--- language: [] library_name: sentence-transformers tags: - sentence-transformers - sentence-similarity - feature-extraction - generated_from_trainer - dataset_size:700000 - loss:DenoisingAutoEncoderLoss base_model: intfloat/e5-base-unsupervised datasets: [] metrics: - pearson_cosine - spearman_cosine - pearson_manhattan - spearman_manhattan - pearson_euclidean - spearman_euclidean - pearson_dot - spearman_dot - pearson_max - spearman_max widget: - source_sentence: in Freeview no extra therefore minimal Also the is wide decent, plus they and. sentences: - 'Pokémon-GX (Japanese: ポケモンGX Pokémon GX), officially written Pokémon-GX, are a variant of Pokémon in the Pokémon Trading Card Game. They were first introduced in the Sun & Moon expansion (the Collection Sun and Collection Moon expansions in Japan). Pokémon-GX have a stylized. graphic on the card name.' - 'The Cape Colony (Dutch: Kaapkolonie) was a Dutch East India Company colony in Southern Africa, centered on the Cape of Good Hope, whence it derived its name. The original colony and its successive states that the colony was incorporated into occupied much of modern South Africa.' - Avtex is expensive, but you get built in Freeview, Freesat and built in DVD player, which means no extra boxes, and therefore minimal wiring. Also the viewing angle is wide and a decent picture quality, plus they are light and designed for mobile use. - source_sentence: as power Yes can use transmission of power steering But, sure you check the manufacturer's the a sentences: - Can you use transmission fluid as a substitute for power steering fluid? Yes, you can use transmission fluid in place of a power steering fluid. But, make sure you check the car manufacturer's recommendations before using the ATF as a substitute. - how much kwh does an xbox one use? - what is the difference between demerara cane sugar and turbinado cane sugar? - source_sentence: '(number ''Step: Ensure date to (and number is set Date 2 formula to add the number months start.''' sentences: - Being a medical doctor is really great. It's stimulating and interesting. Medical doctors have a significant degree of autonomy over their schedules and time. Medical doctors know that they get to help people solve problems every single day. - how much is an air conditioner for a house? - '[''=EDATE(start date, number of months)'', ''Step 1: Ensure the starting date is properly formatted – go to Format Cells (press Ctrl + 1) and make sure the number is set to Date.'', ''Step 2: Use the =EDATE(C3,C5) formula to add the number of specified months to the start date.'']' - source_sentence: many days can sentences: - how many days after can you have morning after pill? - is gender an independent variable? - The current standard is about 30 days, which means that some teachers and support staff may be brought on board before the results of their criminal background check are completed. The issue, as reported in this article, is the lag time between state and federal background checks. - source_sentence: ligand ion channels located? sentences: - Share on Pinterest Recent research suggests that chocolate may have some health benefits. Chocolate receives a lot of bad press because of its high fat and sugar content. Its consumption has been associated with acne, obesity, high blood pressure, coronary artery disease, and diabetes. - where are ligand gated ion channels located? - Duvets tend to be warm but surprisingly lightweight. The duvet cover makes it easier to change bedding looks and styles. You won't need to wash your duvet very often, just wash the cover regularly. Additionally, duvets tend to be fluffier than comforters, and can simplify bed making if you choose the European style. pipeline_tag: sentence-similarity model-index: - name: SentenceTransformer