metadata

language:
  - en
library_name: sentence-transformers
tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - loss:MSELoss
base_model: nreimers/TinyBERT_L-4_H-312_v2
metrics:
  - pearson_cosine
  - spearman_cosine
  - pearson_manhattan
  - spearman_manhattan
  - pearson_euclidean
  - spearman_euclidean
  - pearson_dot
  - spearman_dot
  - pearson_max
  - spearman_max
  - negative_mse
widget:
  - source_sentence: A woman at home.
    sentences:
      - The woman is inside.
      - The woman is performing for an audience.
      - The two men are freinds
  - source_sentence: boys play football
    sentences:
      - Rival college football players are playing a football game.
      - A man looks at his watch at a bus stop.
      - A woman walking on an old bridge near a mountain.
  - source_sentence: Nobody has a pot
    sentences:
      - Nobody has a suit
      - A woman riding a bicycle on the street.
      - The front is decorated with Ethiopian themes and motifs.
  - source_sentence: A dog plays ball.
    sentences:
      - A dog with a ball.
      - A man looking into a microscope in a lab
      - Children go past their parents.
  - source_sentence: A person standing
    sentences:
      - There is a person standing outside
      - A young man plays a racing video game.
      - Two children playing on the floor with toy trains.
pipeline_tag: sentence-similarity
co2_eq_emissions:
  emissions: 3.457859864142588
  energy_consumed: 0.00889591477312334
  source: codecarbon
  training_type: fine-tuning
  on_cloud: false
  cpu_model: 13th Gen Intel(R) Core(TM) i7-13700K
  ram_total_size: 31.777088165283203
  hours_used: 0.054
  hardware_used: 1 x NVIDIA GeForce RTX 3090
model-index:
  - name: SentenceTransformer based on nreimers/TinyBERT_L-4_H-312_v2
    results:
      - task:
          type: semantic-similarity
          name: Semantic Similarity
        dataset:
          name: sts dev
          type: sts-dev
        metrics:
          - type: pearson_cosine
            value: 0.8077673131159315
            name: Pearson Cosine
          - type: spearman_cosine
            value: 0.8208863013753134
            name: Spearman Cosine
          - type: pearson_manhattan
            value: 0.8225516575982812
            name: Pearson Manhattan
          - type: spearman_manhattan
            value: 0.8203236078973807
            name: Spearman Manhattan
          - type: pearson_euclidean
            value: 0.8215663439432439
            name: Pearson Euclidean
          - type: spearman_euclidean
            value: 0.8202318953605339
            name: Spearman Euclidean
          - type: pearson_dot
            value: 0.7901487535994149
            name: Pearson Dot
          - type: spearman_dot
            value: 0.7914362691291718
            name: Spearman Dot
          - type: pearson_max
            value: 0.8225516575982812
            name: Pearson Max
          - type: spearman_max
            value: 0.8208863013753134
            name: Spearman Max
      - task:
          type: knowledge-distillation
          name: Knowledge Distillation
        dataset:
          name: Unknown
          type: unknown
        metrics:
          - type: negative_mse
            value: -50.125449895858765
            name: Negative Mse
      - task:
          type: semantic-similarity
          name: Semantic Similarity
        dataset:
          name: sts test
          type: sts-test
        metrics:
          - type: pearson_cosine
            value: 0.7516961775809978
            name: Pearson Cosine
          - type: spearman_cosine
            value: 0.7558402072520215
            name: Spearman Cosine
          - type: pearson_manhattan
            value: 0.7762734499549059
            name: Pearson Manhattan
          - type: spearman_manhattan
            value: 0.75965556867712
            name: Spearman Manhattan
          - type: pearson_euclidean
            value: 0.7705568379382428
            name: Pearson Euclidean
          - type: spearman_euclidean
            value: 0.7553604477247078
            name: Spearman Euclidean
          - type: pearson_dot
            value: 0.7306801501272192
            name: Pearson Dot
          - type: spearman_dot
            value: 0.7097993872384684
            name: Spearman Dot
          - type: pearson_max
            value: 0.7762734499549059
            name: Pearson Max
          - type: spearman_max
            value: 0.75965556867712
            name: Spearman Max

