chanbistec's picture
Add new SentenceTransformer model.
e011522 verified
metadata
base_model: microsoft/mpnet-base
datasets:
  - sentence-transformers/all-nli
language:
  - en
library_name: sentence-transformers
license: apache-2.0
metrics:
  - cosine_accuracy
  - dot_accuracy
  - manhattan_accuracy
  - euclidean_accuracy
  - max_accuracy
pipeline_tag: sentence-similarity
tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - generated_from_trainer
  - dataset_size:557850
  - loss:MultipleNegativesRankingLoss
widget:
  - source_sentence: A man is jumping unto his filthy bed.
    sentences:
      - A young male is looking at a newspaper while 2 females walks past him.
      - The bed is dirty.
      - The man is on the moon.
  - source_sentence: >-
      A carefully balanced male stands on one foot near a clean ocean beach
      area.
    sentences:
      - A man is ouside near the beach.
      - Three policemen patrol the streets on bikes
      - A man is sitting on his couch.
  - source_sentence: The man is wearing a blue shirt.
    sentences:
      - Near the trashcan the man stood and smoked
      - >-
        A man in a blue shirt leans on a wall beside a road with a blue van and
        red car with water in the background.
      - A man in a black shirt is playing a guitar.
  - source_sentence: The girls are outdoors.
    sentences:
      - Two girls riding on an amusement part ride.
      - a guy laughs while doing laundry
      - >-
        Three girls are standing together in a room, one is listening, one is
        writing on a wall and the third is talking to them.
  - source_sentence: >-
      A construction worker peeking out of a manhole while his coworker sits on
      the sidewalk smiling.
    sentences:
      - A worker is looking out of a manhole.
      - A man is giving a presentation.
      - The workers are both inside the manhole.
model-index:
  - name: MPNet base trained on AllNLI triplets
    results:
      - task:
          type: triplet
          name: Triplet
        dataset:
          name: all nli dev
          type: all-nli-dev
        metrics:
          - type: cosine_accuracy
            value: 0.9141859052247874
            name: Cosine Accuracy
          - type: dot_accuracy
            value: 0.08444714459295262
            name: Dot Accuracy
          - type: manhattan_accuracy
            value: 0.9097812879708383
            name: Manhattan Accuracy
          - type: euclidean_accuracy
            value: 0.9097812879708383
            name: Euclidean Accuracy
          - type: max_accuracy
            value: 0.9141859052247874
            name: Max Accuracy
      - task:
          type: triplet
          name: Triplet
        dataset:
          name: all nli test
          type: all-nli-test
        metrics:
          - type: cosine_accuracy
            value: 0.926463912846119
            name: Cosine Accuracy
          - type: dot_accuracy
            value: 0.07353608715388107
            name: Dot Accuracy
          - type: manhattan_accuracy
            value: 0.9187471629596006
            name: Manhattan Accuracy
          - type: euclidean_accuracy
            value: 0.9179906188530791
            name: Euclidean Accuracy
          - type: max_accuracy
            value: 0.926463912846119
            name: Max Accuracy

MPNet base trained on AllNLI triplets

This is a sentence-transformers model finetuned from microsoft/mpnet-base on the all-nli 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: 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

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("chanbistec/mpnet-base-all-nli-triplet")
# Run inference
sentences = [
    'A construction worker peeking out of a manhole while his coworker sits on the sidewalk smiling.',
    'A worker is looking out of a manhole.',
    'The workers are both inside the manhole.',
]
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

Triplet

Metric Value
cosine_accuracy 0.9142
dot_accuracy 0.0844
manhattan_accuracy 0.9098
euclidean_accuracy 0.9098
max_accuracy 0.9142

Triplet

Metric Value
cosine_accuracy 0.9265
dot_accuracy 0.0735
manhattan_accuracy 0.9187
euclidean_accuracy 0.918
max_accuracy 0.9265

Training Details

Training Dataset

all-nli

  • Dataset: all-nli at d482672
  • Size: 557,850 training samples
  • Columns: anchor, positive, and negative
  • Approximate statistics based on the first 1000 samples:
    anchor positive negative
    type string string string
    details
    • min: 7 tokens
    • mean: 10.46 tokens
    • max: 46 tokens
    • min: 6 tokens
    • mean: 12.81 tokens
    • max: 40 tokens
    • min: 5 tokens
    • mean: 13.4 tokens
    • max: 50 tokens
  • Samples:
    anchor positive negative
    A person on a horse jumps over a broken down airplane. A person is outdoors, on a horse. A person is at a diner, ordering an omelette.
    Children smiling and waving at camera There are children present The kids are frowning
    A boy is jumping on skateboard in the middle of a red bridge. The boy does a skateboarding trick. The boy skates down the sidewalk.
  • Loss: MultipleNegativesRankingLoss with these parameters:
    {
        "scale": 20.0,
        "similarity_fct": "cos_sim"
    }
    

