Shuu12121/CodeSearch-ModernBERT-Owl-4.1-Plus

SentenceTransformer based on Shuu12121/CodeModernBERT-Owl-4.1

This is a sentence-transformers model finetuned from Shuu12121/CodeModernBERT-Owl-4.1. 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: Shuu12121/CodeModernBERT-Owl-4.1
• Maximum Sequence Length: 1024 tokens
• Output Dimensionality: 768 dimensions
• 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': 1024, 'do_lower_case': False}) with Transformer model: ModernBertModel 
  (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})
)

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("sentence_transformers_model_id")
# Run inference
sentences = [
    'Register the router instance.\n\n@return void',
    "protected function registerRouter()\n\t{\n\t\t$this->app['router'] = $this->app->share(function($app)\n\t\t{\n\t\t\treturn new Router($app['events'], $app);\n\t\t});\n\t}",
    'func (d *Driver) DiffSize(id string, idMappings *idtools.IDMappings, parent string, parentMappings *idtools.IDMappings, mountLabel string) (size int64, err error) {\n\tif d.useNaiveDiff() || !d.isParent(id, parent) {\n\t\treturn d.naiveDiff.DiffSize(id, idMappings, parent, parentMappings, mountLabel)\n\t}\n\treturn directory.Size(d.getDiffPath(id))\n}',
]
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]

Training Details

Training Dataset

Unnamed Dataset

• Size: 6,960,000 training samples
• Columns: sentence_0, sentence_1, and label
• Approximate statistics based on the first 1000 samples:

  	sentence_0	sentence_1	label
  type	string	string	float
  details	min: 3 tokens mean: 50.31 tokens max: 1024 tokens	min: 28 tokens mean: 164.73 tokens max: 1024 tokens	min: 1.0 mean: 1.0 max: 1.0

• Samples:

  sentence_0	sentence_1	label
  // GetNodeID returns the NodeID field if it's non-nil, zero value otherwise.	func (a *App) GetNodeID() string { if a == nil
  // SignVote signs a canonical representation of the vote, along with the // chainID. Implements PrivValidator.	func (pv *FilePV) SignVote(chainID string, vote *types.Vote) error { if err := pv.signVote(chainID, vote); err != nil { return fmt.Errorf("error signing vote: %v", err) } return nil }	1.0
  //GetQyAccessToken 获取access_token	func (ctx *Context) GetQyAccessToken() (accessToken string, err error) { ctx.accessTokenLock.Lock() defer ctx.accessTokenLock.Unlock() accessTokenCacheKey := fmt.Sprintf("qy_access_token_%s", ctx.AppID) val := ctx.Cache.Get(accessTokenCacheKey) if val != nil { accessToken = val.(string) return } //从微信服务器获取 var resQyAccessToken ResQyAccessToken resQyAccessToken, err = ctx.GetQyAccessTokenFromServer() if err != nil { return } accessToken = resQyAccessToken.AccessToken return }	1.0

• Loss: MultipleNegativesRankingLoss with these parameters:

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

Training Hyperparameters

Non-Default Hyperparameters

• per_device_train_batch_size: 250
• per_device_eval_batch_size: 250
• fp16: True
• multi_dataset_batch_sampler: round_robin

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: False
• eval_strategy: no
• prediction_loss_only: True
• per_device_train_batch_size: 250
• per_device_eval_batch_size: 250
• 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
• num_train_epochs: 3
• max_steps: -1
• lr_scheduler_type: linear
• lr_scheduler_kwargs: {}
• warmup_ratio: 0.0
• 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
• 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: True
• 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: None
• hub_always_push: False
• hub_revision: None
• 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
• liger_kernel_config: None
• eval_use_gather_object: False
• average_tokens_across_devices: False
• prompts: None
• batch_sampler: batch_sampler
• multi_dataset_batch_sampler: round_robin

