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.

python

from sentence_transformers import SentenceTransformer

# Download from the 🤗 Hub
model = SentenceTransformer("adriansanz/ST-tramits-SITGES-007-5ep")
# Run inference
sentences = [
    'La comunicació és un element important en la cura dels gats, ja que implica la capacitat per a comunicar-se de manera efectiva amb les autoritats competents i amb els altres implicats en la cura dels animals.',
    'Quin és el paper de la comunicació en la cura dels gats?',
    'Qui són considerats titulars o nous exercents en el cas dels espectacles, establiments oberts al públic i les activitats recreatives?',
]
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	0.159
cosine_accuracy@3	0.3033
cosine_accuracy@5	0.3724
cosine_accuracy@10	0.5188
cosine_precision@1	0.159
cosine_precision@3	0.1011
cosine_precision@5	0.0745
cosine_precision@10	0.0519
cosine_recall@1	0.159
cosine_recall@3	0.3033
cosine_recall@5	0.3724
cosine_recall@10	0.5188
cosine_ndcg@10	0.3174
cosine_mrr@10	0.256
cosine_map@100	0.2763

Information Retrieval

Dataset: dim_768
Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.1569
cosine_accuracy@3	0.2971
cosine_accuracy@5	0.3808
cosine_accuracy@10	0.5084
cosine_precision@1	0.1569
cosine_precision@3	0.099
cosine_precision@5	0.0762
cosine_precision@10	0.0508
cosine_recall@1	0.1569
cosine_recall@3	0.2971
cosine_recall@5	0.3808
cosine_recall@10	0.5084
cosine_ndcg@10	0.3139
cosine_mrr@10	0.2541
cosine_map@100	0.2757

Information Retrieval

Dataset: dim_512
Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.1736
cosine_accuracy@3	0.3138
cosine_accuracy@5	0.3954
cosine_accuracy@10	0.5377
cosine_precision@1	0.1736
cosine_precision@3	0.1046
cosine_precision@5	0.0791
cosine_precision@10	0.0538
cosine_recall@1	0.1736
cosine_recall@3	0.3138
cosine_recall@5	0.3954
cosine_recall@10	0.5377
cosine_ndcg@10	0.3324
cosine_mrr@10	0.27
cosine_map@100	0.2901

Information Retrieval

Dataset: dim_256
Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.1506
cosine_accuracy@3	0.2908
cosine_accuracy@5	0.4017
cosine_accuracy@10	0.5356
cosine_precision@1	0.1506
cosine_precision@3	0.0969
cosine_precision@5	0.0803
cosine_precision@10	0.0536
cosine_recall@1	0.1506
cosine_recall@3	0.2908
cosine_recall@5	0.4017
cosine_recall@10	0.5356
cosine_ndcg@10	0.319
cosine_mrr@10	0.2527
cosine_map@100	0.2729

Information Retrieval

Dataset: dim_128
Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.1674
cosine_accuracy@3	0.3201
cosine_accuracy@5	0.4163
cosine_accuracy@10	0.5481
cosine_precision@1	0.1674
cosine_precision@3	0.1067
cosine_precision@5	0.0833
cosine_precision@10	0.0548
cosine_recall@1	0.1674
cosine_recall@3	0.3201
cosine_recall@5	0.4163
cosine_recall@10	0.5481
cosine_ndcg@10	0.3354
cosine_mrr@10	0.27
cosine_map@100	0.2892

Information Retrieval

Dataset: dim_64
Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	0.1548
cosine_accuracy@3	0.2845
cosine_accuracy@5	0.3515
cosine_accuracy@10	0.5209
cosine_precision@1	0.1548
cosine_precision@3	0.0948
cosine_precision@5	0.0703
cosine_precision@10	0.0521
cosine_recall@1	0.1548
cosine_recall@3	0.2845
cosine_recall@5	0.3515
cosine_recall@10	0.5209
cosine_ndcg@10	0.3117
cosine_mrr@10	0.2482
cosine_map@100	0.2686

Training Details

Training Dataset

json

Dataset: json
Size: 6,692 training samples
Columns: positive and anchor
Approximate statistics based on the first 1000 samples:
positive anchor
type string string
details
min: 6 tokens
mean: 44.83 tokens
max: 185 tokens
min: 10 tokens
mean: 20.89 tokens
max: 49 tokens

Samples:

positive	anchor
`Els residus comercials o industrials assimilables als municipals que hauran d'acreditar si disposen d'un gestor autoritzat per a la gestió dels residus.`	`Quins són els residus que es recullen en el servei municipal complementari?`
`L'Ajuntament de Sitges ofereix ajuts econòmics a famílies amb recursos insuficients per accedir a la realització d'activitats de lleure...`	`Quin és el paper de l'Ajuntament de Sitges en la promoció de l'educació no formal i de lleure?`
`Permet comunicar les intervencions necessàries per executar una instal·lació/remodelació d’autoconsum amb energia solar fotovoltaica amb una potència instal·lada inferior a 100 kWp en sòl urbà consolidat.`	`Quin és el propòsit de la remodelació d'una instal·lació d'autoconsum?`

Loss: MatryoshkaLoss with these parameters:
json
1{ 2 "loss": "MultipleNegativesRankingLoss", 3 "matryoshka_dims": [ 4 1024, 5 768, 6 512, 7 256, 8 128, 9 64 10 ], 11 "matryoshka_weights": [ 12 1, 13 1, 14 1, 15 1, 16 1, 17 1 18 ], 19 "n_dims_per_step": -1 20}

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
0.3819	10	3.3449	-	-	-	-	-	-
0.7637	20	2.0557	-	-	-	-	-	-
0.9928	26	-	0.2440	0.2408	0.2590	0.2439	0.2379	0.2512
1.1456	30	1.4634	-	-	-	-	-	-
1.5274	40	0.8163	-	-	-	-	-	-
1.9093	50	0.6103	-	-	-	-	-	-
1.9857	52	-	0.2621	0.2683	0.2483	0.2629	0.2404	0.2472
2.2912	60	0.4854	-	-	-	-	-	-
2.6730	70	0.2796	-	-	-	-	-	-
2.9785	78	-	0.2701	0.2697	0.2761	0.2845	0.2673	0.2709
3.0549	80	0.2458	-	-	-	-	-	-
3.4368	90	0.2616	-	-	-	-	-	-
3.8186	100	0.174	-	-	-	-	-	-
3.9714	104	-	0.2729	0.2863	0.2858	0.2853	0.2656	0.2752
4.2005	110	0.1841	-	-	-	-	-	-
4.5823	120	0.1668	-	-	-	-	-	-
4.9642	130	0.1484	0.2763	0.2892	0.2729	0.2901	0.2686	0.2757

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

bibtex

@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

bibtex

@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

bibtex

@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}
}

	positive	anchor
type	string	string
details	min: 6 tokens mean: 44.83 tokens max: 185 tokens	min: 10 tokens mean: 20.89 tokens max: 49 tokens