Evaluation Dataset

all-nli

  • Dataset: all-nli at d482672
  • Size: 6,584 evaluation samples
  • Columns: anchor, positive, and negative
  • Approximate statistics based on the first 1000 samples:
    anchor positive negative
    type string string string
    details
    • min: 6 tokens
    • mean: 17.95 tokens
    • max: 63 tokens
    • min: 4 tokens
    • mean: 9.78 tokens
    • max: 29 tokens
    • min: 5 tokens
    • mean: 10.35 tokens
    • max: 29 tokens
  • Samples:
    anchor positive negative
    Two women are embracing while holding to go packages. Two woman are holding packages. The men are fighting outside a deli.
    Two young children in blue jerseys, one with the number 9 and one with the number 2 are standing on wooden steps in a bathroom and washing their hands in a sink. Two kids in numbered jerseys wash their hands. Two kids in jackets walk to school.
    A man selling donuts to a customer during a world exhibition event held in the city of Angeles A man selling donuts to a customer. A woman drinks her coffee in a small cafe.
  • Loss: MultipleNegativesRankingLoss with these parameters:
    {
        "scale": 20.0,
        "similarity_fct": "cos_sim"
    }
    

Training Hyperparameters

Non-Default Hyperparameters

  • eval_strategy: steps
  • per_device_train_batch_size: 16
  • per_device_eval_batch_size: 16
  • learning_rate: 2e-05
  • num_train_epochs: 1
  • warmup_ratio: 0.1
  • 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: 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
  • 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: 1
  • 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: 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
  • eval_on_start: False
  • eval_use_gather_object: False
  • batch_sampler: no_duplicates
  • multi_dataset_batch_sampler: proportional

Training Logs

Epoch Step Training Loss loss all-nli-dev_max_accuracy all-nli-test_max_accuracy
0 0 - - 0.6832 -
0.016 100 3.0282 1.5782 0.7752 -
0.032 200 1.2529 0.9154 0.7991 -
0.048 300 1.4472 0.7901 0.8103 -
0.064 400 0.9059 0.7468 0.8114 -
0.08 500 0.8663 0.8423 0.7981 -
0.096 600 1.0836 0.8995 0.8010 -
0.112 700 0.9315 0.8971 0.8100 -
0.128 800 1.1273 0.9654 0.8012 -
0.144 900 1.1194 0.9318 0.8303 -
0.16 1000 1.0911 0.9048 0.8038 -
0.176 1100 1.1332 0.9340 0.8039 -
0.192 1200 1.0154 0.9041 0.8076 -
0.208 1300 0.7995 0.9301 0.7959 -
0.224 1400 0.7614 0.8275 0.8071 -
0.24 1500 0.8724 0.7973 0.8173 -
0.256 1600 0.6751 0.7916 0.8197 -
0.272 1700 0.8933 0.8572 0.8194 -
0.288 1800 0.8585 0.8560 0.8056 -
0.304 1900 0.8354 0.7987 0.8123 -
0.32 2000 0.7484 0.7559 0.8348 -
0.336 2100 0.6047 0.7532 0.8471 -
0.352 2200 0.6221 0.6956 0.8665 -
0.368 2300 0.8332 0.7214 0.8542 -
0.384 2400 0.7755 0.7007 0.8481 -
0.4 2500 0.6912 0.7505 0.8499 -
0.416 2600 0.6169 0.6536 0.8591 -
0.432 2700 0.8907 0.7240 0.8560 -
0.448 2800 0.8576 0.6790 0.8499 -
0.464 2900 0.8057 0.6870 0.8575 -
0.48 3000 0.6928 0.6540 0.8641 -
0.496 3100 0.7566 0.6419 0.8682 -
0.512 3200 0.5757 0.6109 0.8783 -
0.528 3300 0.601 0.5481 0.8914 -
0.544 3400 0.5105 0.5853 0.8820 -
0.56 3500 0.5116 0.5918 0.8961 -
0.576 3600 0.495 0.5546 0.8897 -
0.592 3700 0.5585 0.5457 0.8970 -
0.608 3800 0.4778 0.5056 0.9020 -
0.624 3900 0.5116 0.5203 0.9019 -
0.64 4000 0.753 0.5490 0.9019 -
0.656 4100 0.9207 0.5447 0.9049 -
0.672 4200 0.8695 0.4996 0.9055 -
0.688 4300 0.6867 0.4825 0.9107 -
0.704 4400 0.5961 0.4670 0.9166 -
0.72 4500 0.5547 0.4748 0.9104 -
0.736 4600 0.6145 0.4636 0.9145 -
0.752 4700 0.6643 0.4806 0.9128 -
0.768 4800 0.6134 0.4521 0.9110 -
0.784 4900 0.5847 0.4627 0.9080 -
0.8 5000 0.6482 0.4853 0.9107 -
0.816 5100 0.5103 0.4374 0.9104 -
0.832 5200 0.5639 0.4306 0.9089 -
0.848 5300 0.5247 0.4418 0.9116 -
0.864 5400 0.6094 0.4564 0.9101 -
0.88 5500 0.5296 0.4394 0.9092 -
0.896 5600 0.5469 0.4316 0.9101 -
0.912 5700 0.6061 0.4258 0.9124 -
0.928 5800 0.5456 0.4167 0.9113 -
0.944 5900 0.6776 0.4168 0.9108 -
0.96 6000 0.7401 0.4267 0.9139 -
0.976 6100 0.6568 0.4227 0.9140 -
0.992 6200 0.0002 0.4224 0.9142 -
1.0 6250 - - - 0.9265

Framework Versions

  • Python: 3.12.4
  • Sentence Transformers: 3.1.0
  • Transformers: 4.44.2
  • PyTorch: 2.4.1
  • Accelerate: 0.34.2
  • Datasets: 3.0.0
  • 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}
}