval name: Information Retrieval dataset: name: dim 768 type: dim_768 metrics: - type: cosine_accuracy@1 value: 0.7128571428571429 name: Cosine Accuracy@1 - type: cosine_accuracy@3 value: 0.8428571428571429 name: Cosine Accuracy@3 - type: cosine_accuracy@5 value: 0.8842857142857142 name: Cosine Accuracy@5 - type: cosine_accuracy@10 value: 0.92 name: Cosine Accuracy@10 - type: cosine_precision@1 value: 0.7128571428571429 name: Cosine Precision@1 - type: cosine_precision@3 value: 0.28095238095238095 name: Cosine Precision@3 - type: cosine_precision@5 value: 0.17685714285714288 name: Cosine Precision@5 - type: cosine_precision@10 value: 0.09199999999999998 name: Cosine Precision@10 - type: cosine_recall@1 value: 0.7128571428571429 name: Cosine Recall@1 - type: cosine_recall@3 value: 0.8428571428571429 name: Cosine Recall@3 - type: cosine_recall@5 value: 0.8842857142857142 name: Cosine Recall@5 - type: cosine_recall@10 value: 0.92 name: Cosine Recall@10 - type: cosine_ndcg@10 value: 0.8195233962517928 name: Cosine Ndcg@10 - type: cosine_mrr@10 value: 0.7870022675736963 name: Cosine Mrr@10 - type: cosine_map@100 value: 0.7905145024165581 name: Cosine Map@100 - task: type: information-retrieval name: Information Retrieval dataset: name: dim 512 type: dim_512 metrics: - type: cosine_accuracy@1 value: 0.7157142857142857 name: Cosine Accuracy@1 - type: cosine_accuracy@3 value: 0.8457142857142858 name: Cosine Accuracy@3 - type: cosine_accuracy@5 value: 0.8814285714285715 name: Cosine Accuracy@5 - type: cosine_accuracy@10 value: 0.9228571428571428 name: Cosine Accuracy@10 - type: cosine_precision@1 value: 0.7157142857142857 name: Cosine Precision@1 - type: cosine_precision@3 value: 0.2819047619047619 name: Cosine Precision@3 - type: cosine_precision@5 value: 0.1762857142857143 name: Cosine Precision@5 - type: cosine_precision@10 value: 0.09228571428571428 name: Cosine Precision@10 - type: cosine_recall@1 value: 0.7157142857142857 name: Cosine Recall@1 - type: cosine_recall@3 value: 0.8457142857142858 name: Cosine Recall@3 - type: cosine_recall@5 value: 0.8814285714285715 name: Cosine Recall@5 - type: cosine_recall@10 value: 0.9228571428571428 name: Cosine Recall@10 - type: cosine_ndcg@10 value: 0.821183673183428 name: Cosine Ndcg@10 - type: cosine_mrr@10 value: 0.7884829931972789 name: Cosine Mrr@10 - type: cosine_map@100 value: 0.7916656681436871 name: Cosine Map@100 - task: type: information-retrieval name: Information Retrieval dataset: name: dim 256 type: dim_256 metrics: - type: cosine_accuracy@1 value: 0.7114285714285714 name: Cosine Accuracy@1 - type: cosine_accuracy@3 value: 0.8414285714285714 name: Cosine Accuracy@3 - type: cosine_accuracy@5 value: 0.8842857142857142 name: Cosine Accuracy@5 - type: cosine_accuracy@10 value: 0.9157142857142857 name: Cosine Accuracy@10 - type: cosine_precision@1 value: 0.7114285714285714 name: Cosine Precision@1 - type: cosine_precision@3 value: 0.28047619047619043 name: Cosine Precision@3 - type: cosine_precision@5 value: 0.17685714285714285 name: Cosine Precision@5 - type: cosine_precision@10 value: 0.09157142857142858 name: Cosine Precision@10 - type: cosine_recall@1 value: 0.7114285714285714 name: Cosine Recall@1 - type: cosine_recall@3 value: 0.8414285714285714 name: Cosine Recall@3 - type: cosine_recall@5 value: 0.8842857142857142 name: Cosine Recall@5 - type: cosine_recall@10 value: 0.9157142857142857 name: Cosine Recall@10 - type: cosine_ndcg@10 value: 0.8157881706696753 name: Cosine Ndcg@10 - type: cosine_mrr@10 value: 0.7834812925170066 name: Cosine Mrr@10 - type: cosine_map@100 value: 0.7870779881453726 name: Cosine Map@100 - task: type: information-retrieval name: Information Retrieval dataset: name: dim 128 type: dim_128 metrics: - type: cosine_accuracy@1 value: 0.6957142857142857 name: Cosine Accuracy@1 - type: cosine_accuracy@3 value: 0.82 name: Cosine Accuracy@3 - type: cosine_accuracy@5 value: 0.8685714285714285 name: Cosine Accuracy@5 - type: cosine_accuracy@10 value: 0.9057142857142857 name: Cosine Accuracy@10 - type: cosine_precision@1 value: 0.6957142857142857 name: Cosine Precision@1 - type: cosine_precision@3 value: 0.2733333333333333 name: Cosine Precision@3 - type: cosine_precision@5 value: 0.1737142857142857 name: Cosine Precision@5 - type: cosine_precision@10 value: 0.09057142857142857 name: Cosine Precision@10 - type: cosine_recall@1 value: 0.6957142857142857 name: Cosine Recall@1 - type: cosine_recall@3 value: 0.82 name: Cosine Recall@3 - type: cosine_recall@5 value: 0.8685714285714285 name: Cosine Recall@5 - type: cosine_recall@10 value: 0.9057142857142857 name: Cosine Recall@10 - type: cosine_ndcg@10 value: 0.8018105093606251 name: Cosine Ndcg@10 - type: cosine_mrr@10 value: 0.7683497732426302 name: Cosine Mrr@10 - type: cosine_map@100 value: 0.7722509873826792 name: Cosine Map@100 - task: type: information-retrieval name: Information Retrieval dataset: name: dim 64 type: dim_64 metrics: - type: cosine_accuracy@1 value: 0.6528571428571428 name: Cosine Accuracy@1 - type: cosine_accuracy@3 value: 0.7942857142857143 name: Cosine Accuracy@3 - type: cosine_accuracy@5 value: 0.8314285714285714 name: Cosine Accuracy@5 - type: cosine_accuracy@10 value: 0.8757142857142857 name: Cosine Accuracy@10 - type: cosine_precision@1 value: 0.6528571428571428 name: Cosine Precision@1 - type: cosine_precision@3 value: 0.26476190476190475 name: Cosine Precision@3 - type: cosine_precision@5 value: 0.1662857142857143 name: Cosine Precision@5 - type: cosine_precision@10 value: 0.08757142857142856 name: Cosine Precision@10 - type: cosine_recall@1 value: 0.6528571428571428 name: Cosine Recall@1 - type: cosine_recall@3 value: 0.7942857142857143 name: Cosine Recall@3 - type: cosine_recall@5 value: 0.8314285714285714 name: Cosine Recall@5 - type: cosine_recall@10 value: 0.8757142857142857 name: Cosine Recall@10 - type: cosine_ndcg@10 value: 0.7667522193115596 name: Cosine Ndcg@10 - type: cosine_mrr@10 value: 0.7315833333333331 name: Cosine Mrr@10 - type: cosine_map@100 value: 0.7359673420065519 name: Cosine Map@100 --- # BGE base Financial Matryoshka This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [BAAI/bge-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) on the json dataset. 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:** [BAAI/bge-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) - **Maximum Sequence Length:** 512 tokens - **Output Dimensionality:** 768 dimensions - **Similarity Function:** Cosine Similarity - **Training Dataset:** - json - **Language:** en - **License:** apache-2.0 ### 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': 512, 'do_lower_case': True}) with Transformer model: BertModel (1): Pooling({'word_embedding_dimension': 768, '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: ```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("sud-962081/bge-base-financial-matryoshka") # Run inference sentences = [ 'During 2023, we received approximately $220 of cash collateral, on a net basis.', 'How much cash collateral did AT&T receive on a net basis during 2023?', "What percentage of FedEx Corporation's consolidated revenues did jet fuel costs represent in 2023?", ] 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 #### Information Retrieval * Datasets: `dim_768`, `dim_512`, `dim_256`, `dim_128` and `dim_64` * Evaluated with [InformationRetrievalEvaluator](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.InformationRetrievalEvaluator) | Metric | dim_768 | dim_512 | dim_256 | dim_128 | dim_64 | |:--------------------|:-----------|:-----------|:-----------|:-----------|:-----------| | cosine_accuracy@1 | 0.7129 | 0.7157 | 0.7114 | 0.6957 | 0.6529 | | cosine_accuracy@3 | 0.8429 | 0.8457 | 0.8414 | 0.82 | 0.7943 | | cosine_accuracy@5 | 0.8843 | 0.8814 | 0.8843 | 0.8686 | 0.8314 | | cosine_accuracy@10 | 0.92 | 0.9229 | 0.9157 | 0.9057 | 0.8757 | | cosine_precision@1 | 0.7129 | 0.7157 | 0.7114 | 0.6957 | 0.6529 | | cosine_precision@3 | 0.281 | 0.2819 | 0.2805 | 0.2733 | 0.2648 | | cosine_precision@5 | 0.1769 | 0.1763 | 0.1769 | 0.1737 | 0.1663 | | cosine_precision@10 | 0.092 | 0.0923 | 0.0916 | 0.0906 | 0.0876 | | cosine_recall@1 | 0.7129 | 0.7157 | 0.7114 | 0.6957 | 0.6529 | | cosine_recall@3 | 0.8429 | 0.8457 | 0.8414 | 0.82 | 0.7943 | | cosine_recall@5 | 0.8843 | 0.8814 | 0.8843 | 0.8686 | 0.8314 | | cosine_recall@10 | 0.92 | 0.9229 | 0.9157 | 0.9057 | 0.8757 | | **cosine_ndcg@10** | **0.8195** | **0.8212** | **0.8158** | **0.8018** | **0.7668** | | cosine_mrr@10 | 0.787 | 0.7885 | 0.7835 | 0.7683 | 0.7316 | | cosine_map@100 | 0.7905 | 0.7917 | 0.7871 | 0.7723 | 0.736 | ## Training Details ### Training Dataset #### json * Dataset: json * Size: 6,300 training samples * Columns: positive and anchor * Approximate statistics based on the first 1000 samples: | | positive | anchor | |:--------|:-----------------------------------------------------------------------------------|:----------------------------------------------------------------------------------| | type | string | string | | details | | | * Samples: | positive | anchor | |:------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:------------------------------------------------------------------------------------------------------------------| | SmartFlex benefits and the 'Best of Both' work model at The Hershey Company allow employees to balance professional and personal demands through flexible work arrangements. | How does The Hershey Company ensure flexibility and work-life balance for its employees? | | In February 2024, our Board authorized an additional $2.0 billion stock repurchase program, with no expiration from the date of authorization. | What amount was authorized for common stock repurchase by the company's Board in February 2024? | | Beginning in 2025, the first GM EVs will be constructed using the North American Charging Standard (NACS) hardware. | What significant change is set for General Motors' EVs starting in 2025 regarding charging hardware? | * Loss: [MatryoshkaLoss](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#matryoshkaloss) with these parameters: ```json { "loss": "MultipleNegativesRankingLoss", "matryoshka_dims": [ 768, 512, 256, 128, 64 ], "matryoshka_weights": [ 1, 1, 1, 1, 1 ], "n_dims_per_step": -1 } ``` ### Training Hyperparameters #### Non-Default Hyperparameters - `eval_strategy`: epoch - `per_device_train_batch_size`: 32 - `per_device_eval_batch_size`: 16 - `gradient_accumulation_steps`: 16 - `learning_rate`: 2e-05 - `num_train_epochs`: 4 - `lr_scheduler_type`: cosine - `warmup_ratio`: 0.1 - `bf16`: True - `load_best_model_at_end`: True - `optim`: adamw_torch_fused - `batch_sampler`: no_duplicates #### All Hyperparameters
Click to expand - `overwrite_output_dir`: False - `do_predict`: False - `eval_strategy`: epoch - `prediction_loss_only`: True - `per_device_train_batch_size`: 32 - `per_device_eval_batch_size`: 16 - `per_gpu_train_batch_size`: None - `per_gpu_eval_batch_size`: None - `gradient_accumulation_steps`: 16 - `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`: 4 - `max_steps`: -1 - `lr_scheduler_type`: cosine - `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`: 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`: 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`: True - `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_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 - `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 | Epoch | Step | Training Loss | dim_768_cosine_ndcg@10 | dim_512_cosine_ndcg@10 | dim_256_cosine_ndcg@10 | dim_128_cosine_ndcg@10 | dim_64_cosine_ndcg@10 | |:----------:|:------:|:-------------:|:----------------------:|:----------------------:|:----------------------:|:----------------------:|:---------------------:| | 1.0 | 7 | - | 0.8036 | 0.8049 | 0.7942 | 0.7726 | 0.7375 | | 1.4848 | 10 | 2.2028 | - | - | - | - | - | | 2.0 | 14 | - | 0.8169 | 0.8173 | 0.8127 | 0.8000 | 0.7602 | | 2.9697 | 20 | 0.9836 | - | - | - | - | - | | 3.0 | 21 | - | 0.8187 | 0.8214 | 0.8142 | 0.8017 | 0.7658 | | **3.4848** | **24** | **-** | **0.8195** | **0.8212** | **0.8158** | **0.8018** | **0.7668** | * The bold row denotes the saved checkpoint. ### Framework Versions - Python: 3.10.12 - Sentence Transformers: 3.3.1 - Transformers: 4.47.0 - PyTorch: 2.5.1+cu121 - Accelerate: 1.2.1 - Datasets: 3.2.0 - Tokenizers: 0.21.0 ## 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", } ``` #### MatryoshkaLoss ```bibtex @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 ```bibtex @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} } ```