SentenceTransformer based on BAAI/bge-m3

This is a sentence-transformers model finetuned from BAAI/bge-m3 on the mitrasamgraha 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:
- mitrasamgraha

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("sanganaka/bge-m3-sanskritFT")
# Run inference
sentences = [
    'O Śākyamuni, conquering the powerful host of Māra, You found peace, immortality, and the happiness of that supreme enlightenment',
    'मारस् त्वयास्तु विजितस् सबलो मुनीन्द्रः प्राप्ता शिवा अमृतशान्तवराग्रबोधिः ।',
    'न हि तथता द्वयप्रभाविता नानात्वप्रभाविता ।',
]
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

Translation

Dataset: translate-val
Evaluated with TranslationEvaluator

Metric	Value
src2trg_accuracy	0.9423
trg2src_accuracy	0.9381
mean_accuracy	0.9402

Translation

Dataset: translate-test
Evaluated with TranslationEvaluator

Metric	Value
src2trg_accuracy	0.9305
trg2src_accuracy	0.927
mean_accuracy	0.9288

Training Details

Training Dataset

mitrasamgraha

Dataset: mitrasamgraha
Size: 477,170 training samples
Columns: english and sanskrit_Deva
Approximate statistics based on the first 1000 samples:
english sanskrit_Deva
type string string
details
min: 20 tokens
mean: 43.11 tokens
max: 90 tokens

min: 19 tokens
mean: 33.88 tokens
max: 78 tokens

	english	sanskrit_Deva
type	string	string
details	min: 20 tokens mean: 43.11 tokens max: 90 tokens	min: 19 tokens mean: 33.88 tokens max: 78 tokens

Samples:

english	sanskrit_Deva
`My patience is almost worn out, like that of a creeper under the winter frost. It is decayed, and neither lives nor perishes at once.`	`जर्जरीकृत्य वस्तूनि त्यजन्ती विभ्रती तथा । मार्गशीर्षान्तवल्लीव धृतिर्विधुरतां गता ॥`
`Our minds are partly settled in worldly things, and partly fixed in their giver (the Supreme soul). This divided state of the mind is termed its half waking condition.`	`अपहस्तितसर्वार्थमनवस्थितिरास्थिता । गृहीत्वोत्सृज्य चात्मानं भवस्थितिरवस्थिता ॥`
`My mind is in a state of suspense, being unable to ascertain the real nature of my soul. I am like one in the dark, who is deceived by the stump of a fallen tree at a distance, to think it a human figure.`	`चलिताचलितेनान्तरवष्टम्भेन मे मतिः । दरिद्रा छिन्नवृक्षस्य मूलेनेव विडम्ब्यते ॥`

Loss: MultipleNegativesRankingLoss with these parameters:

{
    "scale": 20.0,
    "similarity_fct": "cos_sim"
}

Evaluation Dataset

mitrasamgraha

Dataset: mitrasamgraha
Size: 5,560 evaluation samples
Columns: english and sanskrit_Deva
Approximate statistics based on the first 1000 samples:
english sanskrit_Deva
type string string
details
min: 5 tokens
mean: 58.68 tokens
max: 387 tokens

min: 6 tokens
mean: 44.88 tokens
max: 257 tokens

	english	sanskrit_Deva
type	string	string
details	min: 5 tokens mean: 58.68 tokens max: 387 tokens	min: 6 tokens mean: 44.88 tokens max: 257 tokens

Samples:

english	sanskrit_Deva
Thereupon he takes the winnowing basket and the Agnihotra ladle , with the text : 'For the work (I take) you, for pervasion (or accomplishment) you two! ' For the sacrifice is a work: hence, in saying 'for the work you two, ' he says, 'for the sacrifice. ' And 'for pervasion you two, ' he says, because he, as it were, pervades (goes through, accomplishes) the sacrifice. He then restrains his speech; for (restrained) speech means undisturbed sacrifice; so that (in so doing) he thinks: 'May I accomplish the sacrifice! ' He now heats (the two objects on the Grhapatya), with the formula : 'Scorched is the Rakshas, scorched are the enemies! ' or : 'Burnt out is the Rakshas, burnt out are the enemies! '	अग्ने व्रतपते व्रतं चरिष्यामि तचकेयं तन्मे राध्यतामित्यग्निर्वै देवानां व्रतपतिस्तस्मा एवैतत्प्राह व्रतं चरिष्यामि तच्चकेयं तन्मे राध्यतामिति नात्र तिरोहितमिवास्ति ॥ अथ संस्थिते विसृजते । अग्ने व्रतपते व्रतमचारिषं तादशकम् तन्मे राधीत्यशकद्येतद्यो यज्ञस्य संस्थामगन्नराधि ह्यस्मै यो यज्ञस्य संस्थामगन्नेतेन न्वेव भूयिष्ठा इव व्रतमुपयन्त्यनेन त्वेवोपेयात् ॥ द्वयं वा इदं न तृतीयमस्ति ।
`For the gods, when they were performing the sacrifice, were afraid of a disturbance on the part of the Asuras and Rakshas: hence by this means he expels from here, at the very opening of the sacrifice, the evil spirits, the Rakshas.`	`एतद्धवै देवा व्रतं चरन्ति यत्सत्यं तस्मात्ते यशो यशो ह भवति य एवं विद्वांत्सत्यंवदति ॥ अथ संस्थिते विसृजते ।`
`He now steps forward (to the cart ), with the text : 'I move along the wide arial realm. ' For the Rakshas roams about in the air, rootless and unfettered in both directions (below and above); and in order that this man (the Adhvaryu) may move about the air, rootless and unfettered in both directions, he by this very prayer renders the atmosphere free from danger and evil spirits.`	स वा आरण्यमेवाश्नीयात् । या वारण्या ओषधयो यद्वा वृक्ष्यं तदु ह स्माहापि बर्कुर्वार्ष्णो मासान्मे पचत न वा एतेसां हविर्गृह्णन्तीति तदु तथा न कुर्याद्व्रीहियवयोर्वा एतदुपजं यचमीधान्यं तद्व्रीहियवावेवैतेन भूयांसौ करोति तस्मादारण्यमेवाश्नीयात् ॥