based on intfloat/e5-base-unsupervised results: - task: type: semantic-similarity name: Semantic Similarity dataset: name: sts test type: sts-test metrics: - type: pearson_cosine value: 0.7651793859211248 name: Pearson Cosine - type: spearman_cosine value: 0.7524804428249002 name: Spearman Cosine - type: pearson_manhattan value: 0.7393361318996702 name: Pearson Manhattan - type: spearman_manhattan value: 0.7326262473219208 name: Spearman Manhattan - type: pearson_euclidean value: 0.7402295162714656 name: Pearson Euclidean - type: spearman_euclidean value: 0.7335305408258518 name: Spearman Euclidean - type: pearson_dot value: 0.5002878735642248 name: Pearson Dot - type: spearman_dot value: 0.4986010870846151 name: Spearman Dot - type: pearson_max value: 0.7651793859211248 name: Pearson Max - type: spearman_max value: 0.7524804428249002 name: Spearman Max --- # SentenceTransformer based on intfloat/e5-base-unsupervised This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [intfloat/e5-base-unsupervised](https://huggingface.co/intfloat/e5-base-unsupervised). 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:** [intfloat/e5-base-unsupervised](https://huggingface.co/intfloat/e5-base-unsupervised) - **Maximum Sequence Length:** 512 tokens - **Output Dimensionality:** 768 tokens - **Similarity Function:** Cosine Similarity ### 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': 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: ```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("bobox/E5-base-unsupervised-TSDAE-2") # Run inference sentences = [ 'ligand ion channels located?', 'where are ligand gated ion channels located?', "Duvets tend to be warm but surprisingly lightweight. The duvet cover makes it easier to change bedding looks and styles. You won't need to wash your duvet very often, just wash the cover regularly. Additionally, duvets tend to be fluffier than comforters, and can simplify bed making if you choose the European style.", ] 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-test` * Evaluated with [EmbeddingSimilarityEvaluator](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.EmbeddingSimilarityEvaluator) | Metric | Value | |:--------------------|:-----------| | pearson_cosine | 0.7652 | | **spearman_cosine** | **0.7525** | | pearson_manhattan | 0.7393 | | spearman_manhattan | 0.7326 | | pearson_euclidean | 0.7402 | | spearman_euclidean | 0.7335 | | pearson_dot | 0.5003 | | spearman_dot | 0.4986 | | pearson_max | 0.7652 | | spearman_max | 0.7525 | ## Training Details ### Training Dataset #### Unnamed Dataset * Size: 700,000 training samples * Columns: sentence_0 and sentence_1 * Approximate statistics based on the first 1000 samples: | | sentence_0 | sentence_1 | |:--------|:----------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------| | type | string | string | | details | | | * Samples: | sentence_0 | sentence_1 | |:---------------------------------------------------------------------------------------------------------------------|:----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------| | Quality such a has components with applicable high objective system measure component improvements | Quality in such a system has three components: high accuracy, compliance with applicable standards, and high customer satisfaction. The objective of the system is to measure each component and achieve improvements. | | include | does qbi include capital gains? | | They have a . parietal is in, as becomes and pigments after four to is believed and in circadian cycles | They have a third eye. The parietal eye is only visible in hatchlings, as it becomes covered in scales and pigments after four to six months. Its function is a subject of ongoing research, but it is believed to be useful in absorbing