SentenceTransformer based on nreimers/TinyBERT_L-4_H-312_v2

This is a sentence-transformers model finetuned from nreimers/TinyBERT_L-4_H-312_v2 on the sentence-transformers/wikipedia-en-sentences dataset. It maps sentences & paragraphs to a 312-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: nreimers/TinyBERT_L-4_H-312_v2
Maximum Sequence Length: 512 tokens
Output Dimensionality: 312 tokens
Similarity Function: Cosine Similarity
Training Dataset:
- sentence-transformers/wikipedia-en-sentences
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: BertModel 
  (1): Pooling({'word_embedding_dimension': 312, '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("tomaarsen/TinyBERT_L-4_H-312_v2-distilled-from-stsb-roberta-base-v2")
# Run inference
sentences = [
    'A person standing',
    'There is a person standing outside',
    'A young man plays a racing video game.',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 312]

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

Evaluation

Metrics

Semantic Similarity

Dataset: sts-dev
Evaluated with EmbeddingSimilarityEvaluator

Metric	Value
pearson_cosine	0.8078
spearman_cosine	0.8209
pearson_manhattan	0.8226
spearman_manhattan	0.8203
pearson_euclidean	0.8216
spearman_euclidean	0.8202
pearson_dot	0.7901
spearman_dot	0.7914
pearson_max	0.8226
spearman_max	0.8209

Knowledge Distillation

Evaluated with MSEEvaluator

Metric	Value
negative_mse	-50.1254

Semantic Similarity

Dataset: sts-test
Evaluated with EmbeddingSimilarityEvaluator

Metric	Value
pearson_cosine	0.7517
spearman_cosine	0.7558
pearson_manhattan	0.7763
spearman_manhattan	0.7597
pearson_euclidean	0.7706
spearman_euclidean	0.7554
pearson_dot	0.7307
spearman_dot	0.7098
pearson_max	0.7763
spearman_max	0.7597

Training Details

Training Dataset

sentence-transformers/wikipedia-en-sentences

Dataset: sentence-transformers/wikipedia-en-sentences at 4a0972d
Size: 200,000 training samples
Columns: sentence and label
Approximate statistics based on the first 1000 samples:
sentence label
type string list
details
min: 4 tokens
mean: 12.24 tokens
max: 52 tokens

size: 312 elements

	sentence	label
type	string	list
details	min: 4 tokens mean: 12.24 tokens max: 52 tokens	size: 312 elements

Samples:

sentence	label
`A person on a horse jumps over a broken down airplane.`	`[-0.09614687412977219, 0.6815224885940552, 2.702199935913086, 1.8371250629425049, -1.2949433326721191, ...]`
`Children smiling and waving at camera`	`[2.769360303878784, 3.074428081512451, -7.291755676269531, 5.248741149902344, 2.85081148147583, ...]`
`A boy is jumping on skateboard in the middle of a red bridge.`	`[-3.0669667720794678, 2.9899890422821045, -1.253997802734375, 6.15218448638916, 0.5838223099708557, ...]`

Loss: MSELoss

Evaluation Dataset

sentence-transformers/wikipedia-en-sentences

Dataset: sentence-transformers/wikipedia-en-sentences at 4a0972d
Size: 10,000 evaluation samples
Columns: sentence and label
Approximate statistics based on the first 1000 samples:
sentence label
type string list
details
min: 5 tokens
mean: 13.23 tokens
max: 57 tokens

size: 312 elements

	sentence	label
type	string	list
details	min: 5 tokens mean: 13.23 tokens max: 57 tokens	size: 312 elements

Samples:

sentence	label
`Two women are embracing while holding to go packages.`	`[6.200135707855225, -2.0865142345428467, -2.1313390731811523, -1.9593913555145264, -1.081985592842102, ...]`
`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.`	`[1.7725015878677368, 0.6873414516448975, -2.5191268920898438, 3.866339683532715, 2.853647470474243, ...]`
`A man selling donuts to a customer during a world exhibition event held in the city of Angeles`	`[-3.317653179168701, 3.0908589363098145, 0.1683920919895172, -2.4405274391174316, -3.1366524696350098, ...]`

