richie-ghost/sbert_facebook_large_mnli_openVino2

SentenceTransformer based on FacebookAI/roberta-large-mnli

This is a sentence-transformers model finetuned from FacebookAI/roberta-large-mnli. It maps sentences & paragraphs to a 1024-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: FacebookAI/roberta-large-mnli
• Maximum Sequence Length: 512 tokens
• Output Dimensionality: 1024 tokens
• Similarity Function: Cosine Similarity

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: RobertaModel 
  (1): Pooling({'word_embedding_dimension': 1024, '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("richie-ghost/sbert_facebook_large_mnli_openVino2")
# Run inference
sentences = [
    'A motorbike rider is barreling across a grass lawn.',
    'The rider is outdoors on a motorbike.',
    'The girl is wearing a shirt.',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 1024]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [3, 3]

Evaluation

Metrics

Semantic Similarity

• Dataset: eval
• Evaluated with EmbeddingSimilarityEvaluator

Metric	Value
pearson_cosine	0.8457
spearman_cosine	0.8101
pearson_manhattan	0.8108
spearman_manhattan	0.7917
pearson_euclidean	0.8106
spearman_euclidean	0.7916
pearson_dot	0.8567
spearman_dot	0.8163
pearson_max	0.8567
spearman_max	0.8163

Training Details

Training Dataset

Unnamed Dataset

• Size: 72,338 training samples
• Columns: sentence_0, sentence_1, and label
• Approximate statistics based on the first 1000 samples:

  	sentence_0	sentence_1	label
  type	string	string	int
  details	min: 5 tokens mean: 18.11 tokens max: 82 tokens	min: 5 tokens mean: 12.82 tokens max: 65 tokens	0: ~50.70% 1: ~49.30%

• Samples:

  sentence_0	sentence_1	label
  Hows would you create strategies and tactics in various combat situations?	I have girlfriend and their parents accepted for my marriage, I m working in Nagpur but her parents wanted me to shift Bangalore? Is it valid wish?	0
  Man from the army speaking with civilian women.	The man is a sergeant	0
  An old man with a white shirt and black pants sits on a chair in the opening of a stone tunnel.	Someone has black pants.	1

• Loss: CosineSimilarityLoss with these parameters:

  {
      "loss_fct": "torch.nn.modules.loss.MSELoss"
  }
  

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 16
• per_device_eval_batch_size: 16
• num_train_epochs: 4
• 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
• torch_empty_cache_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: 4
• 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
• eval_on_start: False
• eval_use_gather_object: False
• batch_sampler: batch_sampler
• multi_dataset_batch_sampler: round_robin

Training Logs

Epoch	Step	Training Loss	eval_spearman_max
0.1106	500	0.1845	0.6681
0.2211	1000	0.0942	0.7711
0.3317	1500	0.0821	0.6355
0.4423	2000	0.0794	0.7283
0.5529	2500	0.0788	0.7129
0.6634	3000	0.0737	0.7853
0.7740	3500	0.07	0.7013
0.8846	4000	0.0686	0.7809
0.9951	4500	0.0683	0.7578
1.0	4522	-	0.7976
1.1057	5000	0.07	0.7749
1.2163	5500	0.0656	0.7826
1.3268	6000	0.0587	0.8032
1.4374	6500	0.0584	0.7666
1.5480	7000	0.0582	0.7917
1.6586	7500	0.0546	0.7945
1.7691	8000	0.0528	0.7786
1.8797	8500	0.051	0.7732
1.9903	9000	0.0527	0.7996
2.0	9044	-	0.7898
2.1008	9500	0.0509	0.7957
2.2114	10000	0.0492	0.7988
2.3220	10500	0.0451	0.8044
2.4326	11000	0.0443	0.7961
2.5431	11500	0.0445	0.7975
2.6537	12000	0.0433	0.8054
2.7643	12500	0.0394	0.7890
2.8748	13000	0.0387	0.8020
2.9854	13500	0.0401	0.8096
3.0	13566	-	0.8087
3.0960	14000	0.0399	0.8098
3.2065	14500	0.039	0.8077
3.3171	15000	0.0346	0.8021
3.4277	15500	0.0339	0.8082
3.5383	16000	0.0347	0.8150
3.6488	16500	0.0352	0.8144
3.7594	17000	0.032	0.8141
3.8700	17500	0.0326	0.8151
3.9805	18000	0.0318	0.8162
4.0	18088	-	0.8163

Framework Versions

• Python: 3.10.12
• Sentence Transformers: 3.2.1
• Transformers: 4.44.2
• PyTorch: 2.4.1+cu121
• Accelerate: 1.0.1
• Datasets: 3.0.1
• 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",
}