ultraviolet rays and in setting circadian and seasonal cycles. | * Loss: [DenoisingAutoEncoderLoss](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#denoisingautoencoderloss) ### Training Hyperparameters #### Non-Default Hyperparameters - `eval_strategy`: steps - `per_device_train_batch_size`: 16 - `per_device_eval_batch_size`: 16 - `num_train_epochs`: 2 - `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`: 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 - `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`: 2 - `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`: 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 - `batch_sampler`: batch_sampler - `multi_dataset_batch_sampler`: round_robin
### Training Logs
Click to expand | Epoch | Step | Training Loss | sts-test_spearman_cosine | |:------:|:-----:|:-------------:|:------------------------:| | 0 | 0 | - | 0.7211 | | 0.0114 | 500 | 9.4957 | - | | 0.0229 | 1000 | 7.4063 | - | | 0.0343 | 1500 | 7.0225 | - | | 0.0457 | 2000 | 6.6991 | - | | 0.0571 | 2500 | 6.4054 | - | | 0.0686 | 3000 | 6.1933 | - | | 0.08 | 3500 | 5.999 | - | | 0.0914 | 4000 | 5.8471 | - | | 0.1 | 4375 | - | 0.4610 | | 0.1029 | 4500 | 5.6876 | - | | 0.1143 | 5000 | 5.5934 | - | | 0.1257 | 5500 | 5.4877 | - | | 0.1371 | 6000 | 5.4034 | - | | 0.1486 | 6500 | 5.3016 | - | | 0.16 | 7000 | 5.2169 | - | | 0.1714 | 7500 | 5.1351 | - | | 0.1829 | 8000 | 5.0605 | - | | 0.1943 | 8500 | 4.9851 | - | | 0.2 | 8750 | - | 0.6490 | | 0.2057 | 9000 | 4.9024 | - | | 0.2171 | 9500 | 4.8722 | - | | 0.2286 | 10000 | 4.7955 | - | | 0.24 | 10500 | 4.7435 | - | | 0.2514 | 11000 | 4.6742 | - | | 0.2629 | 11500 | 4.6447 | - | | 0.2743 | 12000 | 4.5964 | - | | 0.2857 | 12500 | 4.5186 | - | | 0.2971 | 13000 | 4.5024 | - | | 0.3 | 13125 | - | 0.7121 | | 0.3086 | 13500 | 4.4336 | - | | 0.32 | 14000 | 4.3767 | - | | 0.3314 | 14500 | 4.3454 | - | | 0.3429 | 15000 | 4.3067 | - | | 0.3543 | 15500 | 4.2627 | - | | 0.3657 | 16000 | 4.2323 | - | | 0.3771 | 16500 | 4.208 | - | | 0.3886 | 17000 | 4.1622 | - | | 0.4 | 17500 | 4.113 | 0.7375 | | 0.4114 | 18000 | 4.1097 | - | | 0.4229 | 18500 | 4.0666 | - | | 0.4343 | 19000 | 4.0311 | - | | 0.4457 | 19500 | 4.0241 | - | | 0.4571 | 20000 | 3.9991 | - | | 0.4686 | 20500 | 3.9873 | - | | 0.48 | 21000 | 3.9439 | - | | 0.4914 | 21500 | 3.9281 | - | | 0.5 | 21875 | - | 0.7502 | | 0.5029 | 22000 | 3.9047 | - | | 0.5143 | 22500 | 3.89 | - | | 0.5257 | 23000 | 3.8671 | - | | 0.5371 | 23500 | 3.85 | - | | 0.5486 | 24000 | 3.8336 | - | | 0.56 | 24500 | 3.8081 | - | | 0.5714 | 25000 | 3.8049 | - | | 0.5829 | 25500 | 3.7587 | - | | 0.5943 | 26000 | 3.769 | - | | 0.6 | 26250 | - | 0.7530 | | 0.6057 | 26500 | 3.7488 | - | | 0.6171 | 27000 | 3.7218 | - | | 0.6286 | 27500 | 3.7128 | - | | 0.64 | 28000 | 3.7104 | - | | 0.6514 | 28500 | 3.6706 | - | | 0.6629 | 29000 | 3.6602 | - | | 0.6743 | 29500 | 3.658 | - | | 0.6857 | 30000 | 3.665 | - | | 0.6971 | 30500 | 3.6439 | - | | 0.7 | 30625 | - | 0.7561 | | 0.7086 | 31000 | 3.6411 | - | | 0.72 | 31500 | 3.6141 | - | | 0.7314 | 32000 | 3.6172 | - | | 0.7429 | 32500 | 3.5975 | - | | 0.7543 | 33000 | 3.5827 | - | | 0.7657 | 33500 | 3.5836 | - | | 0.7771 | 34000 | 3.5484 | - | | 0.7886 | 34500 | 3.5275 | - | | 0.8 | 35000 | 3.5587 | 0.7553 | | 0.8114 | 35500 | 3.5371 | - | | 0.8229 | 36000 | 3.5334 | - | | 0.8343 | 36500 | 3.5168 | - | | 0.8457 | 37000 | 3.483 | - | | 0.8571 | 37500 | 3.4755 | - | | 0.8686 | 38000 | 3.4943 | - | | 0.88 | 38500 | 3.4699 | - | | 0.8914 | 39000 | 3.4732 | - | | 0.9 | 39375 | - | 0.7560 | | 0.9029 | 39500 | 3.4572 | - | | 0.9143 | 40000 | 3.4518 | - | | 0.9257 | 40500 | 3.4298 | - | | 0.9371 | 41000 | 3.4215 | - | | 0.9486 | 41500 | 3.4176 | - | | 0.96 | 42000 | 3.4353 | - | | 0.9714 | 42500 | 3.4137 | - | | 0.9829 | 43000 | 3.4037 | - | | 