MPNet base trained on GooAQ Question-Answer tuples
This is a sentence-transformers model finetuned from microsoft/mpnet-base on the sentence-transformers/gooaq 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.
This model was trained using the train_script.py code.
Model Details
Model Description
- Model Type: Sentence Transformer
- Base model: microsoft/mpnet-base
- Maximum Sequence Length: 512 tokens
- Output Dimensionality: 768 tokens
- Similarity Function: Cosine Similarity
- Training Dataset:
- Language: en
- License: apache-2.0
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: MPNetModel
(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("tomaarsen/mpnet-base-gooaq")
# Run inference
sentences = [
'11 is what of 8?',
'Convert fraction (ratio) 8 / 11 Answer: 72.727272727273%',
'Old-age pensions are not included in taxable income under the personal income tax.',
]
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
- Dataset:
gooaq-dev
- Evaluated with
InformationRetrievalEvaluator
Metric | Value |
---|---|
cosine_accuracy@1 | 0.7198 |
cosine_accuracy@3 | 0.884 |
cosine_accuracy@5 | 0.9305 |
cosine_accuracy@10 | 0.9709 |
cosine_precision@1 | 0.7198 |
cosine_precision@3 | 0.2947 |
cosine_precision@5 | 0.1861 |
cosine_precision@10 | 0.0971 |
cosine_recall@1 | 0.7198 |
cosine_recall@3 | 0.884 |
cosine_recall@5 | 0.9305 |
cosine_recall@10 | 0.9709 |
cosine_ndcg@10 | 0.8491 |
cosine_mrr@10 | 0.8096 |
cosine_map@100 | 0.8111 |
dot_accuracy@1 | 0.7073 |
dot_accuracy@3 | 0.877 |
dot_accuracy@5 | 0.9244 |
dot_accuracy@10 | 0.9669 |
dot_precision@1 | 0.7073 |
dot_precision@3 | 0.2923 |
dot_precision@5 | 0.1849 |
dot_precision@10 | 0.0967 |
dot_recall@1 | 0.7073 |
dot_recall@3 | 0.877 |
dot_recall@5 | 0.9244 |
dot_recall@10 | 0.9669 |
dot_ndcg@10 | 0.8412 |
dot_mrr@10 | 0.8004 |
dot_map@100 | 0.8023 |
Training Details
Training Dataset
sentence-transformers/gooaq
- Dataset: sentence-transformers/gooaq at b089f72
- Size: 3,002,496 training samples
- Columns:
question
andanswer
- Approximate statistics based on the first 1000 samples:
question answer type string string details - min: 8 tokens
- mean: 11.89 tokens
- max: 22 tokens
- min: 15 tokens
- mean: 60.37 tokens
- max: 147 tokens
- Samples:
question answer biotechnology is best defined as?
Biotechnology is best defined as_______________? The science that involves using living organisms to produce needed materials. Which of the following tools of biotechnology, to do investigation, is used when trying crime?
how to open xye file?
Firstly, use File then Open and make sure that you can see All Files (*. *) and not just Excel files (the default option!) in the folder containing the *. xye file: Select the file you wish to open and Excel will bring up a wizard menu for importing plain text data into Excel (as shown below).
how much does california spend?
Estimated 2016 expenditures The total estimated government spending in California in fiscal year 2016 was $265.9 billion. Per-capita figures are calculated by taking the state's total spending and dividing by the number of state residents according to United States Census Bureau estimates.
- Loss:
MultipleNegativesRankingLoss
with these parameters:{ "scale": 20.0, "similarity_fct": "cos_sim" }
Evaluation Dataset
sentence-transformers/gooaq
- Dataset: sentence-transformers/gooaq at b089f72
- Size: 10,000 evaluation samples
- Columns:
question
andanswer
- Approximate statistics based on the first 1000 samples:
question answer type string string details - min: 8 tokens
- mean: 11.86 tokens
- max: 25 tokens
- min: 14 tokens
- mean: 60.82 tokens
- max: 166 tokens
- Samples:
question answer how to open nx file?
['Click File > Open. The File Open dialog box opens.', 'Select NX File (*. prt) in the Type box. ... ', 'Select an NX . ... ', 'Select Import in the File Open dialog box. ... ', 'If you do not want to retain the import profile in use, select an import profile from the Profile list. ... ', 'Click OK in the Import New Model dialog box.']
how to recover deleted photos from blackberry priv?
['Run Android Data Recovery. ... ', 'Enable USB Debugging Mode. ... ', 'Scan Your BlackBerry PRIV to Find Deleted Photos. ... ', 'Recover Deleted Photos from BlackBerry PRIV.']
which subatomic particles are found within the nucleus of an atom?
