SentenceTransformer based on BAAI/bge-m3

This is a sentence-transformers model finetuned from BAAI/bge-m3 on the json dataset. 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: BAAI/bge-m3
Maximum Sequence Length: 8192 tokens
Output Dimensionality: 1024 tokens
Similarity Function: Cosine Similarity
Training Dataset:
- json

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': 8192, 'do_lower_case': False}) with Transformer model: XLMRobertaModel 
  (1): Pooling({'word_embedding_dimension': 1024, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
  (2): Normalize()
)

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("adriansanz/ST-tramits-sitges-006-5ep")
# Run inference
sentences = [
    'El Decret 97/2002, de 5 de març, regula la concessió de la targeta d’aparcament per a persones amb disminució i altres mesures adreçades a facilitar el desplaçament de les persones amb mobilitat reduïda.',
    "Quin és el benefici de la targeta d'aparcament per a les persones amb disminució?",
    'Quin és el paper de la Junta de Govern Local?',
]
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

Information Retrieval

Dataset: dim_1024
Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.1181
cosine_accuracy@3	0.2328
cosine_accuracy@5	0.3129
cosine_accuracy@10	0.4644
cosine_precision@1	0.1181
cosine_precision@3	0.0776
cosine_precision@5	0.0626
cosine_precision@10	0.0464
cosine_recall@1	0.1181
cosine_recall@3	0.2328
cosine_recall@5	0.3129
cosine_recall@10	0.4644
cosine_ndcg@10	0.2655
cosine_mrr@10	0.2053
cosine_map@100	0.226

Information Retrieval

Dataset: dim_768
Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.1158
cosine_accuracy@3	0.229
cosine_accuracy@5	0.3113
cosine_accuracy@10	0.4657
cosine_precision@1	0.1158
cosine_precision@3	0.0763
cosine_precision@5	0.0623
cosine_precision@10	0.0466
cosine_recall@1	0.1158
cosine_recall@3	0.229
cosine_recall@5	0.3113
cosine_recall@10	0.4657
cosine_ndcg@10	0.2641
cosine_mrr@10	0.2031
cosine_map@100	0.2236

Information Retrieval

Dataset: dim_512
Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.1191
cosine_accuracy@3	0.2328
cosine_accuracy@5	0.3176
cosine_accuracy@10	0.4658
cosine_precision@1	0.1191
cosine_precision@3	0.0776
cosine_precision@5	0.0635
cosine_precision@10	0.0466
cosine_recall@1	0.1191
cosine_recall@3	0.2328
cosine_recall@5	0.3176
cosine_recall@10	0.4658
cosine_ndcg@10	0.2667
cosine_mrr@10	0.2064
cosine_map@100	0.2267

Information Retrieval

Dataset: dim_256
Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.1153
cosine_accuracy@3	0.2266
cosine_accuracy@5	0.3086
cosine_accuracy@10	0.4567
cosine_precision@1	0.1153
cosine_precision@3	0.0755
cosine_precision@5	0.0617
cosine_precision@10	0.0457
cosine_recall@1	0.1153
cosine_recall@3	0.2266
cosine_recall@5	0.3086
cosine_recall@10	0.4567
cosine_ndcg@10	0.2604
cosine_mrr@10	0.201
cosine_map@100	0.2217

Information Retrieval

Dataset: dim_128
Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.1118
cosine_accuracy@3	0.2233
cosine_accuracy@5	0.3025
cosine_accuracy@10	0.4529
cosine_precision@1	0.1118
cosine_precision@3	0.0744
cosine_precision@5	0.0605
cosine_precision@10	0.0453
cosine_recall@1	0.1118
cosine_recall@3	0.2233
cosine_recall@5	0.3025
cosine_recall@10	0.4529
cosine_ndcg@10	0.2566
cosine_mrr@10	0.1972
cosine_map@100	0.2178

Information Retrieval

Dataset: dim_64
Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.1069
cosine_accuracy@3	0.2125
cosine_accuracy@5	0.2885
cosine_accuracy@10	0.4297
cosine_precision@1	0.1069
cosine_precision@3	0.0708
cosine_precision@5	0.0577
cosine_precision@10	0.043
cosine_recall@1	0.1069
cosine_recall@3	0.2125
cosine_recall@5	0.2885
cosine_recall@10	0.4297
cosine_ndcg@10	0.2438
cosine_mrr@10	0.1876
cosine_map@100	0.2081

Training Details

Training Dataset

json

Dataset: json
Size: 2,844 training samples
Columns: positive and anchor
Approximate statistics based on the first 1000 samples:
positive anchor
type string string
details
min: 3 tokens
mean: 49.45 tokens
max: 148 tokens

min: 10 tokens
mean: 20.94 tokens
max: 45 tokens

	positive	anchor
type	string	string
details	min: 3 tokens mean: 49.45 tokens max: 148 tokens	min: 10 tokens mean: 20.94 tokens max: 45 tokens