0.9943 | 43500 | 3.4157 | - | | 1.0 | 43750 | - | 0.7554 | | 1.0057 | 44000 | 3.393 | - | | 1.0171 | 44500 | 3.4092 | - | | 1.0286 | 45000 | 3.3861 | - | | 1.04 | 45500 | 3.3976 | - | | 1.0514 | 46000 | 3.3769 | - | | 1.0629 | 46500 | 3.3444 | - | | 1.0743 | 47000 | 3.3598 | - | | 1.0857 | 47500 | 3.3556 | - | | 1.0971 | 48000 | 3.3548 | - | | 1.1 | 48125 | - | 0.7549 | | 1.1086 | 48500 | 3.3278 | - | | 1.12 | 49000 | 3.3309 | - | | 1.1314 | 49500 | 3.3459 | - | | 1.1429 | 50000 | 3.3353 | - | | 1.1543 | 50500 | 3.3192 | - | | 1.1657 | 51000 | 3.3022 | - | | 1.1771 | 51500 | 3.3189 | - | | 1.1886 | 52000 | 3.301 | - | | 1.2 | 52500 | 3.2785 | 0.7542 | | 1.2114 | 53000 | 3.2996 | - | | 1.2229 | 53500 | 3.2863 | - | | 1.2343 | 54000 | 3.2916 | - | | 1.2457 | 54500 | 3.272 | - | | 1.2571 | 55000 | 3.2896 | - | | 1.2686 | 55500 | 3.2694 | - | | 1.28 | 56000 | 3.2848 | - | | 1.2914 | 56500 | 3.2528 | - | | 1.3 | 56875 | - | 0.7554 | | 1.3029 | 57000 | 3.2622 | - | | 1.3143 | 57500 | 3.2515 | - | | 1.3257 | 58000 | 3.2385 | - | | 1.3371 | 58500 | 3.2341 | - | | 1.3486 | 59000 | 3.2275 | - | | 1.3600 | 59500 | 3.2538 | - | | 1.3714 | 60000 | 3.2329 | - | | 1.3829 | 60500 | 3.2322 | - | | 1.3943 | 61000 | 3.2039 | - | | 1.4 | 61250 | - | 0.7530 | | 1.4057 | 61500 | 3.212 | - | | 1.4171 | 62000 | 3.2127 | - | | 1.4286 | 62500 | 3.1956 | - | | 1.44 | 63000 | 3.202 | - | | 1.4514 | 63500 | 3.2046 | - | | 1.4629 | 64000 | 3.2105 | - | | 1.4743 | 64500 | 3.1915 | - | | 1.4857 | 65000 | 3.176 | - | | 1.4971 | 65500 | 3.1852 | - | | 1.5 | 65625 | - | 0.7541 | | 1.5086 | 66000 | 3.1988 | - | | 1.52 | 66500 | 3.1714 | - | | 1.5314 | 67000 | 3.1816 | - | | 1.5429 | 67500 | 3.1745 | - | | 1.5543 | 68000 | 3.1674 | - | | 1.5657 | 68500 | 3.1887 | - | | 1.5771 | 69000 | 3.1567 | - | | 1.5886 | 69500 | 3.1775 | - | | 1.6 | 70000 | 3.1696 | 0.7535 | | 1.6114 | 70500 | 3.154 | - | | 1.6229 | 71000 | 3.1553 | - | | 1.6343 | 71500 | 3.1675 | - | | 1.6457 | 72000 | 3.1516 | - | | 1.6571 | 72500 | 3.1569 | - | | 1.6686 | 73000 | 3.1403 | - | | 1.6800 | 73500 | 3.1667 | - | | 1.6914 | 74000 | 3.1545 | - | | 1.7 | 74375 | - | 0.7529 | | 1.7029 | 74500 | 3.1736 | - | | 1.7143 | 75000 | 3.1447 | - | | 1.7257 | 75500 | 3.1567 | - | | 1.7371 | 76000 | 3.1682 | - | | 1.7486 | 76500 | 3.149 | - | | 1.76 | 77000 | 3.1522 | - | | 1.7714 | 77500 | 3.1412 | - | | 1.7829 | 78000 | 3.1268 | - | | 1.7943 | 78500 | 3.1476 | - | | 1.8 | 78750 | - | 0.7524 | | 1.8057 | 79000 | 3.1669 | - | | 1.8171 | 79500 | 3.1432 | - | | 1.8286 | 80000 | 3.1603 | - | | 1.8400 | 80500 | 3.1347 | - | | 1.8514 | 81000 | 3.1209 | - | | 1.8629 | 81500 | 3.1302 | - | | 1.8743 | 82000 | 3.1423 | - | | 1.8857 | 82500 | 3.1481 | - | | 1.8971 | 83000 | 3.1262 | - | | 1.9 | 83125 | - | 0.7525 | | 1.9086 | 83500 | 3.1484 | - | | 1.92 | 84000 | 3.1331 | - | | 1.9314 | 84500 | 3.122 | - | | 1.9429 | 85000 | 3.1272 | - | | 1.9543 | 85500 | 3.1435 | - | | 1.9657 | 86000 | 3.1431 | - | | 1.9771 | 86500 | 3.1457 | - | | 1.9886 | 87000 | 3.1286 | - | | 2.0 | 87500 | 3.1352 | 0.7525 |
### Framework Versions - Python: 3.10.13 - Sentence Transformers: 3.0.1 - Transformers: 4.41.2 - PyTorch: 2.1.2 - Accelerate: 0.31.0 - Datasets: 2.19.2 - Tokenizers: 0.19.1 ## 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", } ``` #### DenoisingAutoEncoderLoss ```bibtex @inproceedings{wang-2021-TSDAE, title = "TSDAE: Using Transformer-based Sequential Denoising Auto-Encoderfor Unsupervised Sentence Embedding Learning", author = "Wang, Kexin and Reimers, Nils and Gurevych, Iryna", booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2021", month = nov, year = "2021", address = "Punta Cana, Dominican Republic", publisher = "Association for Computational Linguistics", pages = "671--688", url = "https://arxiv.org/abs/2104.06979", } ```