In the middle of every atom is the nucleus. The nucleus contains two types of subatomic particles, protons and neutrons. The protons have a positive electrical charge and the neutrons have no electrical charge. A third type of subatomic particle, electrons, move around the nucleus.
- Loss:
MultipleNegativesRankingLoss
with these parameters:{ "scale": 20.0, "similarity_fct": "cos_sim" }
Training Hyperparameters
Non-Default Hyperparameters
eval_strategy
: stepsper_device_train_batch_size
: 64per_device_eval_batch_size
: 64learning_rate
: 2e-05num_train_epochs
: 1warmup_ratio
: 0.1bf16
: Truebatch_sampler
: no_duplicates
All Hyperparameters
Click to expand
overwrite_output_dir
: Falsedo_predict
: Falseeval_strategy
: stepsprediction_loss_only
: Trueper_device_train_batch_size
: 64per_device_eval_batch_size
: 64per_gpu_train_batch_size
: Noneper_gpu_eval_batch_size
: Nonegradient_accumulation_steps
: 1eval_accumulation_steps
: Nonelearning_rate
: 2e-05weight_decay
: 0.0adam_beta1
: 0.9adam_beta2
: 0.999adam_epsilon
: 1e-08max_grad_norm
: 1.0num_train_epochs
: 1max_steps
: -1lr_scheduler_type
: linearlr_scheduler_kwargs
: {}warmup_ratio
: 0.1warmup_steps
: 0log_level
: passivelog_level_replica
: warninglog_on_each_node
: Truelogging_nan_inf_filter
: Truesave_safetensors
: Truesave_on_each_node
: Falsesave_only_model
: Falserestore_callback_states_from_checkpoint
: Falseno_cuda
: Falseuse_cpu
: Falseuse_mps_device
: Falseseed
: 42data_seed
: Nonejit_mode_eval
: Falseuse_ipex
: Falsebf16
: Truefp16
: Falsefp16_opt_level
: O1half_precision_backend
: autobf16_full_eval
: Falsefp16_full_eval
: Falsetf32
: Nonelocal_rank
: 0ddp_backend
: Nonetpu_num_cores
: Nonetpu_metrics_debug
: Falsedebug
: []dataloader_drop_last
: Falsedataloader_num_workers
: 0dataloader_prefetch_factor
: Nonepast_index
: -1disable_tqdm
: Falseremove_unused_columns
: Truelabel_names
: Noneload_best_model_at_end
: Falseignore_data_skip
: Falsefsdp
: []fsdp_min_num_params
: 0fsdp_config
: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}fsdp_transformer_layer_cls_to_wrap
: Noneaccelerator_config
: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}deepspeed
: Nonelabel_smoothing_factor
: 0.0optim
: adamw_torchoptim_args
: Noneadafactor
: Falsegroup_by_length
: Falselength_column_name
: lengthddp_find_unused_parameters
: Noneddp_bucket_cap_mb
: Noneddp_broadcast_buffers
: Falsedataloader_pin_memory
: Truedataloader_persistent_workers
: Falseskip_memory_metrics
: Trueuse_legacy_prediction_loop
: Falsepush_to_hub
: Falseresume_from_checkpoint
: Nonehub_model_id
: Nonehub_strategy
: every_savehub_private_repo
: Falsehub_always_push
: Falsegradient_checkpointing
: Falsegradient_checkpointing_kwargs
: Noneinclude_inputs_for_metrics
: Falseeval_do_concat_batches
: Truefp16_backend
: autopush_to_hub_model_id
: Nonepush_to_hub_organization
: Nonemp_parameters
:auto_find_batch_size
: Falsefull_determinism
