Lampistero

This is a sentence-transformers model finetuned from jinaai/jina-embeddings-v3 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: jinaai/jina-embeddings-v3
Maximum Sequence Length: 8194 tokens
Output Dimensionality: 1024 dimensions
Similarity Function: Cosine Similarity
Training Dataset:
- json
Language: es
License: apache-2.0

Model Sources

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

Full Model Architecture

SentenceTransformer(
  (transformer): Transformer(
    (auto_model): XLMRobertaLoRA(
      (roberta): XLMRobertaModel(
        (embeddings): XLMRobertaEmbeddings(
          (word_embeddings): ParametrizedEmbedding(
            250002, 1024, padding_idx=1
            (parametrizations): ModuleDict(
              (weight): ParametrizationList(
                (0): LoRAParametrization()
              )
            )
          )
          (token_type_embeddings): ParametrizedEmbedding(
            1, 1024
            (parametrizations): ModuleDict(
              (weight): ParametrizationList(
                (0): LoRAParametrization()
              )
            )
          )
        )
        (emb_drop): Dropout(p=0.1, inplace=False)
        (emb_ln): LayerNorm((1024,), eps=1e-05, elementwise_affine=True)
        (encoder): XLMRobertaEncoder(
          (layers): ModuleList(
            (0-23): 24 x Block(
              (mixer): MHA(
                (rotary_emb): RotaryEmbedding()
                (Wqkv): ParametrizedLinearResidual(
                  in_features=1024, out_features=3072, bias=True
                  (parametrizations): ModuleDict(
                    (weight): ParametrizationList(
                      (0): LoRAParametrization()
                    )
                  )
                )
                (inner_attn): FlashSelfAttention(
                  (drop): Dropout(p=0.1, inplace=False)
                )
                (inner_cross_attn): FlashCrossAttention(
                  (drop): Dropout(p=0.1, inplace=False)
                )
                (out_proj): ParametrizedLinear(
                  in_features=1024, out_features=1024, bias=True
                  (parametrizations): ModuleDict(
                    (weight): ParametrizationList(
                      (0): LoRAParametrization()
                    )
                  )
                )
              )
              (dropout1): Dropout(p=0.1, inplace=False)
              (drop_path1): StochasticDepth(p=0.0, mode=row)
              (norm1): LayerNorm((1024,), eps=1e-05, elementwise_affine=True)
              (mlp): Mlp(
                (fc1): ParametrizedLinear(
                  in_features=1024, out_features=4096, bias=True
                  (parametrizations): ModuleDict(
                    (weight): ParametrizationList(
                      (0): LoRAParametrization()
                    )
                  )
                )
                (fc2): ParametrizedLinear(
                  in_features=4096, out_features=1024, bias=True
                  (parametrizations): ModuleDict(
                    (weight): ParametrizationList(
                      (0): LoRAParametrization()
                    )
                  )
                )
              )
              (dropout2): Dropout(p=0.1, inplace=False)
              (drop_path2): StochasticDepth(p=0.0, mode=row)
              (norm2): LayerNorm((1024,), eps=1e-05, elementwise_affine=True)
            )
          )
        )
        (pooler): XLMRobertaPooler(
          (dense): ParametrizedLinear(
            in_features=1024, out_features=1024, bias=True
            (parametrizations): ModuleDict(
              (weight): ParametrizationList(
                (0): LoRAParametrization()
              )
            )
          )
          (activation): Tanh()
        )
      )
    )
  )
  (pooler): Pooling({'word_embedding_dimension': 1024, '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})
  (normalizer): 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("csanz91/lampistero_rag_embeddings_2")
# Run inference
sentences = [
    "¿En qué año se demarcó y reconoció la mina 'El Pilar'?",
    "La mina 'El Pilar' se demarcó y reconoció en 1857.",
    'Según la quinta demanda del SOMM, todas compañías mineras debían entregar a todos sus obreros un libramiento de liquidación mensual',
]
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 with these parameters:
```
{
    "truncate_dim": 1024
}
```

