kiarashmo/chembberta-77m-sBERT-finetuned-on-bace-with-sBERTBatchAllTripletLoss

SentenceTransformer based on kiarashmo/chembberta-77m-mlm-safetensors

This is a sentence-transformers model finetuned from kiarashmo/chembberta-77m-mlm-safetensors. 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: kiarashmo/chembberta-77m-mlm-safetensors
• Maximum Sequence Length: 512 tokens
• Output Dimensionality: 384 dimensions
• 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, 'architecture': 'RobertaModel'})
  (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})
)

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("sentence_transformers_model_id")
# Run inference
sentences = [
    'CC(=O)NC(Cc1cc(F)cc(F)c1)C(O)C[NH2+]C1(c2cccc(-c3cscn3)c2)CCCCC1',
    'CC1(C)Cc2cc(Cl)ccc2C(NC(Cc2ccccc2)c2nc3ccncc3c(=O)[nH]2)=N1',
    'COc1ccc(C2(c3cccc(-c4cccnc4)c3)N=C(N)N(C)C2=O)cc1C1CCCC1',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 384]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities)
# tensor([[ 1.0000,  0.9662, -0.4132],
#         [ 0.9662,  1.0000, -0.3956],
#         [-0.4132, -0.3956,  1.0000]])

Evaluation

Metrics

Binary Classification

• Dataset: val-sim
• Evaluated with BinaryClassificationEvaluator

Metric	Value
cosine_accuracy	0.597
cosine_accuracy_threshold	-0.0509
cosine_f1	0.6744
cosine_f1_threshold	-0.6213
cosine_precision	0.5124
cosine_recall	0.986
cosine_ap	0.6063
cosine_mcc	0.1185

Training Details

Training Dataset

Unnamed Dataset

• Size: 1,390 training samples
• Columns: text and label
• Approximate statistics based on the first 1000 samples:

  	text	label
  type	string	int
  details	min: 17 tokens mean: 61.3 tokens max: 129 tokens	0: ~49.90% 1: ~50.10%

• Samples:

  text	label
  CN1C(=O)C(c2ccc(OC(F)F)cc2)(c2ccc(F)c(C#CCCCF)c2)N=C1N	1
  CN1C(=O)C(c2ccc(OC(F)F)cc2)(c2cccc(C#CCF)c2)N=C1N	1
  CN1C(=O)C(c2ccc(OC(F)F)cc2)(c2cccc(C#CCCF)c2)N=C1N	1

• Loss: BatchAllTripletLoss

Evaluation Dataset

Unnamed Dataset

• Size: 168 evaluation samples
• Columns: text and label
• Approximate statistics based on the first 168 samples:

  	text	label
  type	string	int
  details	min: 46 tokens mean: 63.26 tokens max: 110 tokens	0: ~50.00% 1: ~50.00%

• Samples:

  text	label
  CC(C)(C)c1cccc(C[NH2+]C2CS(=O)(=O)CC(Cc3cc(F)c4c(c3)C3(CCC3)CN4)C2O)c1	0
  CC(=O)NC(Cc1cc(F)cc(F)c1)C(O)C[NH2+]C1(c2cccc(C3CCOCOC3)c2)CCCCC1	0
  CC1(c2cccc(NC(=O)c3ccc(Cl)cn3)c2)N=C(N)c2ccccc21	0

• Loss: BatchAllTripletLoss

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 32
• per_device_eval_batch_size: 32
• num_train_epochs: 100
• warmup_steps: 100
• load_best_model_at_end: True

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: False
• eval_strategy: steps
• 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: 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: 1.0
• num_train_epochs: 100
• max_steps: -1
• lr_scheduler_type: linear
• lr_scheduler_kwargs: {}
• warmup_ratio: 0.0
• warmup_steps: 100
• 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
• 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: None
• hub_always_push: False
• hub_revision: None
• gradient_checkpointing: False
• gradient_checkpointing_kwargs: None
• include_inputs_for_metrics: False
• include_for_metrics: []
• 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
• 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
• use_liger_kernel: False
• liger_kernel_config: None
• eval_use_gather_object: False
• average_tokens_across_devices: False
• prompts: None
• batch_sampler: batch_sampler
• multi_dataset_batch_sampler: proportional
• router_mapping: {}
• learning_rate_mapping: {}

Training Logs

Epoch	Step	Training Loss	Validation Loss	val-sim_cosine_ap
11.3636	500	4.4432	5.65	0.5975
22.7273	1000	4.4873	5.9314	0.6090
34.0909	1500	4.3507	6.3564	0.6071
45.4545	2000	4.2365	6.3077	0.6007
56.8182	2500	4.076	6.2073	0.6116
68.1818	3000	3.9436	6.1120	0.6063

• The bold row denotes the saved checkpoint.

Framework Versions

• Python: 3.9.23
• Sentence Transformers: 5.0.0
• Transformers: 4.53.3
• PyTorch: 2.5.0+cu118
• Accelerate: 1.9.0
• Datasets: 4.0.0
• Tokenizers: 0.21.2

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",
}

BatchAllTripletLoss

@misc{hermans2017defense,
    title={In Defense of the Triplet Loss for Person Re-Identification},
    author={Alexander Hermans and Lucas Beyer and Bastian Leibe},
    year={2017},
    eprint={1703.07737},
    archivePrefix={arXiv},
    primaryClass={cs.CV}
}