Training Logs

Click to expand

Epoch	Step	Training Loss
0.0180	500	0.9746
0.0359	1000	0.1636
0.0539	1500	0.1502
0.0718	2000	0.1374
0.0898	2500	0.1314
0.1078	3000	0.1241
0.1257	3500	0.1152
0.1437	4000	0.1146
0.1616	4500	0.1065
0.1796	5000	0.1014
0.1976	5500	0.0983
0.2155	6000	0.0987
0.2335	6500	0.0917
0.2514	7000	0.0912
0.2694	7500	0.0896
0.2874	8000	0.086
0.3053	8500	0.0811
0.3233	9000	0.0813
0.3412	9500	0.082
0.3592	10000	0.0759
0.3772	10500	0.0753
0.3951	11000	0.0722
0.4131	11500	0.0707
0.4310	12000	0.0699
0.4490	12500	0.0698
0.4670	13000	0.0679
0.4849	13500	0.0653
0.5029	14000	0.0641
0.5208	14500	0.063
0.5388	15000	0.0621
0.5568	15500	0.061
0.5747	16000	0.0581
0.5927	16500	0.0555
0.6106	17000	0.0552
0.6286	17500	0.0551
0.6466	18000	0.0533
0.6645	18500	0.0521
0.6825	19000	0.051
0.7004	19500	0.0509
0.7184	20000	0.0499
0.7364	20500	0.0468
0.7543	21000	0.0484
0.7723	21500	0.0466
0.7902	22000	0.0446
0.8082	22500	0.0453
0.8261	23000	0.0442
0.8441	23500	0.0424
0.8621	24000	0.0434
0.8800	24500	0.0416
0.8980	25000	0.0406
0.9159	25500	0.0404
0.9339	26000	0.0398
0.9519	26500	0.0406
0.9698	27000	0.0387
0.9878	27500	0.0386
1.0057	28000	0.0311
1.0237	28500	0.0193
1.0417	29000	0.0197
1.0596	29500	0.0186
1.0776	30000	0.0192
1.0955	30500	0.0194
1.1135	31000	0.0196
1.1315	31500	0.0198
1.1494	32000	0.0203
1.1674	32500	0.02
1.1853	33000	0.0184
1.2033	33500	0.0181
1.2213	34000	0.0195
1.2392	34500	0.0186
1.2572	35000	0.0184
1.2751	35500	0.0184
1.2931	36000	0.0194
1.3111	36500	0.0191
1.3290	37000	0.0183
1.3470	37500	0.0179
1.3649	38000	0.0179
1.3829	38500	0.0178
1.4009	39000	0.018
1.4188	39500	0.0182
1.4368	40000	0.0188
1.4547	40500	0.0172
1.4727	41000	0.0169
1.4907	41500	0.0173
1.5086	42000	0.0166
1.5266	42500	0.0157
1.5445	43000	0.0168
1.5625	43500	0.0158
1.5805	44000	0.016
1.5984	44500	0.0166
1.6164	45000	0.0168
1.6343	45500	0.0162
1.6523	46000	0.0153
1.6703	46500	0.0149
1.6882	47000	0.0158
1.7062	47500	0.0152
1.7241	48000	0.0147
1.7421	48500	0.0146
1.7601	49000	0.0145
1.7780	49500	0.0148
1.7960	50000	0.015
1.8139	50500	0.0145
1.8319	51000	0.0142
1.8499	51500	0.014
1.8678	52000	0.0139
1.8858	52500	0.0133
1.9037	53000	0.0135
1.9217	53500	0.0131
1.9397	54000	0.0134
1.9576	54500	0.013
1.9756	55000	0.0132
1.9935	55500	0.0122
2.0115	56000	0.0089
2.0295	56500	0.0061
2.0474	57000	0.0061
2.0654	57500	0.006
2.0833	58000	0.0062
2.1013	58500	0.0058
2.1193	59000	0.0059
2.1372	59500	0.0059
2.1552	60000	0.0059
2.1731	60500	0.0058
2.1911	61000	0.0059
2.2091	61500	0.0058
2.2270	62000	0.0059
2.2450	62500	0.0058
2.2629	63000	0.0057
2.2809	63500	0.0055
2.2989	64000	0.0056
2.3168	64500	0.0056
2.3348	65000	0.0056
2.3527	65500	0.0057
2.3707	66000	0.0055
2.3886	66500	0.0056
2.4066	67000	0.0054
2.4246	67500	0.0055
2.4425	68000	0.0052
2.4605	68500	0.0053
2.4784	69000	0.0052
2.4964	69500	0.0053
2.5144	70000	0.0052
2.5323	70500	0.0052
2.5503	71000	0.0051
2.5682	71500	0.0049
2.5862	72000	0.005
2.6042	72500	0.0047
2.6221	73000	0.0048
2.6401	73500	0.0047
2.6580	74000	0.0048
2.6760	74500	0.0048
2.6940	75000	0.0048
2.7119	75500	0.0047
2.7299	76000	0.0047
2.7478	76500	0.0046
2.7658	77000	0.0046
2.7838	77500	0.0044
2.8017	78000	0.0046
2.8197	78500	0.0047
2.8376	79000	0.0045
2.8556	79500	0.0043
2.8736	80000	0.0045
2.8915	80500	0.0044
2.9095	81000	0.0045
2.9274	81500	0.0045
2.9454	82000	0.0043
2.9634	82500	0.0042
2.9813	83000	0.0041
2.9993	83500	0.0044

Framework Versions

• Python: 3.11.13
• Sentence Transformers: 4.1.0
• Transformers: 4.53.0
• PyTorch: 2.6.0+cu124
• Accelerate: 1.8.1
• Datasets: 3.6.0
• Tokenizers: 0.21.2

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}
}