Loss: MSELoss

Training Hyperparameters

Non-Default Hyperparameters

eval_strategy: steps
per_device_train_batch_size: 64
per_device_eval_batch_size: 64
learning_rate: 0.0001
num_train_epochs: 1
warmup_ratio: 0.1
fp16: True
load_best_model_at_end: True

All Hyperparameters

Click to expand

overwrite_output_dir: False
do_predict: False
eval_strategy: steps
prediction_loss_only: False
per_device_train_batch_size: 64
per_device_eval_batch_size: 64
per_gpu_train_batch_size: None
per_gpu_eval_batch_size: None
gradient_accumulation_steps: 1
eval_accumulation_steps: None
learning_rate: 0.0001
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
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: 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: None
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_sampler: batch_sampler
multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	loss	negative_mse	sts-dev_spearman_cosine	sts-test_spearman_cosine
0.032	100	0.8847	-	-	-	-
0.064	200	0.8136	-	-	-	-
0.096	300	0.697	-	-	-	-
0.128	400	0.6128	-	-	-	-
0.16	500	0.5634	0.6324	-63.2356	0.7564	-
0.192	600	0.5294	-	-	-	-
0.224	700	0.5035	-	-	-	-
0.256	800	0.4861	-	-	-	-
0.288	900	0.4668	-	-	-	-
0.32	1000	0.4515	0.5673	-56.7263	0.7965	-
0.352	1100	0.4376	-	-	-	-
0.384	1200	0.4274	-	-	-	-
0.416	1300	0.4178	-	-	-	-
0.448	1400	0.4098	-	-	-	-
0.48	1500	0.4053	0.5354	-53.5381	0.8091	-
0.512	1600	0.3934	-	-	-	-
0.544	1700	0.391	-	-	-	-
0.576	1800	0.3848	-	-	-	-
0.608	1900	0.3785	-	-	-	-
0.64	2000	0.3737	0.5168	-51.6829	0.8159	-
0.672	2100	0.3716	-	-	-	-
0.704	2200	0.3695	-	-	-	-
0.736	2300	0.3666	-	-	-	-
0.768	2400	0.3616	-	-	-	-
0.8	2500	0.358	0.5067	-50.6687	0.8189	-
0.832	2600	0.3551	-	-	-	-
0.864	2700	0.3544	-	-	-	-
0.896	2800	0.3524	-	-	-	-
0.928	2900	0.3524	-	-	-	-
0.96	3000	0.3529	0.5013	-50.1254	0.8209	-
0.992	3100	0.3496	-	-	-	-
1.0	3125	-	-	-	-	0.7558

The bold row denotes the saved checkpoint.

Environmental Impact

Carbon emissions were measured using CodeCarbon.

Energy Consumed: 0.009 kWh
Carbon Emitted: 0.003 kg of CO2
Hours Used: 0.054 hours

Training Hardware

On Cloud: No
GPU Model: 1 x NVIDIA GeForce RTX 3090
CPU Model: 13th Gen Intel(R) Core(TM) i7-13700K
RAM Size: 31.78 GB

Framework Versions

Python: 3.11.6
Sentence Transformers: 3.0.0.dev0
Transformers: 4.41.0.dev0
PyTorch: 2.3.0+cu121
Accelerate: 0.26.1
Datasets: 2.18.0
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",
}

MSELoss

@inproceedings{reimers-2020-multilingual-sentence-bert,
    title = "Making Monolingual Sentence Embeddings Multilingual using Knowledge Distillation",
    author = "Reimers, Nils and Gurevych, Iryna",
    booktitle = "Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing",
    month = "11",
    year = "2020",
    publisher = "Association for Computational Linguistics",
    url = "https://arxiv.org/abs/2004.09813",
}