Samples:

positive	anchor
`L'Ajuntament de Sitges atorga subvencions per a projectes i activitats d'interès públic o social que tinguin per finalitat les activitats esportives federades, escolars o populars desenvolupades per les entitats esportives i esportistes del municipi de Sitges.`	`Quin és el benefici de les subvencions per a les entitats esportives?`
`Per a poder ser beneficiari d'una subvenció per a un projecte o activitat cultural, les entitats o associacions culturals de Sitges han de tenir una seu social a la ciutat de Sitges i estar inscrites en el Registre d'Entitats de la Generalitat de Catalunya.`	`Quin és el requisit per a poder ser beneficiari d'una subvenció per a un projecte o activitat cultural?`
`La cessió entre tercers, només es contempla en el cas de sepultures de construcció particular que hagin estat donades d'alta amb una anterioritat de 10 anys a la data de sol·licitud de la cessió.`	`Quin és el paper de la persona que, legalment hi tingui dret, en la cessió entre tercers?`

Loss: MatryoshkaLoss with these parameters:

{
    "loss": "MultipleNegativesRankingLoss",
    "matryoshka_dims": [
        1024,
        768,
        512,
        256,
        128,
        64
    ],
    "matryoshka_weights": [
        1,
        1,
        1,
        1,
        1,
        1
    ],
    "n_dims_per_step": -1
}

Training Hyperparameters

Non-Default Hyperparameters

eval_strategy: epoch
per_device_train_batch_size: 16
per_device_eval_batch_size: 16
gradient_accumulation_steps: 16
learning_rate: 2e-05
num_train_epochs: 5
lr_scheduler_type: cosine
warmup_ratio: 0.2
bf16: True
tf32: True
load_best_model_at_end: True
optim: adamw_torch_fused
batch_sampler: no_duplicates

All Hyperparameters

Click to expand

overwrite_output_dir: False
do_predict: False
eval_strategy: epoch
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: 16
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: cosine
lr_scheduler_kwargs: {}
warmup_ratio: 0.2
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: True
fp16: False
fp16_opt_level: O1
half_precision_backend: auto
bf16_full_eval: False
fp16_full_eval: False
tf32: True
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_fused
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: no_duplicates
multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	dim_1024_cosine_map@100	dim_128_cosine_map@100	dim_256_cosine_map@100	dim_512_cosine_map@100	dim_64_cosine_map@100	dim_768_cosine_map@100
0.8989	10	3.2114	-	-	-	-	-	-
0.9888	11	-	0.2144	0.2008	0.2070	0.2126	0.1842	0.2126
1.7978	20	1.5622	-	-	-	-	-	-
1.9775	22	-	0.2179	0.2101	0.2169	0.2180	0.2012	0.2193
2.6966	30	0.7882	-	-	-	-	-	-
2.9663	33	-	0.2239	0.2162	0.2220	0.2238	0.2070	0.2222
3.5955	40	0.4956	-	-	-	-	-	-
3.9551	44	-	0.2270	0.2177	0.2231	0.2278	0.2084	0.2255
4.4944	50	0.392	-	-	-	-	-	-
4.9438	55	-	0.226	0.2178	0.2217	0.2267	0.2081	0.2236

The bold row denotes the saved checkpoint.

Framework Versions

Python: 3.10.12
Sentence Transformers: 3.1.1
Transformers: 4.44.2
PyTorch: 2.4.1+cu121
Accelerate: 0.35.0.dev0
Datasets: 3.0.1
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",
}

MatryoshkaLoss

@misc{kusupati2024matryoshka,
    title={Matryoshka Representation Learning},
    author={Aditya Kusupati and Gantavya Bhatt and Aniket Rege and Matthew Wallingford and Aditya Sinha and Vivek Ramanujan and William Howard-Snyder and Kaifeng Chen and Sham Kakade and Prateek Jain and Ali Farhadi},
    year={2024},
    eprint={2205.13147},
    archivePrefix={arXiv},
    primaryClass={cs.LG}
}

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}
}

adriansanz
/

ST-tramits-sitges-006-5ep

SentenceTransformer based on BAAI/bge-m3

Model Details

Model Description

Model Sources

Full Model Architecture

Usage

Direct Usage (Sentence Transformers)

Evaluation

Metrics

Information Retrieval

Information Retrieval

Information Retrieval

Information Retrieval

Information Retrieval

Information Retrieval

Training Details

Training Dataset

json

Training Hyperparameters

Non-Default Hyperparameters

All Hyperparameters

Training Logs

Framework Versions

Citation

BibTeX

Sentence Transformers

MatryoshkaLoss

MultipleNegativesRankingLoss

Model tree for adriansanz/ST-tramits-sitges-006-5ep

Evaluation results