This is a sentence-transformers model finetuned from BAAI/bge-m3 on the train 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:
- train
Language: hu
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(
  (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("karsar/bge-m3-hu")
# Run inference
sentences = [
    'Az emberek alszanak.',
    'Egy apa és a fia ölelgeti alvás közben.',
    'Egy csoport ember ül egy nyitott, térszerű területen, mögötte nagy bokrok és egy sor viktoriánus stílusú épület, melyek közül sokat a kép jobb oldalán lévő erős elmosódás tesz kivehetetlenné.',
]
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

Triplet

Dataset: all-nli-dev
Evaluated with TripletEvaluator

Metric	Value
cosine_accuracy	0.979
dot_accuracy	0.021
manhattan_accuracy	0.9804
euclidean_accuracy	0.979
max_accuracy	0.9804

Triplet

Dataset: all-nli-test
Evaluated with TripletEvaluator

Metric	Value
cosine_accuracy	0.979
dot_accuracy	0.021
manhattan_accuracy	0.9804
euclidean_accuracy	0.979
max_accuracy	0.9804

Training Details

Training Dataset

train

Dataset: train
Size: 200,000 training samples
Columns: anchor, positive, and negative
Approximate statistics based on the first 1000 samples:
anchor positive negative
type string string string
details
min: 7 tokens
mean: 11.73 tokens
max: 56 tokens
min: 6 tokens
mean: 15.24 tokens
max: 47 tokens
min: 7 tokens
mean: 16.07 tokens
max: 53 tokens

Samples:

anchor	positive	negative
`Egy lóháton ülő ember átugrik egy lerombolt repülőgép felett.`	`Egy ember a szabadban, lóháton.`	`Egy ember egy étteremben van, és omlettet rendel.`
`Gyerekek mosolyogva és integetett a kamera`	`Gyermekek vannak jelen`	`A gyerekek homlokot rántanak`
`Egy fiú ugrál a gördeszkát a közepén egy piros híd.`	`A fiú gördeszkás trükköt csinál.`	`A fiú korcsolyázik a járdán.`

Loss: MultipleNegativesRankingLoss with these parameters:
json
1{ 2 "scale": 20.0, 3 "similarity_fct": "cos_sim" 4}

Evaluation Dataset

train

Dataset: train
Size: 5,000 evaluation samples
Columns: anchor, positive, and negative
Approximate statistics based on the first 1000 samples:
anchor positive negative
type string string string
details
min: 7 tokens
mean: 11.73 tokens
max: 56 tokens
min: 6 tokens
mean: 15.24 tokens
max: 47 tokens
min: 7 tokens
mean: 16.07 tokens
max: 53 tokens

Samples:

anchor	positive	negative
`Egy lóháton ülő ember átugrik egy lerombolt repülőgép felett.`	`Egy ember a szabadban, lóháton.`	`Egy ember egy étteremben van, és omlettet rendel.`
`Gyerekek mosolyogva és integetett a kamera`	`Gyermekek vannak jelen`	`A gyerekek homlokot rántanak`
`Egy fiú ugrál a gördeszkát a közepén egy piros híd.`	`A fiú gördeszkás trükköt csinál.`	`A fiú korcsolyázik a járdán.`

Loss: MultipleNegativesRankingLoss with these parameters:
json
1{ 2 "scale": 20.0, 3 "similarity_fct": "cos_sim" 4}

Training Hyperparameters

Non-Default Hyperparameters

eval_strategy: steps
per_device_train_batch_size: 16
per_device_eval_batch_size: 16
num_train_epochs: 1
warmup_ratio: 0.1
bf16: True
batch_sampler: no_duplicates

All Hyperparameters

Click to expand

overwrite_output_dir: False
do_predict: False
eval_strategy: steps
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: 1
eval_accumulation_steps: None
torch_empty_cache_steps: None
learning_rate: 5e-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: 1
max_steps: -1
lr_scheduler_type: linear
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: True
fp16: False
fp16_opt_level: O1
half_precision_backend: auto
bf16_full_eval: False
fp16_full_eval: False
tf32: None
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: False
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
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

