This is a sentence-transformers model finetuned from BAAI/bge-base-en-v1.5. It maps sentences & paragraphs to a 768-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-base-en-v1.5
Maximum Sequence Length: 512 tokens
Output Dimensionality: 768 tokens
Similarity Function: Cosine Similarity

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': 512, 'do_lower_case': True}) with Transformer model: BertModel 
  (1): Pooling({'word_embedding_dimension': 768, '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("Nutanix/bge-base-mbpp-processed")
# Run inference
sentences = [
    'Write a function to find sum and average of first n natural numbers.',
    'def sum_average(number):\r\n total = 0\r\n for value in range(1, number + 1):\r\n    total = total + value\r\n average = total / number\r\n return (total,average)',
    'def long_words(n, str):\r\n    word_len = []\r\n    txt = str.split(" ")\r\n    for x in txt:\r\n        if len(x) > n:\r\n            word_len.append(x)\r\n    return word_len\t',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 768]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [3, 3]

Evaluation

Metrics

Triplet

Dataset: sts-dev
Evaluated with TripletEvaluator

Metric	Value
cosine_accuracy	0.9642
dot_accuracy	0.0358
manhattan_accuracy	0.9623
euclidean_accuracy	0.9642
max_accuracy	0.9642

Training Details

Training Hyperparameters

Non-Default Hyperparameters

per_device_train_batch_size: 16
per_device_eval_batch_size: 16
num_train_epochs: 1
bf16: True
batch_sampler: no_duplicates

All Hyperparameters

Click to expand

overwrite_output_dir: False
do_predict: False
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
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
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
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, '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_sampler: no_duplicates
multi_dataset_batch_sampler: proportional

