This is a sentence-transformers model finetuned from DeepChem/ChemBERTa-77M-MLM. 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: DeepChem/ChemBERTa-77M-MLM
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}) with Transformer model: 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.

python

from sentence_transformers import SentenceTransformer

# Download from the 🤗 Hub
model = SentenceTransformer("HassanCS/chemBERTa-tuned-on-ClinTox-using-triplet-loss")
# Run inference
sentences = [
    'CON=C(C(=O)NC1C(=O)N2C(C(=O)[O-])=C(C[N+]3(C)CCCC3)CSC12)c1csc(N)n1',
    'CC1CNc2c(cccc2S(=O)(=O)NC(CCC[NH+]=C(N)N)C(=O)N2CCC(C)CC2C(=O)[O-])C1',
    'CC(C)C1(C(=O)NC2CC(=O)OC2(O)CF)CC(c2nccc3ccccc23)=NO1',
]
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

Triplet

Dataset: all-dev
Evaluated with TripletEvaluator

Metric	Value
cosine_accuracy	0.7135

Training Details

Training Dataset

Unnamed Dataset

Size: 118,400 training samples
Columns: anchor, positive, and negative
Approximate statistics based on the first 1000 samples:
anchor positive negative
type string string string
details
min: 20 tokens
mean: 33.0 tokens
max: 60 tokens
min: 3 tokens
mean: 47.34 tokens
max: 212 tokens
min: 3 tokens
mean: 53.88 tokens
max: 212 tokens

Samples:

anchor	positive	negative
`CC(C)CC(NC(=O)CNC(=O)c1cc(Cl)ccc1Cl)B(O)O`	`CC(=O)OC1CCC2(C)C(=CCC3C2CCC2(C)C(c4cccnc4)=CCC32)C1`	`CCOC(=O)c1ncn2c1CN(C)C(=O)c1cc(F)ccc1-2`
`CC(C)CC(NC(=O)CNC(=O)c1cc(Cl)ccc1Cl)B(O)O`	`COc1ccc(C(CN(C)C)C2(O)CCCCC2)cc1`	`C[NH2+]C1(C)C2CCC(C2)C1(C)C`
`CC(C)CC(NC(=O)CNC(=O)c1cc(Cl)ccc1Cl)B(O)O`	`CNC(=O)c1cc(Oc2ccc(NC(=O)Nc3ccc(Cl)c(C(F)(F)F)c3)cc2)ccn1.Cc1ccc(S(=O)(=O)O)cc1`	`Nc1ncnc2c1ncn2C1OC(CO)C(O)C1O`

Loss: TripletLoss with these parameters:
json
1{ 2 "distance_metric": "TripletDistanceMetric.EUCLIDEAN", 3 "triplet_margin": 5 4}

Evaluation Dataset

Unnamed Dataset

Size: 1,480 evaluation samples
Columns: anchor, positive, and negative
Approximate statistics based on the first 1000 samples:
anchor positive negative
type string string string
details
min: 18 tokens
mean: 54.07 tokens
max: 169 tokens
min: 18 tokens
mean: 60.4 tokens
max: 244 tokens
min: 30 tokens
mean: 71.25 tokens
max: 141 tokens

Samples:

anchor	positive	negative
`CC(C)OC(=O)CCCC=CCC1C(O)CC(O)C1C=CC(O)COc1cccc(C(F)(F)F)c1`	`CC12CCCCCC(Cc3ccc(O)cc31)C2[NH3+]`	`CC(C)C(CN1CCC(C)(c2cccc(O)c2)C(C)C1)NC(=O)C1Cc2ccc(O)cc2CN1`
`CC(C)OC(=O)CCCC=CCC1C(O)CC(O)C1C=CC(O)COc1cccc(C(F)(F)F)c1`	`COc1cc2c(cc1OC)C1CC(=O)C(CC(C)C)C[NH+]1CC2`	`CC(C)C(CN1CCC(C)(c2cccc(O)c2)C(C)C1)NC(=O)C1Cc2ccc(O)cc2CN1`
`CC(C)OC(=O)CCCC=CCC1C(O)CC(O)C1C=CC(O)COc1cccc(C(F)(F)F)c1`	`C[object Object]CCC=C1c2ccccc2COc2ccc(CC(=O)[O-])cc21`	`CC(C)C1(C(=O)NC2CC(=O)OC2(O)CF)CC(c2nccc3ccccc23)=NO1`

Loss: TripletLoss with these parameters:
json
1{ 2 "distance_metric": "TripletDistanceMetric.EUCLIDEAN", 3 "triplet_margin": 5 4}

