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.

python

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:
json
1{ 2 "loss_fct": "torch.nn.modules.loss.MSELoss" 3}

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

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",
}

	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%