Training Logs

Click to expand

Epoch	Step	Training Loss	sts-dev_max_accuracy
0.0050	100	4.2871	-
0.0101	200	4.1063	-
0.0151	300	4.0535	-
0.0202	400	4.0162	-
0.0252	500	3.9802	-
0.0302	600	3.982	-
0.0353	700	3.9349	-
0.0403	800	3.9372	-
0.0453	900	3.9175	-
0.0504	1000	3.9201	-
0.0554	1100	3.9036	-
0.0605	1200	3.8985	-
0.0655	1300	3.8994	-
0.0705	1400	3.8979	-
0.0756	1500	3.8846	-
0.0806	1600	3.897	-
0.0857	1700	3.8779	-
0.0907	1800	3.895	-
0.0957	1900	3.8793	-
0.1008	2000	3.8677	-
0.1058	2100	3.8672	-
0.1109	2200	3.8744	-
0.1159	2300	3.8721	-
0.1209	2400	3.8554	-
0.1260	2500	3.8379	-
0.1310	2600	3.8477	-
0.1360	2700	3.8625	-
0.1411	2800	3.8426	-
0.1461	2900	3.8475	-
0.1512	3000	3.8462	-
0.1562	3100	3.8314	-
0.1612	3200	3.8227	-
0.1663	3300	3.8311	-
0.1713	3400	3.8305	-
0.1764	3500	3.8249	-
0.1814	3600	3.842	-
0.1864	3700	3.8259	-
0.1915	3800	3.8295	-
0.1965	3900	3.8217	-
0.2016	4000	3.8362	-
0.2066	4100	3.8354	-
0.2116	4200	3.8175	-
0.2167	4300	3.8108	-
0.2217	4400	3.84	-
0.2267	4500	3.8206	-
0.2318	4600	3.8065	-
0.2368	4700	3.8161	-
0.2419	4800	3.8234	-
0.2469	4900	3.8112	-
0.2519	5000	3.8006	-
0.2570	5100	3.7996	-
0.2620	5200	3.8104	-
0.2671	5300	3.7975	-
0.2721	5400	3.7944	-
0.2771	5500	3.8037	-
0.2822	5600	3.7996	-
0.2872	5700	3.7883	-
0.2923	5800	3.7978	-
0.2973	5900	3.7889	-
0.3023	6000	3.7794	-
0.3074	6100	3.7669	-
0.3124	6200	3.7844	-
0.3174	6300	3.7687	-
0.3225	6400	3.7915	-
0.3275	6500	3.7505	-
0.3326	6600	3.7979	-
0.3376	6700	3.7681	-
0.3426	6800	3.7738	-
0.3477	6900	3.7702	-
0.3527	7000	3.7679	-
0.3578	7100	3.7862	-
0.3628	7200	3.7718	-
0.3678	7300	3.7898	-
0.3729	7400	3.7832	-
0.3779	7500	3.7701	-
0.3829	7600	3.7805	-
0.3880	7700	3.7725	-
0.3930	7800	3.7476	-
0.3981	7900	3.7612	-
0.4031	8000	3.7555	-
0.4081	8100	3.7489	-
0.4132	8200	3.7507	-
0.4182	8300	3.741	-
0.4233	8400	3.7465	-
0.4283	8500	3.7445	-
0.4333	8600	3.7536	-
0.4384	8700	3.7279	-
0.4434	8800	3.745	-
0.4485	8900	3.7712	-
0.4535	9000	3.7429	-
0.4585	9100	3.7386	-
0.4636	9200	3.7328	-
0.4686	9300	3.735	-
0.4736	9400	3.7451	-
0.4787	9500	3.7292	-
0.4837	9600	3.7381	-
0.4888	9700	3.7346	-
0.4938	9800	3.7396	-
0.4988	9900	3.7122	-
0.5039	10000	3.7295	-
0.5089	10100	3.738	-
0.5140	10200	3.7272	-
0.5190	10300	3.7233	-
0.5240	10400	3.7066	-
0.5291	10500	3.713	-
0.5341	10600	3.7185	-
0.5392	10700	3.7236	-
0.5442	10800	3.7086	-
0.5492	10900	3.718	-
0.5543	11000	3.7296	-
0.5593	11100	3.7276	-
0.5643	11200	3.7164	-
0.5694	11300	3.7119	-
0.5744	11400	3.6979	-
0.5795	11500	3.7095	-
0.5845	11600	3.7067	-
0.5895	11700	3.7018	-
0.5946	11800	3.727	-
0.5996	11900	3.7136	-
0.6047	12000	3.7233	-
0.6097	12100	3.7076	-
0.6147	12200	3.7243	-
0.6198	12300	3.6966	-
0.6248	12400	3.7058	-
0.6298	12500	3.698	-
0.6349	12600	3.6934	-
0.6399	12700	3.7046	-
0.6450	12800	3.6986	-
0.6500	12900	3.7134	-
0.6550	13000	3.7019	-
0.6601	13100	3.7154	-
0.6651	13200	3.7056	-
0.6702	13300	3.6948	-
0.6752	13400	3.697	-
0.6802	13500	3.6998	-
0.6853	13600	3.6881	-
0.6903	13700	3.6982	-
0.6954	13800	3.6861	-
0.7004	13900	3.6816	-
0.7054	14000	3.7004	-
0.7105	14100	3.6895	-
0.7155	14200	3.7045	-
0.7205	14300	3.7078	-
0.7256	14400	3.6884	-
0.7306	14500	3.6839	-
0.7357	14600	3.6891	-
0.7407	14700	3.6864	-
0.7457	14800	3.7069	-
0.7508	14900	3.6879	-
0.7558	15000	3.7049	-
0.7609	15100	3.7099	-
0.7659	15200	3.6908	-
0.7709	15300	3.6973	-
0.7760	15400	3.6775	-
0.7810	15500	3.6776	-
0.7861	15600	3.706	-
0.7911	15700	3.6941	-
0.7961	15800	3.6974	-
0.8012	15900	3.6706	-
0.8062	16000	3.6922	-
0.8112	16100	3.6898	-
0.8163	16200	3.7005	-
0.8213	16300	3.691	-
0.8264	16400	3.7066	-
0.8314	16500	3.6959	-
0.8364	16600	3.6944	-
0.8415	16700	3.6724	-
0.8465	16800	3.6783	-
0.8516	16900	3.683	-
0.8566	17000	3.6929	-
0.8616	17100	3.6823	-
0.8667	17200	3.6737	-
0.8717	17300	3.6847	-
0.8768	17400	3.6786	-
0.8818	17500	3.7018	-
0.8868	17600	3.6944	-
0.8919	17700	3.687	-
0.8969	17800	3.6841	-
0.9019	17900	3.6764	-
0.9070	18000	3.6779	-
0.9120	18100	3.689	-
0.9171	18200	3.6837	-
0.9221	18300	3.7034	-
0.9271	18400	3.669	-
0.9322	18500	3.67	-
0.9372	18600	3.6868	-
0.9423	18700	3.6916	-
0.9473	18800	3.6751	-
0.9523	18900	3.6935	-
0.9574	19000	3.702	-
0.9624	19100	3.6761	-
0.9674	19200	3.6798	-
0.9725	19300	3.6844	-
0.9775	19400	3.6775	-
0.9826	19500	3.6679	-
0.9876	19600	3.6793	-
0.9926	19700	3.6833	-
0.9977	19800	3.6717	-
1.0	19846	-	0.9642

Framework Versions

Python: 3.10.14
Sentence Transformers: 3.0.1
Transformers: 4.40.0
PyTorch: 2.3.0+cu121
Accelerate: 0.33.0
Datasets: 2.20.0
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",
}

TripletLoss

bibtex

@misc{hermans2017defense,
    title={In Defense of the Triplet Loss for Person Re-Identification}, 
    author={Alexander Hermans and Lucas Beyer and Bastian Leibe},
    year={2017},
    eprint={1703.07737},
    archivePrefix={arXiv},
    primaryClass={cs.CV}
}