srikarvar/fine_tuned_model_9

SentenceTransformer based on intfloat/multilingual-e5-small

This is a sentence-transformers model finetuned from intfloat/multilingual-e5-small. It maps sentences & paragraphs to a 384-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/multilingual-e5-small
• Maximum Sequence Length: 512 tokens
• Output Dimensionality: 384 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: BertModel 
  (1): Pooling({'word_embedding_dimension': 384, '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})
  (2): Normalize()
)

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("srikarvar/fine_tuned_model_9")
# Run inference
sentences = [
    'Who wrote the book "1984"?',
    'Who wrote the book "To Kill a Mockingbird"?',
    'What is the speed of light?',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 384]

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

Evaluation

Metrics

Binary Classification

• Dataset: pair-class-dev
• Evaluated with BinaryClassificationEvaluator

Metric	Value
cosine_accuracy	0.8623
cosine_accuracy_threshold	0.8492
cosine_f1	0.8856
cosine_f1_threshold	0.8245
cosine_precision	0.8436
cosine_recall	0.9321
cosine_ap	0.9267
dot_accuracy	0.8623
dot_accuracy_threshold	0.8492
dot_f1	0.8856
dot_f1_threshold	0.8245
dot_precision	0.8436
dot_recall	0.9321
dot_ap	0.9267
manhattan_accuracy	0.8623
manhattan_accuracy_threshold	8.5996
manhattan_f1	0.8856
manhattan_f1_threshold	9.2211
manhattan_precision	0.8436
manhattan_recall	0.9321
manhattan_ap	0.926
euclidean_accuracy	0.8623
euclidean_accuracy_threshold	0.5492
euclidean_f1	0.8856
euclidean_f1_threshold	0.5924
euclidean_precision	0.8436
euclidean_recall	0.9321
euclidean_ap	0.9267
max_accuracy	0.8623
max_accuracy_threshold	8.5996
max_f1	0.8856
max_f1_threshold	9.2211
max_precision	0.8436
max_recall	0.9321
max_ap	0.9267

Binary Classification

• Dataset: pair-class-test
• Evaluated with BinaryClassificationEvaluator

Metric	Value
cosine_accuracy	0.8659
cosine_accuracy_threshold	0.8321
cosine_f1	0.8875
cosine_f1_threshold	0.8321
cosine_precision	0.8743
cosine_recall	0.9012
cosine_ap	0.9258
dot_accuracy	0.8659
dot_accuracy_threshold	0.8321
dot_f1	0.8875
dot_f1_threshold	0.8321
dot_precision	0.8743
dot_recall	0.9012
dot_ap	0.9258
manhattan_accuracy	0.8623
manhattan_accuracy_threshold	8.8548
manhattan_f1	0.8876
manhattan_f1_threshold	9.3493
manhattan_precision	0.8523
manhattan_recall	0.9259
manhattan_ap	0.9255
euclidean_accuracy	0.8659
euclidean_accuracy_threshold	0.5796
euclidean_f1	0.8875
euclidean_f1_threshold	0.5796
euclidean_precision	0.8743
euclidean_recall	0.9012
euclidean_ap	0.9258
max_accuracy	0.8659
max_accuracy_threshold	8.8548
max_f1	0.8876
max_f1_threshold	9.3493
max_precision	0.8743
max_recall	0.9259
max_ap	0.9258

Training Details

Training Dataset

Unnamed Dataset

• Size: 2,476 training samples
• Columns: sentence2, label, and sentence1
• Approximate statistics based on the first 1000 samples:

  	sentence2	label	sentence1
  type	string	int	string
  details	min: 4 tokens mean: 16.06 tokens max: 98 tokens	0: ~40.20% 1: ~59.80%	min: 6 tokens mean: 16.35 tokens max: 98 tokens

• Samples:

  sentence2	label	sentence1
  A model is trained using the ImageNet dataset to classify images into distinct categories.	1	The ImageNet dataset is used for training models to classify images into various categories.
  Version 5.3.1 does not contain it.	1	No, it doesn't exist in version 5.3.1.
  Can you do my homework for me?	0	Can you help me with my homework?

• Loss: OnlineContrastiveLoss

Evaluation Dataset

Unnamed Dataset

• Size: 276 evaluation samples
• Columns: sentence2, label, and sentence1
• Approximate statistics based on the first 276 samples:

  	sentence2	label	sentence1
  type	string	int	string
  details	min: 5 tokens mean: 15.34 tokens max: 86 tokens	0: ~41.30% 1: ~58.70%	min: 6 tokens mean: 15.56 tokens max: 87 tokens

• Samples:

  sentence2	label	sentence1
  How is AI used to enhance cybersecurity?	0	What are the challenges of AI in cybersecurity?
  The SYSTEM log documentation can be accessed by clicking on the link which will take you to the main version.	1	You can find the SYSTEM log documentation on the main version. Click on the provided link to redirect to the main version of the documentation.
  Name the capital city of Italy	1	What is the capital of Italy?

• Loss: OnlineContrastiveLoss

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: epoch
• per_device_train_batch_size: 32
• per_device_eval_batch_size: 32
• gradient_accumulation_steps: 2
• num_train_epochs: 4
• warmup_ratio: 0.1
• 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: 32
• per_gpu_train_batch_size: None
• per_gpu_eval_batch_size: None
• gradient_accumulation_steps: 2
• 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: 1.0
• num_train_epochs: 4
• 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: 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: 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: no_duplicates
• multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	loss	pair-class-dev_max_ap	pair-class-test_max_ap
0	0	-	-	0.7876	-
0.2564	10	1.6257	-	-	-
0.5128	20	0.8138	-	-	-
0.7692	30	0.7276	-	-	-
1.0	39	-	0.8190	0.9089	-
1.0256	40	0.6423	-	-	-
1.2821	50	0.5168	-	-	-
1.5385	60	0.3583	-	-	-
1.7949	70	0.3182	-	-	-
2.0	78	-	0.7351	0.9215	-
2.0513	80	0.3521	-	-	-
2.3077	90	0.2037	-	-	-
2.5641	100	0.1293	-	-	-
2.8205	110	0.1374	-	-	-
3.0	117	-	0.7223	0.9258	-
3.0769	120	0.198	-	-	-
3.3333	130	0.0667	-	-	-
3.5897	140	0.0526	-	-	-
3.8462	150	0.0652	-	-	-
4.0	156	-	0.7327	0.9267	0.9258

• The bold row denotes the saved checkpoint.

Framework Versions

• Python: 3.10.12
• Sentence Transformers: 3.1.0
• Transformers: 4.41.2
• PyTorch: 2.1.2+cu121
• Accelerate: 0.34.2
• 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",
}