This is a sentence-transformers model finetuned from BAAI/bge-small-zh-v1.5 on the json dataset. It maps sentences & paragraphs to a 512-dimensional dense vector space and can be used for retrieval.

Model Details

Model Description

Model Type: Sentence Transformer
Base model: BAAI/bge-small-zh-v1.5
Maximum Sequence Length: 512 tokens
Output Dimensionality: 512 dimensions
Similarity Function: Cosine Similarity
Supported Modality: Text
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({'transformer_task': 'feature-extraction', 'modality_config': {'text': {'method': 'forward', 'method_output_name': 'last_hidden_state'}}, 'module_output_name': 'token_embeddings', 'architecture': 'BertModel'})
  (1): Pooling({'embedding_dimension': 512, 'pooling_mode': 'cls', '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("johnyy212/moe-girl-v3")
# Run inference
sentences = [
    '御姐、长短袜、下乳、长靴、红瞳',
    '角色：少女前线:PTRD\n本名：PTRD41反坦克步枪\n别名：卤蛋、南半球\n声优：清水爱\n萌点：长发、巨乳、黑丝、御姐、下乳、呆毛、哥萨克帽\n发色：亚麻\n瞳色：红',
    '少女前线:PTRD',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 512]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities)
# tensor([[ 1.0000,  0.5500, -0.1585],
#         [ 0.5500,  1.0000, -0.1041],
#         [-0.1585, -0.1041,  1.0000]])

Training Details

Training Dataset

json

Dataset: json
Size: 58,482 training samples
Columns: anchor, positive, and char_name
Approximate statistics based on the first 1000 samples:
anchor positive char_name
type string string string
details
min: 3 tokens
mean: 11.96 tokens
max: 36 tokens
min: 13 tokens
mean: 77.78 tokens
max: 478 tokens
min: 3 tokens
mean: 7.23 tokens
max: 23 tokens

Samples:

anchor	positive	char_name
`杀手、吐槽、红发、混血儿`	`角色：风吹雪[object Object]本名：风吹雪[object Object]别名：雪妹、花、天意之花[object Object]年龄：16[object Object]瞳色：棕[object Object]发色：红[object Object]血型：O[object Object]萌点：长直、傲娇、女神系、吐槽、妹妹、裹胸布`	`风吹雪`
`黑发、百合、高中生、半马尾、黑色连裤袜`	`角色：高远凛[object Object]本名：高远、凛[object Object]别名：凛、Rin[object Object]声优：长妻树里[object Object]生日：10月10日[object Object]瞳色：棕[object Object]发色：黑[object Object]三围：85、60、86[object Object]萌点：高中生、短发、穿越者、百合、黑丝、半马尾`	`高远凛`
`小野凉子、北见六花`	`角色：世计丸[object Object]本名：世计、丸[object Object]声优：北见六花配音角色、小野凉子配音角色[object Object]发色：紫[object Object]瞳色：紫[object Object]三围：90、58、81[object Object]生日：12月1日[object Object]星座：射手[object Object]萌点：过膝袜、长发、下双马尾、毒舌、绝对领域、姐控、遮眼发、巨乳、傲娇`	`世计丸`

Loss: MultipleNegativesRankingLoss with these parameters:
json
1{ 2 "scale": 20.0, 3 "similarity_fct": "cos_sim", 4 "gather_across_devices": false, 5 "directions": [ 6 "query_to_doc" 7 ], 8 "partition_mode": "joint", 9 "hardness_mode": null, 10 "hardness_strength": 0.0 11}

Evaluation Dataset

json

Dataset: json
Size: 3,615 evaluation samples
Columns: anchor, positive, and char_name
Approximate statistics based on the first 1000 samples:
anchor positive char_name
type string string string
details
min: 9 tokens
mean: 15.88 tokens
max: 29 tokens
min: 21 tokens
mean: 77.17 tokens
max: 307 tokens
min: 3 tokens
mean: 7.31 tokens
max: 24 tokens

Samples:

anchor	positive	char_name
`萝莉控、黑色吊带袜、手套、绿瞳`	`角色：碧蓝航线:皇家方舟[object Object]本名：HMS Ark Royal[object Object]别名：狱友[object Object]发色：紫[object Object]瞳色：绿[object Object]萌点：御姐、黑色吊带袜、萝莉控、巨乳、遮眼发[object Object]生日：4月13日`	`碧蓝航线:皇家方舟`
`孤儿、紫瞳、披风、长直、黑发、黑长直`	`角色：江昙[object Object]本名：江昙、Jiang Tan[object Object]别名：“圣女”、小昙花[object Object]声优：白杺瓒、皛四白、富田美忧[object Object]发色：黑[object Object]瞳色：紫[object Object]萌点：披风、针织衫、黑长直、孤儿`	`江昙`
`尾巴、自大、嗜杀`	`角色：紫影狒狒[object Object]本名：Purprill、the、Mandroid、紫影狒狒[object Object]声优：高木涉[object Object]瞳色：绿[object Object]萌点：狒狒、尾巴、自大、嗜杀`	`紫影狒狒`

Loss: MultipleNegativesRankingLoss with these parameters:
json
1{ 2 "scale": 20.0, 3 "similarity_fct": "cos_sim", 4 "gather_across_devices": false, 5 "directions": [ 6 "query_to_doc" 7 ], 8 "partition_mode": "joint", 9 "hardness_mode": null, 10 "hardness_strength": 0.0 11}