Loss: MultipleNegativesRankingLoss with these parameters:

{
    "scale": 20.0,
    "similarity_fct": "cos_sim"
}

Training Hyperparameters

Non-Default Hyperparameters

eval_strategy: epoch
per_device_train_batch_size: 32
per_device_eval_batch_size: 128
learning_rate: 2e-05
num_train_epochs: 5
warmup_ratio: 0.1
bf16: True
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: 32
per_device_eval_batch_size: 128
per_gpu_train_batch_size: None
per_gpu_eval_batch_size: None
gradient_accumulation_steps: 1
eval_accumulation_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: 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
batch_sampler: no_duplicates
multi_dataset_batch_sampler: proportional

Training Logs

Click to expand

Epoch	Step	Training Loss	Validation Loss	translate-val_mean_accuracy	translate-test_mean_accuracy
0	0	-	-	0.3724	-
0.0335	500	1.7546	-	-	-
0.0671	1000	0.8427	-	-	-
0.1006	1500	0.301	-	-	-
0.1341	2000	0.1892	-	-	-
0.1677	2500	0.0985	-	-	-
0.2012	3000	0.283	-	-	-
0.2347	3500	0.4843	-	-	-
0.2682	4000	0.3884	-	-	-
0.3018	4500	0.5331	-	-	-
0.3353	5000	0.6926	-	-	-
0.3688	5500	0.4398	-	-	-
0.4024	6000	0.3152	-	-	-
0.4359	6500	0.2488	-	-	-
0.4694	7000	0.3297	-	-	-
0.5030	7500	0.2496	-	-	-
0.5365	8000	0.2058	-	-	-
0.5700	8500	0.2032	-	-	-
0.6035	9000	0.3762	-	-	-
0.6371	9500	0.4035	-	-	-
0.6706	10000	0.5922	-	-	-
0.7041	10500	0.0894	-	-	-
0.7377	11000	0.2658	-	-	-
0.7712	11500	0.2099	-	-	-
0.8047	12000	0.4648	-	-	-
0.8383	12500	0.5967	-	-	-
0.8718	13000	0.0863	-	-	-
0.9053	13500	0.0626	-	-	-
0.9388	14000	0.2336	-	-	-
0.9724	14500	0.3032	-	-	-
1.0	14912	-	0.2874	0.8858	-
1.0059	15000	0.2268	-	-	-
1.0394	15500	0.4782	-	-	-
1.0730	16000	0.2226	-	-	-
1.1065	16500	0.0766	-	-	-
1.1400	17000	0.0589	-	-	-
1.1736	17500	0.0248	-	-	-
1.2071	18000	0.1875	-	-	-
1.2406	18500	0.2958	-	-	-
1.2741	19000	0.2065	-	-	-
1.3077	19500	0.4541	-	-	-
1.3412	20000	0.5509	-	-	-
1.3747	20500	0.1221	-	-	-
1.4083	21000	0.1986	-	-	-
1.4418	21500	0.1263	-	-	-
1.4753	22000	0.1777	-	-	-
1.5089	22500	0.1165	-	-	-
1.5424	23000	0.1017	-	-	-
1.5759	23500	0.1309	-	-	-
1.6094	24000	0.2304	-	-	-
1.6430	24500	0.3245	-	-	-
1.6765	25000	0.3282	-	-	-
1.7100	25500	0.0163	-	-	-
1.7436	26000	0.1357	-	-	-
1.7771	26500	0.1302	-	-	-
1.8106	27000	0.4238	-	-	-
1.8442	27500	0.3066	-	-	-
1.8777	28000	0.0305	-	-	-
1.9112	28500	0.0279	-	-	-
1.9447	29000	0.1823	-	-	-
1.9783	29500	0.151	-	-	-
2.0	29824	-	0.2112	0.9160	-
2.0118	30000	0.169	-	-	-
2.0453	30500	0.2848	-	-	-
2.0789	31000	0.0858	-	-	-
2.1124	31500	0.0363	-	-	-
2.1459	32000	0.0208	-	-	-
