metadata

language:
  - en
library_name: sentence-transformers
tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - dataset_size:100K<n<1M
  - loss:MultipleNegativesRankingLoss
base_model: FacebookAI/xlm-roberta-base
metrics:
  - cosine_accuracy
  - dot_accuracy
  - manhattan_accuracy
  - euclidean_accuracy
  - max_accuracy
widget:
  - source_sentence: a baby smiling
    sentences:
      - The boy is smiling
      - He is playing a song.
      - A man is sleeping.
  - source_sentence: an eagle flies
    sentences:
      - The person is amused.
      - There is a land race.
      - A man is sleeping.
  - source_sentence: There's a dock
    sentences:
      - The animal is outdoors.
      - The biker is a human.
      - A man is sleeping.
  - source_sentence: A woman sings.
    sentences:
      - The woman is singing.
      - Girls dance together.
      - A man playing ice hockey.
  - source_sentence: The boy scowls
    sentences:
      - The boy is outside.
      - The person is inside.
      - two men stand alone
pipeline_tag: sentence-similarity
model-index:
  - name: SentenceTransformer based on FacebookAI/xlm-roberta-base
    results:
      - task:
          type: triplet
          name: Triplet
        dataset:
          name: all nli dev
          type: all-nli-dev
        metrics:
          - type: cosine_accuracy
            value: 0.935
            name: Cosine Accuracy
          - type: dot_accuracy
            value: 0.061
            name: Dot Accuracy
          - type: manhattan_accuracy
            value: 0.929
            name: Manhattan Accuracy
          - type: euclidean_accuracy
            value: 0.93
            name: Euclidean Accuracy
          - type: max_accuracy
            value: 0.935
            name: Max Accuracy

SentenceTransformer based on FacebookAI/xlm-roberta-base

This is a sentence-transformers model finetuned from FacebookAI/xlm-roberta-base on the sentence-transformers/all-nli dataset. It maps sentences & paragraphs to a 768-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/xlm-roberta-base
Maximum Sequence Length: 512 tokens
Output Dimensionality: 768 tokens
Similarity Function: Cosine Similarity
Training Dataset:
- sentence-transformers/all-nli
Language: en

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: XLMRobertaModel 
  (1): Pooling({'word_embedding_dimension': 768, '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 = [
    'The boy scowls',
    'The boy is outside.',
    'The person is inside.',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 768]

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

Evaluation

Metrics

Triplet

Dataset: all-nli-dev
Evaluated with TripletEvaluator

Metric	Value
cosine_accuracy	0.935
dot_accuracy	0.061
manhattan_accuracy	0.929
euclidean_accuracy	0.93
max_accuracy	0.935

Training Details

Training Dataset

sentence-transformers/all-nli

Dataset: sentence-transformers/all-nli at d482672
Size: 100,000 training samples
Columns: anchor, positive, and negative

Approximate statistics based on the first 1000 samples:

	anchor	positive	negative
type	string	string	string
details	min: 7 tokens mean: 10.9 tokens max: 52 tokens	min: 6 tokens mean: 13.62 tokens max: 42 tokens	min: 5 tokens mean: 14.76 tokens max: 55 tokens

Samples:

anchor	positive	negative
`A person on a horse jumps over a broken down airplane.`	`A person is outdoors, on a horse.`	`A person is at a diner, ordering an omelette.`
`Children smiling and waving at camera`	`There are children present`	`The kids are frowning`
`A boy is jumping on skateboard in the middle of a red bridge.`	`The boy does a skateboarding trick.`	`The boy skates down the sidewalk.`

Loss: MultipleNegativesRankingLoss with these parameters:

{
    "scale": 20.0,
    "similarity_fct": "cos_sim"
}

Evaluation Dataset

sentence-transformers/all-nli

Dataset: sentence-transformers/all-nli at d482672
Size: 1,000 evaluation samples
Columns: anchor, positive, and negative

Approximate statistics based on the first 1000 samples:

	anchor	positive	negative
type	string	string	string
details	min: 6 tokens mean: 20.31 tokens max: 83 tokens	min: 5 tokens mean: 10.71 tokens max: 35 tokens	min: 5 tokens mean: 11.39 tokens max: 32 tokens

Samples:

anchor	positive	negative
`Two women are embracing while holding to go packages.`	`Two woman are holding packages.`	`The men are fighting outside a deli.`
`Two young children in blue jerseys, one with the number 9 and one with the number 2 are standing on wooden steps in a bathroom and washing their hands in a sink.`	`Two kids in numbered jerseys wash their hands.`	`Two kids in jackets walk to school.`
`A man selling donuts to a customer during a world exhibition event held in the city of Angeles`	`A man selling donuts to a customer.`	`A woman drinks her coffee in a small cafe.`