Metric	Value
cosine_accuracy@1	0.7701
cosine_accuracy@3	0.8926
cosine_accuracy@5	0.9155
cosine_accuracy@10	0.933
cosine_precision@1	0.7701
cosine_precision@3	0.2975
cosine_precision@5	0.1831
cosine_precision@10	0.0933
cosine_recall@1	0.7701
cosine_recall@3	0.8926
cosine_recall@5	0.9155
cosine_recall@10	0.933
cosine_ndcg@10	0.8579
cosine_mrr@10	0.8331
cosine_map@100	0.8343

Information Retrieval

Dataset: dim_768
Evaluated with InformationRetrievalEvaluator with these parameters:
```
{
    "truncate_dim": 768
}
```

Metric	Value
cosine_accuracy@1	0.7695
cosine_accuracy@3	0.889
cosine_accuracy@5	0.9125
cosine_accuracy@10	0.933
cosine_precision@1	0.7695
cosine_precision@3	0.2963
cosine_precision@5	0.1825
cosine_precision@10	0.0933
cosine_recall@1	0.7695
cosine_recall@3	0.889
cosine_recall@5	0.9125
cosine_recall@10	0.933
cosine_ndcg@10	0.8571
cosine_mrr@10	0.8321
cosine_map@100	0.8333

Information Retrieval

Dataset: dim_512
Evaluated with InformationRetrievalEvaluator with these parameters:
```
{
    "truncate_dim": 512
}
```

Metric	Value
cosine_accuracy@1	0.7683
cosine_accuracy@3	0.8865
cosine_accuracy@5	0.9113
cosine_accuracy@10	0.9306
cosine_precision@1	0.7683
cosine_precision@3	0.2955
cosine_precision@5	0.1823
cosine_precision@10	0.0931
cosine_recall@1	0.7683
cosine_recall@3	0.8865
cosine_recall@5	0.9113
cosine_recall@10	0.9306
cosine_ndcg@10	0.8555
cosine_mrr@10	0.8307
cosine_map@100	0.8321

Information Retrieval

Dataset: dim_256
Evaluated with InformationRetrievalEvaluator with these parameters:
```
{
    "truncate_dim": 256
}
```

Metric	Value
cosine_accuracy@1	0.764
cosine_accuracy@3	0.8902
cosine_accuracy@5	0.9083
cosine_accuracy@10	0.93
cosine_precision@1	0.764
cosine_precision@3	0.2967
cosine_precision@5	0.1817
cosine_precision@10	0.093
cosine_recall@1	0.764
cosine_recall@3	0.8902
cosine_recall@5	0.9083
cosine_recall@10	0.93
cosine_ndcg@10	0.8535
cosine_mrr@10	0.8283
cosine_map@100	0.8296

Information Retrieval

Dataset: dim_128
Evaluated with InformationRetrievalEvaluator with these parameters:
```
{
    "truncate_dim": 128
}
```

Metric	Value
cosine_accuracy@1	0.7447
cosine_accuracy@3	0.8769
cosine_accuracy@5	0.9028
cosine_accuracy@10	0.9215
cosine_precision@1	0.7447
cosine_precision@3	0.2923
cosine_precision@5	0.1806
cosine_precision@10	0.0922
cosine_recall@1	0.7447
cosine_recall@3	0.8769
cosine_recall@5	0.9028
cosine_recall@10	0.9215
cosine_ndcg@10	0.8403
cosine_mrr@10	0.8134
cosine_map@100	0.8149

Information Retrieval

Dataset: dim_64
Evaluated with InformationRetrievalEvaluator with these parameters:
```
{
    "truncate_dim": 64
}
```

Metric	Value
cosine_accuracy@1	0.7103
cosine_accuracy@3	0.8491
cosine_accuracy@5	0.8781
cosine_accuracy@10	0.8998
cosine_precision@1	0.7103
cosine_precision@3	0.283
cosine_precision@5	0.1756
cosine_precision@10	0.09
cosine_recall@1	0.7103
cosine_recall@3	0.8491
cosine_recall@5	0.8781
cosine_recall@10	0.8998
cosine_ndcg@10	0.8119
cosine_mrr@10	0.7829
cosine_map@100	0.7851

Training Details

Training Dataset

json

Dataset: json
Size: 14,907 training samples
Columns: query and answer
Approximate statistics based on the first 1000 samples:
query answer
type string string
details
min: 9 tokens
mean: 26.09 tokens
max: 66 tokens

min: 4 tokens
mean: 34.02 tokens
max: 405 tokens

	query	answer
type	string	string
details	min: 9 tokens mean: 26.09 tokens max: 66 tokens	min: 4 tokens mean: 34.02 tokens max: 405 tokens

