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/sqv-v5-5ep")
# Run inference
sentences = [
    'Permet tramitar la baixa de les activitats esportives municipals.',
    'Quin és el procés per a donar de baixa una activitat esportiva?',
    'Quin és el benefici fiscal que es pot obtenir?',
]
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	0.1
cosine_accuracy@3	0.2261
cosine_accuracy@5	0.3043
cosine_accuracy@10	0.4957
cosine_precision@1	0.1
cosine_precision@3	0.0754
cosine_precision@5	0.0609
cosine_precision@10	0.0496
cosine_recall@1	0.1
cosine_recall@3	0.2261
cosine_recall@5	0.3043
cosine_recall@10	0.4957
cosine_ndcg@10	0.2645
cosine_mrr@10	0.1949
cosine_map@100	0.2142

Information Retrieval

Dataset: dim_768
Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.1
cosine_accuracy@3	0.213
cosine_accuracy@5	0.3
cosine_accuracy@10	0.4913
cosine_precision@1	0.1
cosine_precision@3	0.071
cosine_precision@5	0.06
cosine_precision@10	0.0491
cosine_recall@1	0.1
cosine_recall@3	0.213
cosine_recall@5	0.3
cosine_recall@10	0.4913
cosine_ndcg@10	0.2612
cosine_mrr@10	0.1922
cosine_map@100	0.2117

Information Retrieval

Dataset: dim_512
Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.0957
cosine_accuracy@3	0.2522
cosine_accuracy@5	0.3217
cosine_accuracy@10	0.5043
cosine_precision@1	0.0957
cosine_precision@3	0.0841
cosine_precision@5	0.0643
cosine_precision@10	0.0504
cosine_recall@1	0.0957
cosine_recall@3	0.2522
cosine_recall@5	0.3217
cosine_recall@10	0.5043
cosine_ndcg@10	0.2737
cosine_mrr@10	0.2033
cosine_map@100	0.2225

Information Retrieval

Dataset: dim_256
Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.0913
cosine_accuracy@3	0.2435
cosine_accuracy@5	0.3261
cosine_accuracy@10	0.4783
cosine_precision@1	0.0913
cosine_precision@3	0.0812
cosine_precision@5	0.0652
cosine_precision@10	0.0478
cosine_recall@1	0.0913
cosine_recall@3	0.2435
cosine_recall@5	0.3261
cosine_recall@10	0.4783
cosine_ndcg@10	0.2584
cosine_mrr@10	0.1911
cosine_map@100	0.2126

Information Retrieval

Dataset: dim_128
Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.0957
cosine_accuracy@3	0.2217
cosine_accuracy@5	0.3261
cosine_accuracy@10	0.513
cosine_precision@1	0.0957
cosine_precision@3	0.0739
cosine_precision@5	0.0652
cosine_precision@10	0.0513
cosine_recall@1	0.0957
cosine_recall@3	0.2217
cosine_recall@5	0.3261
cosine_recall@10	0.513
cosine_ndcg@10	0.2704
cosine_mrr@10	0.1969
cosine_map@100	0.2158

Information Retrieval

Dataset: dim_64
Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.1043
cosine_accuracy@3	0.2348
cosine_accuracy@5	0.3217
cosine_accuracy@10	0.4913
cosine_precision@1	0.1043
cosine_precision@3	0.0783
cosine_precision@5	0.0643
cosine_precision@10	0.0491
cosine_recall@1	0.1043
cosine_recall@3	0.2348
cosine_recall@5	0.3217
cosine_recall@10	0.4913
cosine_ndcg@10	0.2687
cosine_mrr@10	0.201
cosine_map@100	0.2206

Training Details

Training Dataset

json

Dataset: json
Size: 5,520 training samples
Columns: positive and anchor
Approximate statistics based on the first 1000 samples:
positive anchor
type string string
details
min: 5 tokens
mean: 43.7 tokens
max: 117 tokens

min: 9 tokens
mean: 20.51 tokens
max: 51 tokens

	positive	anchor
type	string	string
details	min: 5 tokens mean: 43.7 tokens max: 117 tokens	min: 9 tokens mean: 20.51 tokens max: 51 tokens

Samples:

positive	anchor
`L’Ajuntament vol crear un banc de recursos on recollir tots els oferiments de la població i que servirà per atendre les necessitats de les famílies refugiades acollides al poble.`	`Quin és el paper de l’Ajuntament en la integració de les persones refugiades acollides?`
`Aquest tipus d'actuació requereix la intervenció d'una persona tècnica competent que subscrigui el projecte o la documentació tècnica corresponent i que assumeixi la direcció facultativa de l'execució de les obres.`	`Quin és el requisit per a la intervenció d'una persona tècnica competent en les obres d'intervenció parcial interior en edificis amb elements catalogats?`
`Aquest títol, adreçat a persones empadronades a Sant Quirze del Vallès, es concedirà segons el nivell d’ingressos, la condició d’edat o de discapacitat, en base als criteris específics que recull l’ordenança reguladora del sistema de tarifació social del transport públic municipal en autobús a Sant Quirze del Vallès.`	`Quin és el benefici de la TBUS GRATUÏTA per a les persones majors?`

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
0.4638	10	4.122	-	-	-	-	-	-
0.9275	20	2.7131	-	-	-	-	-	-
0.9739	21	-	0.2085	0.1973	0.1884	0.2087	0.1886	0.2177
1.3913	30	1.6964	-	-	-	-	-	-
1.8551	40	1.2311	-	-	-	-	-	-
1.9942	43	-	0.2148	0.2135	0.2170	0.2351	0.2091	0.2386
2.3188	50	0.9216	-	-	-	-	-	-
2.7826	60	0.737	-	-	-	-	-	-
2.9681	64	-	0.2145	0.2058	0.2072	0.2277	0.2127	0.2085
3.2464	70	0.6678	-	-	-	-	-	-
3.7101	80	0.555	-	-	-	-	-	-
3.9884	86	-	0.2028	0.2154	0.2117	0.2331	0.2113	0.2028
4.1739	90	0.5542	-	-	-	-	-	-
4.6377	100	0.5058	-	-	-	-	-	-
4.8696	105	-	0.2142	0.2158	0.2126	0.2225	0.2206	0.2117

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-SQV-005-5ep

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-SQV-005-5ep

Evaluation results