Loss: MultipleNegativesRankingLoss with these parameters:

{
    "scale": 20.0,
    "similarity_fct": "cos_sim"
}

Training Hyperparameters

Non-Default Hyperparameters

eval_strategy: steps
per_device_train_batch_size: 16
per_device_eval_batch_size: 16
num_train_epochs: 1
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
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: 1
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: 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	all-nli-dev_max_accuracy
0	0	-	-	0.587
0.016	100	3.4547	2.2853	0.801
0.032	200	1.6761	1.3493	0.856
0.048	300	1.5528	1.4181	0.83
0.064	400	1.0069	1.3277	0.835
0.08	500	1.0611	1.4610	0.847
0.096	600	1.1424	1.7394	0.805
0.112	700	1.3545	1.4179	0.83
0.128	800	1.3587	1.6350	0.84
0.144	900	1.237	1.6794	0.801
0.16	1000	1.2029	1.6733	0.811
0.176	1100	1.2748	1.6360	0.818
0.192	1200	1.1433	1.7952	0.806
0.208	1300	1.0113	1.4315	0.817
0.224	1400	0.8216	1.6300	0.776
0.24	1500	1.3451	1.1566	0.856
0.256	1600	0.8745	1.2075	0.838
0.272	1700	0.9945	1.3296	0.831
0.288	1800	0.9827	1.3052	0.844
0.304	1900	0.974	1.1643	0.85
0.32	2000	0.7555	1.2738	0.869
0.336	2100	0.7176	1.3749	0.832
0.352	2200	0.834	1.0712	0.879
0.368	2300	1.0819	1.2763	0.849
0.384	2400	0.9515	1.1384	0.848
0.4	2500	0.7828	1.0879	0.861
0.416	2600	0.7268	0.9835	0.868
0.432	2700	0.9228	1.1840	0.851
0.448	2800	1.0017	1.1968	0.853
0.464	2900	0.9138	0.9931	0.869
0.48	3000	0.8498	0.9926	0.876
0.496	3100	0.9682	1.0004	0.866
0.512	3200	0.7227	0.8490	0.883
0.528	3300	0.7134	0.8215	0.884
0.544	3400	0.6645	0.8889	0.877
0.56	3500	0.7073	0.8374	0.888
0.576	3600	0.6679	0.7780	0.911
0.592	3700	0.6609	0.8129	0.896
0.608	3800	0.687	0.7216	0.913
0.624	3900	0.5725	0.7618	0.92
0.64	4000	0.87	0.7070	0.909
0.656	4100	1.0892	0.7424	0.901
0.672	4200	1.048	0.6750	0.909
0.688	4300	0.8571	0.6474	0.903
0.704	4400	0.7945	0.6095	0.911
0.72	4500	0.6717	0.5664	0.93
0.736	4600	0.8161	0.5479	0.919
0.752	4700	0.7917	0.6420	0.911
0.768	4800	0.7711	0.5856	0.916
0.784	4900	0.6441	0.5775	0.916
0.8	5000	0.7766	0.5785	0.922
0.816	5100	0.6009	0.5680	0.921
0.832	5200	0.6711	0.5487	0.921
0.848	5300	0.618	0.5450	0.926
0.864	5400	0.6702	0.5498	0.926
0.88	5500	0.7039	0.5192	0.927
0.896	5600	0.6114	0.5045	0.932
0.912	5700	0.7761	0.5033	0.934
0.928	5800	0.6248	0.5013	0.932
0.944	5900	0.8359	0.4976	0.93
0.96	6000	0.8764	0.4976	0.936
0.976	6100	0.763	0.4845	0.935

Framework Versions

Python: 3.9.10
Sentence Transformers: 3.0.0
Transformers: 4.41.2
PyTorch: 2.3.0+cu121
Accelerate: 0.26.1
Datasets: 2.16.1
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",
}

MultipleNegativesRankingLoss

@misc{henderson2017efficient,
    title={Efficient Natural Language Response Suggestion for Smart Reply}, 
    author={Matthew Henderson and Rami Al-Rfou and Brian Strope and Yun-hsuan Sung and Laszlo Lukacs and Ruiqi Guo and Sanjiv Kumar and Balint Miklos and Ray Kurzweil},
    year={2017},
    eprint={1705.00652},
    archivePrefix={arXiv},
    primaryClass={cs.CL}
}