Samples:

query	answer
`¿Qué tipos de palas se utilizan para cargar el carbón y el mineral?`	`Se utiliza una pala convencional y una pala hidráulica, esta última descarga sobre un páncer, puede hacerlo lateralmente y se desplaza sobre ruedas u oruga.`
`Tras el cierre de la tejería de Florencio Salvador, ¿de dónde procedieron finalmente los ladrillos para las doscientas diez viviendas construidas en Utrillas?`	`Los ladrillos y material para las doscientas diez viviendas construidas en Utrillas procedieron finalmente de Letux, Zaragoza .`
`¿Cuál es el formato de los juegos infantiles que se están preparando para el verano en Escucha en 2021?`	`Los juegos infantiles que se están preparando para el verano en Escucha en 2021 están en formato revista.`

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: 64
per_device_eval_batch_size: 16
gradient_accumulation_steps: 32
learning_rate: 2e-05
num_train_epochs: 8
lr_scheduler_type: cosine
warmup_ratio: 0.1
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: 64
per_device_eval_batch_size: 16
per_gpu_train_batch_size: None
per_gpu_eval_batch_size: None
gradient_accumulation_steps: 32
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: 8
max_steps: -1
lr_scheduler_type: cosine
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
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: False
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}
tp_size: 0
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: None
hub_always_push: False
gradient_checkpointing: False
gradient_checkpointing_kwargs: None
include_inputs_for_metrics: False
include_for_metrics: []
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
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
use_liger_kernel: False
eval_use_gather_object: False
average_tokens_across_devices: False
prompts: None
batch_sampler: no_duplicates
multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	dim_1024_cosine_ndcg@10	dim_768_cosine_ndcg@10	dim_512_cosine_ndcg@10	dim_256_cosine_ndcg@10	dim_128_cosine_ndcg@10	dim_64_cosine_ndcg@10
1.0	8	-	0.7841	0.7835	0.7836	0.7791	0.7665	0.7226
1.2747	10	58.1187	-	-	-	-	-	-
2.0	16	-	0.8348	0.8366	0.8345	0.8301	0.8184	0.7861
2.5494	20	24.4181	-	-	-	-	-	-
3.0	24	-	0.8521	0.8504	0.8503	0.8457	0.8319	0.8007
3.8240	30	16.1488	-	-	-	-	-	-
4.0	32	-	0.8561	0.8548	0.8555	0.8509	0.8387	0.8073
5.0	40	13.4897	0.8585	0.8556	0.8545	0.8528	0.8397	0.8111
6.0	48	-	0.8578	0.8563	0.8550	0.8535	0.8410	0.8110
6.2747	50	13.7469	-	-	-	-	-	-
7.0	56	-	0.8579	0.8571	0.8555	0.8535	0.8403	0.8119

Framework Versions

Python: 3.12.10
Sentence Transformers: 4.1.0
Transformers: 4.51.3
PyTorch: 2.7.0+cu126
Accelerate: 1.7.0
Datasets: 3.6.0
Tokenizers: 0.21.1

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}
}

Downloads last month: 4

Safetensors

Model size

0.6B params

Tensor type

BF16

Model tree for csanz91/lampistero_rag_embeddings_2

Base model

jinaai/jina-embeddings-v3

Finetuned

(29)

this model

Evaluation results

Cosine Accuracy@1 on dim 1024
self-reported

0.770
Cosine Accuracy@3 on dim 1024
self-reported

0.893
Cosine Accuracy@5 on dim 1024
self-reported

0.916
Cosine Accuracy@10 on dim 1024
self-reported

0.933
Cosine Precision@1 on dim 1024
self-reported

0.770
Cosine Precision@3 on dim 1024
self-reported

0.298
Cosine Precision@5 on dim 1024
self-reported

0.183
Cosine Precision@10 on dim 1024
self-reported

0.093
Cosine Recall@1 on dim 1024
self-reported

0.770
Cosine Recall@3 on dim 1024
self-reported

0.893

View on Papers With Code