: Falsetorchdynamo
: Noneray_scope
: lastddp_timeout
: 1800torch_compile
: Falsetorch_compile_backend
: Nonetorch_compile_mode
: Nonedispatch_batches
: Nonesplit_batches
: Noneinclude_tokens_per_second
: Falseinclude_num_input_tokens_seen
: Falseneftune_noise_alpha
: Noneoptim_target_modules
: Nonebatch_eval_metrics
: Falsebatch_sampler
: no_duplicatesmulti_dataset_batch_sampler
: proportional
Training Logs
Click to expand
Epoch | Step | Training Loss | loss | gooaq-dev_cosine_map@100 |
---|---|---|---|---|
0 | 0 | - | - | 0.1379 |
0.0000 | 1 | 3.6452 | - | - |
0.0053 | 250 | 2.4418 | - | - |
0.0107 | 500 | 0.373 | - | - |
0.0160 | 750 | 0.183 | - | - |
0.0213 | 1000 | 0.1286 | 0.0805 | 0.6796 |
0.0266 | 1250 | 0.1099 | - | - |
0.0320 | 1500 | 0.091 | - | - |
0.0373 | 1750 | 0.0768 | - | - |
0.0426 | 2000 | 0.0665 | 0.0526 | 0.7162 |
0.0480 | 2250 | 0.0659 | - | - |
0.0533 | 2500 | 0.0602 | - | - |
0.0586 | 2750 | 0.0548 | - | - |
0.0639 | 3000 | 0.0543 | 0.0426 | 0.7328 |
0.0693 | 3250 | 0.0523 | - | - |
0.0746 | 3500 | 0.0494 | - | - |
0.0799 | 3750 | 0.0468 | - | - |
0.0853 | 4000 | 0.0494 | 0.0362 | 0.7450 |
0.0906 | 4250 | 0.048 | - | - |
0.0959 | 4500 | 0.0442 | - | - |
0.1012 | 4750 | 0.0442 | - | - |
0.1066 | 5000 | 0.0408 | 0.0332 | 0.7519 |
0.1119 | 5250 | 0.0396 | - | - |
0.1172 | 5500 | 0.0379 | - | - |
0.1226 | 5750 | 0.0392 | - | - |
0.1279 | 6000 | 0.0395 | 0.0300 | 0.7505 |
0.1332 | 6250 | 0.0349 | - | - |
0.1386 | 6500 | 0.0383 | - | - |
0.1439 | 6750 | 0.0335 | - | - |
0.1492 | 7000 | 0.0323 | 0.0253 | 0.7624 |
0.1545 | 7250 | 0.0342 | - | - |
0.1599 | 7500 | 0.0292 | - | - |
0.1652 | 7750 | 0.0309 | - | - |
0.1705 | 8000 | 0.0335 | 0.0249 | 0.7631 |
0.1759 | 8250 | 0.0304 | - | - |
0.1812 | 8500 | 0.0318 | - | - |
0.1865 | 8750 | 0.0271 | - | - |
0.1918 | 9000 | 0.029 | 0.0230 | 0.7615 |
0.1972 | 9250 | 0.0309 | - | - |
0.2025 | 9500 | 0.0305 | - | - |
0.2078 | 9750 | 0.0237 | - | - |
0.2132 | 10000 | 0.0274 | 0.0220 | 0.7667 |
0.2185 | 10250 | 0.0248 | - | - |
0.2238 | 10500 | 0.0249 | - | - |
0.2291 | 10750 | 0.0272 | - | - |
0.2345 | 11000 | 0.0289 | 0.0230 | 0.7664 |
0.2398 | 11250 | 0.027 | - | - |
0.2451 | 11500 | 0.0259 | - | - |
0.2505 | 11750 | 0.0237 | - | - |
0.2558 | 12000 | 0.0245 | 0.0220 | 0.7694 |
0.2611 | 12250 | 0.0251 | - | - |
0.2664 | 12500 | 0.0243 | - | - |
0.2718 | 12750 | 0.0229 | - | - |
0.2771 | 13000 | 0.0273 | 0.0201 | 0.7725 |
0.2824 | 13250 | 0.0244 | - | - |
0.2878 | 13500 | 0.0248 | - | - |
0.2931 | 13750 | 0.0255 | - | - |
0.2984 | 14000 | 0.0244 | 0.0192 | 0.7729 |
0.3037 | 14250 | 0.0242 | - | - |
0.3091 | 14500 | 0.0235 | - | - |
0.3144 | 14750 | 0.0231 | - | - |
0.3197 | 15000 | 0.0228 | 0.0190 | 0.7823 |
0.3251 | 15250 | 0.0229 | - | - |
0.3304 | 15500 | 0.0224 | - | - |
0.3357 | 15750 | 0.0216 | - | - |
0.3410 | 16000 | 0.0218 | 0.0186 | 0.7787 |
0.3464 | 16250 | 0.022 | - | - |
0.3517 | 16500 | 0.0233 | - | - |