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: 5
warmup_ratio: 0.1
fp16: True
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: 5
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: True
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: None
hub_always_push: False
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
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
use_liger_kernel: False
eval_use_gather_object: False
average_tokens_across_devices: False
prompts: None
batch_sampler: no_duplicates
multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	Validation Loss	all-dev_cosine_accuracy
0.0676	500	5.0821	5.1737	0.4047
0.1351	1000	4.9869	5.1766	0.4230
0.2027	1500	4.5562	4.9102	0.5345
0.2703	2000	3.2364	4.3712	0.6534
0.3378	2500	2.0738	4.0704	0.6736
0.4054	3000	1.4239	4.0200	0.6635
0.4730	3500	1.1578	3.7202	0.6791
0.5405	4000	0.9669	3.7197	0.6831
0.6081	4500	0.714	3.8818	0.6547
0.6757	5000	0.5359	4.0987	0.6243
0.7432	5500	0.5663	3.8127	0.6500
0.8108	6000	0.4827	3.8346	0.6676
0.8784	6500	0.4758	3.8333	0.6507
0.9459	7000	0.4759	3.6872	0.6912
1.0135	7500	0.4651	3.7229	0.6831
1.0811	8000	0.4739	3.8041	0.6662
1.1486	8500	0.4458	3.8235	0.6703
1.2162	9000	0.4189	3.7957	0.6716
1.2838	9500	0.4504	3.7422	0.6784
1.3514	10000	0.413	3.7588	0.6770
1.4189	10500	0.3808	3.9750	0.6615
1.4865	11000	0.3853	3.7417	0.6953
1.5541	11500	0.379	3.7319	0.6993
1.6216	12000	0.429	3.5620	0.7209
1.6892	12500	0.3735	3.6900	0.7020
1.7568	13000	0.3908	3.8182	0.6932
1.8243	13500	0.3848	3.7228	0.7101
1.8919	14000	0.3777	3.6604	0.7149
1.9595	14500	0.3912	3.7849	0.6946
2.0269	15000	0.3282	3.8607	0.7014
2.0945	15500	0.3324	3.8573	0.6953
2.1620	16000	0.3852	3.9420	0.7000
2.2296	16500	0.3633	3.7928	0.7189
2.2972	17000	0.3493	3.8217	0.7216
2.3647	17500	0.3554	3.8546	0.6993
2.4323	18000	0.3363	3.7764	0.6993
2.4999	18500	0.377	3.8224	0.6959
2.5674	19000	0.3569	3.8376	0.7155
2.635	19500	0.3414	4.0017	0.7034
2.7026	20000	0.3567	3.7405	0.7135
2.7701	20500	0.3524	3.9446	0.7189
2.8377	21000	0.3347	3.8140	0.7169
2.9053	21500	0.3458	4.0700	0.7088
2.9728	22000	0.3632	3.7930	0.7081
3.0404	22500	0.3496	3.9884	0.7236
3.1080	23000	0.3426	3.7102	0.7155
3.1755	23500	0.3579	3.9201	0.7135
3.2431	24000	0.3553	4.2237	0.7270
3.3107	24500	0.345	3.8090	0.7189
3.3782	25000	0.3475	3.7802	0.7284
3.4458	25500	0.3326	3.7549	0.7250
3.5134	26000	0.3228	3.6717	0.7216
3.5809	26500	0.3311	3.8241	0.7155
3.6485	27000	0.3215	3.8151	0.7142
3.7161	27500	0.3534	3.8639	0.7149
3.7836	28000	0.3369	4.0947	0.7101
3.8512	28500	0.3229	4.0495	0.7101
3.9188	29000	0.3442	4.0408	0.7169
3.9864	29500	0.3059	3.9493	0.6959
4.0538	30000	0.3349	4.0431	0.7108
4.1214	30500	0.3266	4.0224	0.7189
4.1889	31000	0.3501	3.9502	0.7169
4.2565	31500	0.3676	3.8903	0.7196
4.3241	32000	0.3191	3.7994	0.7162
4.3916	32500	0.3317	3.7889	0.7182
4.4592	33000	0.3304	3.8661	0.7108
4.5268	33500	0.3332	3.8822	0.7115
4.5943	34000	0.3435	3.7945	0.7088
4.6619	34500	0.317	3.8721	0.7243
4.7295	35000	0.3038	3.8615	0.7209
4.7970	35500	0.3093	3.8360	0.7162
4.8646	36000	0.3309	3.8277	0.7155
4.9322	36500	0.3378	3.7988	0.7128
4.9997	37000	0.311	3.8015	0.7135

Framework Versions

Python: 3.11.11
Sentence Transformers: 3.4.1
Transformers: 4.47.1
PyTorch: 2.5.1+cu124
Accelerate: 1.2.1
Datasets: 3.2.0
Tokenizers: 0.21.0

Citation

BibTeX

Sentence Transformers

bibtex

@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",
}

TripletLoss

bibtex

@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}
}

	anchor	positive	negative
type	string	string	string
details	min: 20 tokens mean: 33.0 tokens max: 60 tokens	min: 3 tokens mean: 47.34 tokens max: 212 tokens	min: 3 tokens mean: 53.88 tokens max: 212 tokens