pascalymb/hack_ai_embbedding_model

3ac0d41 verified 3 months ago

16.7 kB

tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - generated_from_trainer
  - dataset_size:16000
  - loss:CoSENTLoss
base_model: abdeljalilELmajjodi/model
widget:
  - source_sentence: A man and woman are walking in a restaurant that has signs in Chinese.
    sentences:
      - A newlywed couple is walking through a Chinese restaurant.
      - The woman is sitting on the ground.
      - they are playing basketball
  - source_sentence: Several Asian cooks in a kitchen wearing white Dress Shirts.
    sentences:
      - >-
        A group of women carry woven baskets and large red rugs as they walk
        down a street.
      - A group of people is observing aquatic life.
      - Severl cooks are in a kitchen.
  - source_sentence: An Asian fish market with fish being cut up for sale.
    sentences:
      - A man in a blue and white shirt buys fish.
      - Fish are being sold.
      - The only amish person to own a cellphone.
  - source_sentence: An Asian man pushing a wheelchair.
    sentences:
      - A surfer is in the water.
      - There are at least four people.
      - A girl sitting in a wheelchair waiting for a friend.
  - source_sentence: Seven people are wading in a natural pool.
    sentences:
      - there were five people
      - A man is standing alone.
      - The man is riding a unicycle.
pipeline_tag: sentence-similarity
library_name: sentence-transformers
metrics:
  - pearson_cosine
  - spearman_cosine
model-index:
  - name: SentenceTransformer based on abdeljalilELmajjodi/model
    results:
      - task:
          type: semantic-similarity
          name: Semantic Similarity
        dataset:
          name: pair score evaluator dev
          type: pair-score-evaluator-dev
        metrics:
          - type: pearson_cosine
            value: 0.6451319585193018
            name: Pearson Cosine
          - type: spearman_cosine
            value: 0.6628169159235177
            name: Spearman Cosine

SentenceTransformer based on abdeljalilELmajjodi/model

This is a sentence-transformers model finetuned from abdeljalilELmajjodi/model on the all-nli dataset. It maps sentences & paragraphs to a 1024-dimensional dense vector space and can be used for retrieval.

Model Details

Model Description

Model Type: Sentence Transformer
Base model: abdeljalilELmajjodi/model
Maximum Sequence Length: 512 tokens
Output Dimensionality: 1024 dimensions
Similarity Function: Cosine Similarity
Supported Modality: Text
Training Dataset:
- all-nli

Model Sources

Documentation: Sentence Transformers Documentation
Repository: Sentence Transformers on GitHub
Hugging Face: Sentence Transformers on Hugging Face

Full Model Architecture

SentenceTransformer(
  (0): Transformer({'transformer_task': 'feature-extraction', 'modality_config': {'text': {'method': 'forward', 'method_output_name': 'last_hidden_state'}}, 'module_output_name': 'token_embeddings', 'architecture': 'XLMRobertaModel'})
  (1): Pooling({'embedding_dimension': 1024, 'pooling_mode': 'mean', '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 = [
    'Seven people are wading in a natural pool.',
    'there were five people',
    'The man is riding a unicycle.',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 1024]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities)
# tensor([[1.0000, 0.7814, 0.6510],
#         [0.7814, 1.0000, 0.7047],
#         [0.6510, 0.7047, 1.0000]])

Evaluation

Metrics

Semantic Similarity

Dataset: pair-score-evaluator-dev
Evaluated with EmbeddingSimilarityEvaluator

Metric	Value
pearson_cosine	0.6451
spearman_cosine	0.6628

Training Details

Training Dataset

all-nli

Dataset: all-nli
Size: 16,000 training samples
Columns: sentence1, sentence2, and score
Approximate statistics based on the first 1000 samples:
sentence1 sentence2 score
type string string float
details
min: 7 tokens
mean: 18.91 tokens
max: 66 tokens

min: 4 tokens
mean: 11.42 tokens
max: 36 tokens

min: 0.0
mean: 0.53
max: 1.0

	sentence1	sentence2	score
type	string	string	float
details	min: 7 tokens mean: 18.91 tokens max: 66 tokens	min: 4 tokens mean: 11.42 tokens max: 36 tokens	min: 0.0 mean: 0.53 max: 1.0

Samples:

sentence1	sentence2	score
`A man in a white shirt is on a rooftop lifting a board.`	`A man has a white shirt.`	`1.0`
`Two men in shorts and sandals are carrying food and drinks at a farmers market.`	`Men shopping their local market.`	`1.0`
`A woman with her head down is in a very run down area.`	`the man in a suit talks on his cellphone`	`0.0`

Loss: CoSENTLoss with these parameters:

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

Evaluation Dataset

all-nli

Dataset: all-nli
Size: 4,000 evaluation samples
Columns: sentence1, sentence2, and score
Approximate statistics based on the first 1000 samples:
sentence1 sentence2 score
type string string float
details
min: 7 tokens
mean: 19.28 tokens
max: 70 tokens

min: 5 tokens
mean: 11.46 tokens
max: 46 tokens

min: 0.0
mean: 0.49
max: 1.0

	sentence1	sentence2	score
type	string	string	float
details	min: 7 tokens mean: 19.28 tokens max: 70 tokens	min: 5 tokens mean: 11.46 tokens max: 46 tokens	min: 0.0 mean: 0.49 max: 1.0

Samples:

sentence1	sentence2	score
`A person dressed in natural clothing taking their picture in a mirror.`	`the person is nude at the bay`	`0.0`
`A basketball team of 8 girls is doing a hand huddle.`	`An all girls basketball team gets ready to start a game.`	`1.0`
`A person in a tan and blue sweater hanging clothes on a clothesline outside the window of her building.`	`There are no clothes.`	`0.0`

Loss: CoSENTLoss with these parameters:

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

Training Hyperparameters

Non-Default Hyperparameters

per_device_eval_batch_size: 16
gradient_accumulation_steps: 4
learning_rate: 2e-05
num_train_epochs: 4
warmup_steps: 0.05
bf16: True
bf16_full_eval: True
dataloader_num_workers: 4
load_best_model_at_end: True
push_to_hub: True
gradient_checkpointing: True

All Hyperparameters

Click to expand

do_predict: False
prediction_loss_only: True
per_device_train_batch_size: 8
per_device_eval_batch_size: 16
gradient_accumulation_steps: 4
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: 4
max_steps: -1
lr_scheduler_type: linear
lr_scheduler_kwargs: None
warmup_ratio: None
warmup_steps: 0.05
log_level: passive
log_level_replica: warning
log_on_each_node: True
logging_nan_inf_filter: True
enable_jit_checkpoint: False
save_on_each_node: False
save_only_model: False
restore_callback_states_from_checkpoint: False
use_cpu: False
seed: 42
data_seed: None
bf16: True
fp16: False
bf16_full_eval: True
fp16_full_eval: False
tf32: None
local_rank: -1
ddp_backend: None
debug: []
dataloader_drop_last: False
dataloader_num_workers: 4
dataloader_prefetch_factor: None
disable_tqdm: False
remove_unused_columns: True
label_names: None
load_best_model_at_end: True
ignore_data_skip: False
fsdp: []
fsdp_config: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}
accelerator_config: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
parallelism_config: None
deepspeed: None
label_smoothing_factor: 0.0
optim: adamw_torch_fused
optim_args: None
group_by_length: False
length_column_name: length
project: huggingface
trackio_space_id: trackio
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
push_to_hub: True
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: True
gradient_checkpointing_kwargs: None
include_for_metrics: []
eval_do_concat_batches: True
auto_find_batch_size: False
full_determinism: False
ddp_timeout: 1800
torch_compile: False
torch_compile_backend: None
torch_compile_mode: None
include_num_input_tokens_seen: no
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: True
use_cache: 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	pair-score-evaluator-dev_spearman_cosine
-1	-1	-	-	0.1511
0.002	1	3.0124	-	-
0.4	200	2.8949	-	-
0.8	400	2.7600	-	-
1.2	600	2.6245	-	-
1.6	800	2.5053	-	-
2.0	1000	2.4926	-	-
2.4	1200	2.2290	-	-
2.8	1400	2.1972	-	-
3.2	1600	2.0796	-	-
3.6	1800	1.9814	-	-
4.0	2000	1.9559	4.461	0.6628
-1	-1	-	-	0.6628

The bold row denotes the saved checkpoint.

Training Time

Training: 1.6 hours

Framework Versions

Python: 3.12.13
Sentence Transformers: 5.4.1
Transformers: 5.0.0
PyTorch: 2.10.0+cu128
Accelerate: 1.13.0
Datasets: 4.8.5
Tokenizers: 0.22.2

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",
}

CoSENTLoss

@article{10531646,
    author={Huang, Xiang and Peng, Hao and Zou, Dongcheng and Liu, Zhiwei and Li, Jianxin and Liu, Kay and Wu, Jia and Su, Jianlin and Yu, Philip S.},
    journal={IEEE/ACM Transactions on Audio, Speech, and Language Processing},
    title={CoSENT: Consistent Sentence Embedding via Similarity Ranking},
    year={2024},
    doi={10.1109/TASLP.2024.3402087}
}