Training Hyperparameters

Non-Default Hyperparameters

per_device_train_batch_size: 64
learning_rate: 2e-05
num_train_epochs: 5
warmup_steps: 0.1
fp16: True
load_best_model_at_end: True
batch_sampler: no_duplicates

All Hyperparameters

Click to expand

do_predict: False
prediction_loss_only: True
per_device_train_batch_size: 64
per_device_eval_batch_size: 8
gradient_accumulation_steps: 1
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: linear
lr_scheduler_kwargs: None
warmup_ratio: None
warmup_steps: 0.1
log_level: passive
log_level_replica: warning
log_on_each_node: True
logging_nan_inf_filter: True
enable_jit_checkpoint: False
save_on_each_node: False
save_only_model: False
restore_callback_states_from_checkpoint: False
use_cpu: False
seed: 42
data_seed: None
bf16: False
fp16: True
bf16_full_eval: False
fp16_full_eval: False
tf32: None
local_rank: -1
ddp_backend: None
debug: []
dataloader_drop_last: False
dataloader_num_workers: 0
dataloader_prefetch_factor: None
disable_tqdm: False
remove_unused_columns: True
label_names: None
load_best_model_at_end: True
ignore_data_skip: False
fsdp: []
fsdp_config: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}
accelerator_config: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
parallelism_config: None
deepspeed: None
label_smoothing_factor: 0.0
optim: adamw_torch_fused
optim_args: None
group_by_length: False
length_column_name: length
project: huggingface
trackio_space_id: trackio
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
push_to_hub: False
resume_from_checkpoint: None
hub_model_id: None
hub_strategy: every_save
hub_private_repo: None
hub_always_push: False
hub_revision: None
gradient_checkpointing: False
gradient_checkpointing_kwargs: None
include_for_metrics: []
eval_do_concat_batches: True
auto_find_batch_size: False
full_determinism: False
ddp_timeout: 1800
torch_compile: False
torch_compile_backend: None
torch_compile_mode: None
include_num_input_tokens_seen: no
neftune_noise_alpha: None
optim_target_modules: None
batch_eval_metrics: False
eval_on_start: False
use_liger_kernel: False
liger_kernel_config: None
eval_use_gather_object: False
average_tokens_across_devices: True
use_cache: False
prompts: None
batch_sampler: no_duplicates
multi_dataset_batch_sampler: proportional
router_mapping: {}
learning_rate_mapping: {}

Training Logs

Epoch	Step	Training Loss	Validation Loss
0.1094	100	2.4634	-
0.2188	200	1.1738	-
0.3282	300	1.0204	-
0.4376	400	0.8706	-
0.5470	500	0.8237	-
0.6565	600	0.8114	-
0.7659	700	0.7423	-
0.8753	800	0.7134	-
0.9847	900	0.6652	-
1.0	914	-	0.1627
1.0941	1000	0.6096	-
1.2035	1100	0.6270	-
1.3129	1200	0.6147	-
1.4223	1300	0.5881	-
1.5317	1400	0.6169	-
1.6411	1500	0.5780	-
1.7505	1600	0.5719	-
1.8600	1700	0.5977	-
1.9694	1800	0.5682	-
2.0	1828	-	0.1404
2.0788	1900	0.5367	-
2.1882	2000	0.5234	-
2.2976	2100	0.4955	-
2.4070	2200	0.5051	-
2.5164	2300	0.5314	-
2.6258	2400	0.4933	-
2.7352	2500	0.4881	-
2.8446	2600	0.4982	-
2.9540	2700	0.4992	-
3.0	2742	-	0.1312
3.0635	2800	0.4945	-
3.1729	2900	0.4468	-
3.2823	3000	0.4524	-
3.3917	3100	0.4675	-
3.5011	3200	0.4591	-
3.6105	3300	0.4627	-
3.7199	3400	0.4545	-
3.8293	3500	0.4718	-
3.9387	3600	0.4558	-
4.0	3656	-	0.1237
4.0481	3700	0.4439	-
4.1575	3800	0.4375	-
4.2670	3900	0.4384	-
4.3764	4000	0.4486	-
4.4858	4100	0.4373	-
4.5952	4200	0.4356	-
4.7046	4300	0.4425	-
4.8140	4400	0.4285	-
4.9234	4500	0.4219	-
5.0	4570	-	0.1243

The bold row denotes the saved checkpoint.

Training Time

Training: 15.3 minutes

Framework Versions

Python: 3.12.13
Sentence Transformers: 5.4.1
Transformers: 5.0.0
PyTorch: 2.10.0+cu128
Accelerate: 1.13.0
Datasets: 4.8.5
Tokenizers: 0.22.2

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{oord2019representationlearningcontrastivepredictive,
      title={Representation Learning with Contrastive Predictive Coding},
      author={Aaron van den Oord and Yazhe Li and Oriol Vinyals},
      year={2019},
      eprint={1807.03748},
      archivePrefix={arXiv},
      primaryClass={cs.LG},
      url={https://arxiv.org/abs/1807.03748},
}

	anchor	positive	char_name
type	string	string	string
details	min: 3 tokens mean: 11.96 tokens max: 36 tokens	min: 13 tokens mean: 77.78 tokens max: 478 tokens	min: 3 tokens mean: 7.23 tokens max: 23 tokens