0.3570 | 16750 | 0.0216 | - | - |
0.3624 | 17000 | 0.0226 | 0.0169 | 0.7862 |
0.3677 | 17250 | 0.0215 | - | - |
0.3730 | 17500 | 0.0212 | - | - |
0.3784 | 17750 | 0.0178 | - | - |
0.3837 | 18000 | 0.0217 | 0.0161 | 0.7813 |
0.3890 | 18250 | 0.0217 | - | - |
0.3943 | 18500 | 0.0191 | - | - |
0.3997 | 18750 | 0.0216 | - | - |
0.4050 | 19000 | 0.022 | 0.0157 | 0.7868 |
0.4103 | 19250 | 0.0223 | - | - |
0.4157 | 19500 | 0.021 | - | - |
0.4210 | 19750 | 0.0176 | - | - |
0.4263 | 20000 | 0.021 | 0.0162 | 0.7873 |
0.4316 | 20250 | 0.0206 | - | - |
0.4370 | 20500 | 0.0196 | - | - |
0.4423 | 20750 | 0.0186 | - | - |
0.4476 | 21000 | 0.0197 | 0.0158 | 0.7907 |
0.4530 | 21250 | 0.0156 | - | - |
0.4583 | 21500 | 0.0178 | - | - |
0.4636 | 21750 | 0.0175 | - | - |
0.4689 | 22000 | 0.0187 | 0.0151 | 0.7937 |
0.4743 | 22250 | 0.0182 | - | - |
0.4796 | 22500 | 0.0185 | - | - |
0.4849 | 22750 | 0.0217 | - | - |
0.4903 | 23000 | 0.0179 | 0.0156 | 0.7937 |
0.4956 | 23250 | 0.0193 | - | - |
0.5009 | 23500 | 0.015 | - | - |
0.5062 | 23750 | 0.0181 | - | - |
0.5116 | 24000 | 0.0173 | 0.0150 | 0.7924 |
0.5169 | 24250 | 0.0177 | - | - |
0.5222 | 24500 | 0.0183 | - | - |
0.5276 | 24750 | 0.0171 | - | - |
0.5329 | 25000 | 0.0185 | 0.0140 | 0.7955 |
0.5382 | 25250 | 0.0178 | - | - |
0.5435 | 25500 | 0.015 | - | - |
0.5489 | 25750 | 0.017 | - | - |
0.5542 | 26000 | 0.0171 | 0.0139 | 0.7931 |
0.5595 | 26250 | 0.0164 | - | - |
0.5649 | 26500 | 0.0175 | - | - |
0.5702 | 26750 | 0.0175 | - | - |
0.5755 | 27000 | 0.0163 | 0.0133 | 0.7954 |
0.5809 | 27250 | 0.0179 | - | - |
0.5862 | 27500 | 0.016 | - | - |
0.5915 | 27750 | 0.0155 | - | - |
0.5968 | 28000 | 0.0162 | 0.0138 | 0.7979 |
0.6022 | 28250 | 0.0164 | - | - |
0.6075 | 28500 | 0.0148 | - | - |
0.6128 | 28750 | 0.0152 | - | - |
0.6182 | 29000 | 0.0166 | 0.0134 | 0.7987 |
0.6235 | 29250 | 0.0159 | - | - |
0.6288 | 29500 | 0.0168 | - | - |
0.6341 | 29750 | 0.0187 | - | - |
0.6395 | 30000 | 0.017 | 0.0137 | 0.7980 |
0.6448 | 30250 | 0.0168 | - | - |
0.6501 | 30500 | 0.0149 | - | - |
0.6555 | 30750 | 0.0159 | - | - |
0.6608 | 31000 | 0.0149 | 0.0131 | 0.8017 |
0.6661 | 31250 | 0.0149 | - | - |
0.6714 | 31500 | 0.0147 | - | - |
0.6768 | 31750 | 0.0157 | - | - |
0.6821 | 32000 | 0.0151 | 0.0125 | 0.8011 |
0.6874 | 32250 | 0.015 | - | - |
0.6928 | 32500 | 0.0157 | - | - |
0.6981 | 32750 | 0.0153 | - | - |
0.7034 | 33000 | 0.0141 | 0.0123 | 0.8012 |
0.7087 | 33250 | 0.0143 | - | - |
0.7141 | 33500 | 0.0121 | - | - |
0.7194 | 33750 | 0.0164 | - | - |
0.7247 | 34000 | 0.014 | 0.0121 | 0.8014 |
0.7301 | 34250 | 0.0147 | - | - |
0.7354 | 34500 | 0.0149 | - | - |
0.7407 | 34750 | 0.014 | - | - |
0.7460 | 35000 | 0.0156 | 0.0117 | 0.8022 |
0.7514 | 35250 | 0.0153 | - | - |
0.7567 | 35500 | 0.0146 | - | - |
0.7620 | 35750 | 0.0144 | - | - |
0.7674 | 36000 | 0.0139 | 0.0111 | 0.8035 |
0.7727 | 36250 | 0.0134 | - | - |
0.7780 | 36500 | 0.013 | - | - |