Click to expand

Epoch	Step	Training Loss	train loss	all-nli-dev_max_accuracy	all-nli-test_max_accuracy
0	0	-	-	0.7176	-
0.008	100	1.0753	-	-	-
0.016	200	0.7611	-	-	-
0.024	300	1.0113	-	-	-
0.032	400	0.6224	-	-	-
0.04	500	0.8465	0.6159	0.8938	-
0.048	600	0.7761	-	-	-
0.056	700	0.8738	-	-	-
0.064	800	0.9393	-	-	-
0.072	900	0.9743	-	-	-
0.08	1000	0.8445	0.4556	0.8916	-
0.088	1100	0.7237	-	-	-
0.096	1200	0.8064	-	-	-
0.104	1300	0.607	-	-	-
0.112	1400	0.7632	-	-	-
0.12	1500	0.7477	1.6880	0.6748	-
0.128	1600	1.018	-	-	-
0.136	1700	0.9046	-	-	-
0.144	1800	0.728	-	-	-
0.152	1900	0.7219	-	-	-
0.16	2000	0.632	0.6459	0.8622	-
0.168	2100	0.6067	-	-	-
0.176	2200	0.7267	-	-	-
0.184	2300	0.781	-	-	-
0.192	2400	0.662	-	-	-
0.2	2500	0.6192	1.0124	0.8328	-
0.208	2600	0.7943	-	-	-
0.216	2700	0.8762	-	-	-
0.224	2800	0.7913	-	-	-
0.232	2900	0.8049	-	-	-
0.24	3000	0.858	0.6378	0.8046	-
0.248	3100	0.679	-	-	-
0.256	3200	0.7213	-	-	-
0.264	3300	0.6028	-	-	-
0.272	3400	0.5778	-	-	-
0.28	3500	0.5434	0.6784	0.8496	-
0.288	3600	0.6726	-	-	-
0.296	3700	0.7347	-	-	-
0.304	3800	0.8413	-	-	-
0.312	3900	0.7993	-	-	-
0.32	4000	0.8899	0.7732	0.8092	-
0.328	4100	1.1505	-	-	-
0.336	4200	0.8871	-	-	-
0.344	4300	0.8423	-	-	-
0.352	4400	0.8288	-	-	-
0.36	4500	0.6728	0.6341	0.8436	-
0.368	4600	0.7534	-	-	-
0.376	4700	0.8276	-	-	-
0.384	4800	0.7677	-	-	-
0.392	4900	0.588	-	-	-
0.4	5000	0.7742	0.4389	0.8808	-
0.408	5100	0.6782	-	-	-
0.416	5200	0.6688	-	-	-
0.424	5300	0.5579	-	-	-
0.432	5400	0.6891	-	-	-
0.44	5500	0.5764	0.4192	0.902	-
0.448	5600	0.6152	-	-	-
0.456	5700	0.6864	-	-	-
0.464	5800	0.6429	-	-	-
0.472	5900	0.9379	-	-	-
0.48	6000	0.7607	0.4744	0.8736	-
0.488	6100	0.819	-	-	-
0.496	6200	0.6316	-	-	-
0.504	6300	0.8175	-	-	-
0.512	6400	0.8485	-	-	-
0.52	6500	0.5374	0.4860	0.916	-
0.528	6600	0.781	-	-	-
0.536	6700	0.7722	-	-	-
0.544	6800	0.7281	-	-	-
0.552	6900	0.8453	-	-	-
0.56	7000	0.8541	0.2612	0.9322	-
0.568	7100	0.9698	-	-	-
0.576	7200	0.7184	-	-	-
0.584	7300	0.699	-	-	-
0.592	7400	0.5574	-	-	-
0.6	7500	0.5374	0.1939	0.9472	-
0.608	7600	0.6485	-	-	-
0.616	7700	0.5177	-	-	-
0.624	7800	0.814	-	-	-
0.632	7900	0.6442	-	-	-
0.64	8000	0.5301	0.1192	0.9616	-
0.648	8100	0.4948	-	-	-
0.656	8200	0.426	-	-	-
0.664	8300	0.4781	-	-	-
0.672	8400	0.4188	-	-	-
0.68	8500	0.5695	0.1523	0.9492	-
0.688	8600	0.3895	-	-	-
0.696	8700	0.5041	-	-	-
0.704	8800	0.7599	-	-	-
0.712	8900	0.5893	-	-	-
0.72	9000	0.6678	0.1363	0.9588	-
0.728	9100	0.5917	-	-	-
0.736	9200	0.6201	-	-	-
0.744	9300	0.5072	-	-	-
0.752	9400	0.4233	-	-	-
0.76	9500	0.396	0.2490	0.937	-
0.768	9600	0.3699	-	-	-
0.776	9700	0.3734	-	-	-
0.784	9800	0.4145	-	-	-
0.792	9900	0.4422	-	-	-
0.8	10000	0.4427	0.1394	0.9634	-
0.808	10100	0.678	-	-	-
0.816	10200	0.6771	-	-	-
0.824	10300	0.8249	-	-	-
0.832	10400	0.5003	-	-	-
0.84	10500	0.5586	0.1006	0.9726	-
0.848	10600	0.4649	-	-	-
0.856	10700	0.5322	-	-	-
0.864	10800	0.4837	-	-	-
0.872	10900	0.5717	-	-	-
0.88	11000	0.4403	0.1009	0.9688	-
0.888	11100	0.5044	-	-	-
0.896	11200	0.4771	-	-	-
0.904	11300	0.4426	-	-	-
0.912	11400	0.3705	-	-	-
0.92	11500	0.4445	0.0992	0.978	-
0.928	11600	0.3707	-	-	-
0.936	11700	0.4322	-	-	-
0.944	11800	0.4619	-	-	-
0.952	11900	0.4772	-	-	-
0.96	12000	0.5756	0.0950	0.9804	-
0.968	12100	0.5649	-	-	-
0.976	12200	0.5037	-	-	-
0.984	12300	0.0317	-	-	-
0.992	12400	0.0001	-	-	-
1.0	12500	0.0001	0.0948	0.9804	0.9804

Framework Versions

Python: 3.11.8
Sentence Transformers: 3.1.1
Transformers: 4.44.0
PyTorch: 2.3.0.post101
Accelerate: 0.33.0
Datasets: 2.18.0
Tokenizers: 0.19.0

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",
}

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}
}

	anchor	positive	negative
type	string	string	string
details	min: 7 tokens mean: 11.73 tokens max: 56 tokens	min: 6 tokens mean: 15.24 tokens max: 47 tokens	min: 7 tokens mean: 16.07 tokens max: 53 tokens