2.1795	32500	0.01	-	-	-
2.2130	33000	0.1198	-	-	-
2.2465	33500	0.2025	-	-	-
2.2800	34000	0.1131	-	-	-
2.3136	34500	0.3647	-	-	-
2.3471	35000	0.3397	-	-	-
2.3806	35500	0.0507	-	-	-
2.4142	36000	0.1101	-	-	-
2.4477	36500	0.0832	-	-	-
2.4812	37000	0.0977	-	-	-
2.5148	37500	0.0666	-	-	-
2.5483	38000	0.0546	-	-	-
2.5818	38500	0.0868	-	-	-
2.6153	39000	0.1504	-	-	-
2.6489	39500	0.2462	-	-	-
2.6824	40000	0.1835	-	-	-
2.7159	40500	0.0279	-	-	-
2.7495	41000	0.0594	-	-	-
2.7830	41500	0.0889	-	-	-
2.8165	42000	0.4076	-	-	-
2.8501	42500	0.1206	-	-	-
2.8836	43000	0.0143	-	-	-
2.9171	43500	0.013	-	-	-
2.9506	44000	0.1479	-	-	-
2.9842	44500	0.0626	-	-	-
3.0	44736	-	0.1816	0.9262	-
3.0177	45000	0.1422	-	-	-
3.0512	45500	0.1636	-	-	-
3.0848	46000	0.0266	-	-	-
3.1183	46500	0.0145	-	-	-
3.1518	47000	0.0096	-	-	-
3.1854	47500	0.0055	-	-	-
3.2189	48000	0.0728	-	-	-
3.2524	48500	0.1368	-	-	-
3.2859	49000	0.0739	-	-	-
3.3195	49500	0.2677	-	-	-
3.3530	50000	0.2339	-	-	-
3.3865	50500	0.0283	-	-	-
3.4201	51000	0.0654	-	-	-
3.4536	51500	0.0659	-	-	-
3.4871	52000	0.0445	-	-	-
3.5207	52500	0.0355	-	-	-
3.5542	53000	0.0307	-	-	-
3.5877	53500	0.0577	-	-	-
3.6212	54000	0.129	-	-	-
3.6548	54500	0.1727	-	-	-
3.6883	55000	0.0952	-	-	-
3.7218	55500	0.03	-	-	-
3.7554	56000	0.0263	-	-	-
3.7889	56500	0.059	-	-	-
3.8224	57000	0.3222	-	-	-
3.8560	57500	0.0727	-	-	-
3.8895	58000	0.0072	-	-	-
3.9230	58500	0.0229	-	-	-
3.9565	59000	0.0877	-	-	-
3.9901	59500	0.0273	-	-	-
4.0	59648	-	0.1633	0.9357	-
4.0236	60000	0.111	-	-	-
4.0571	60500	0.0897	-	-	-
4.0907	61000	0.0117	-	-	-
4.1242	61500	0.0077	-	-	-
4.1577	62000	0.005	-	-	-
4.1913	62500	0.0115	-	-	-
4.2248	63000	0.0463	-	-	-
4.2583	63500	0.097	-	-	-
4.2918	64000	0.0713	-	-	-
4.3254	64500	0.1869	-	-	-
4.3589	65000	0.1845	-	-	-
4.3924	65500	0.0267	-	-	-
4.4260	66000	0.041	-	-	-
4.4595	66500	0.0463	-	-	-
4.4930	67000	0.0239	-	-	-
4.5266	67500	0.0276	-	-	-
4.5601	68000	0.0176	-	-	-
4.5936	68500	0.0409	-	-	-
4.6271	69000	0.107	-	-	-
4.6607	69500	0.1604	-	-	-
4.6942	70000	0.0495	-	-	-
4.7277	70500	0.0268	-	-	-
4.7613	71000	0.0259	-	-	-
4.7948	71500	0.0478	-	-	-
4.8283	72000	0.3	-	-	-
4.8619	72500	0.0436	-	-	-
4.8954	73000	0.0059	-	-	-
4.9289	73500	0.0295	-	-	-
4.9624	74000	0.0926	-	-	-
4.9960	74500	0.0191	-	-	-
5.0	74560	-	0.1558	0.9402	0.9288

Framework Versions

Python: 3.12.4
Sentence Transformers: 3.2.1
Transformers: 4.42.4
PyTorch: 2.3.1
Accelerate: 0.31.0
Datasets: 2.20.0
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",
}

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}
}