0.7833 | 36750 | 0.0156 | - | - |
0.7887 | 37000 | 0.0144 | 0.0108 | 0.8048 |
0.7940 | 37250 | 0.0133 | - | - |
0.7993 | 37500 | 0.0154 | - | - |
0.8047 | 37750 | 0.0132 | - | - |
0.8100 | 38000 | 0.013 | 0.0108 | 0.8063 |
0.8153 | 38250 | 0.0126 | - | - |
0.8207 | 38500 | 0.0135 | - | - |
0.8260 | 38750 | 0.014 | - | - |
0.8313 | 39000 | 0.013 | 0.0109 | 0.8086 |
0.8366 | 39250 | 0.0136 | - | - |
0.8420 | 39500 | 0.0141 | - | - |
0.8473 | 39750 | 0.0155 | - | - |
0.8526 | 40000 | 0.0153 | 0.0106 | 0.8075 |
0.8580 | 40250 | 0.0131 | - | - |
0.8633 | 40500 | 0.0128 | - | - |
0.8686 | 40750 | 0.013 | - | - |
0.8739 | 41000 | 0.0133 | 0.0109 | 0.8060 |
0.8793 | 41250 | 0.0119 | - | - |
0.8846 | 41500 | 0.0144 | - | - |
0.8899 | 41750 | 0.0142 | - | - |
0.8953 | 42000 | 0.0138 | 0.0105 | 0.8083 |
0.9006 | 42250 | 0.014 | - | - |
0.9059 | 42500 | 0.0134 | - | - |
0.9112 | 42750 | 0.0134 | - | - |
0.9166 | 43000 | 0.0124 | 0.0106 | 0.8113 |
0.9219 | 43250 | 0.0122 | - | - |
0.9272 | 43500 | 0.0126 | - | - |
0.9326 | 43750 | 0.0121 | - | - |
0.9379 | 44000 | 0.0137 | 0.0103 | 0.8105 |
0.9432 | 44250 | 0.0132 | - | - |
0.9485 | 44500 | 0.012 | - | - |
0.9539 | 44750 | 0.0136 | - | - |
0.9592 | 45000 | 0.0133 | 0.0104 | 0.8112 |
0.9645 | 45250 | 0.0118 | - | - |
0.9699 | 45500 | 0.0132 | - | - |
0.9752 | 45750 | 0.0118 | - | - |
0.9805 | 46000 | 0.012 | 0.0102 | 0.8104 |
0.9858 | 46250 | 0.0127 | - | - |
0.9912 | 46500 | 0.0134 | - | - |
0.9965 | 46750 | 0.0121 | - | - |
1.0 | 46914 | - | - | 0.8111 |
Environmental Impact
Carbon emissions were measured using CodeCarbon.
- Energy Consumed: 1.637 kWh
- Carbon Emitted: 0.636 kg of CO2
- Hours Used: 4.514 hours
Training Hardware
- On Cloud: No
- GPU Model: 1 x NVIDIA GeForce RTX 3090
- CPU Model: 13th Gen Intel(R) Core(TM) i7-13700K
- RAM Size: 31.78 GB
Framework Versions
- Python: 3.11.6
- Sentence Transformers: 3.1.0.dev0
- Transformers: 4.41.2
- PyTorch: 2.3.0+cu121
- Accelerate: 0.30.1
- Datasets: 2.19.1
- Tokenizers: 0.19.1
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",
}
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}
}
- Downloads last month
- 264
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.
Model tree for tomaarsen/mpnet-base-gooaq
Base model
microsoft/mpnet-base
Finetuned
this model
Dataset used to train tomaarsen/mpnet-base-gooaq
Space using tomaarsen/mpnet-base-gooaq 1
Evaluation results
- Cosine Accuracy@1 on gooaq devself-reported0.720
- Cosine Accuracy@3 on gooaq devself-reported0.884
- Cosine Accuracy@5 on gooaq devself-reported0.930
- Cosine Accuracy@10 on gooaq devself-reported0.971
- Cosine Precision@1 on gooaq devself-reported0.720
- Cosine Precision@3 on gooaq devself-reported0.295
- Cosine Precision@5 on gooaq devself-reported0.186
- Cosine Precision@10 on gooaq devself-reported0.097
- Cosine Recall@1 on gooaq devself-reported0.720
- Cosine Recall@3 on gooaq devself-reported0.884