thealper2/graphcodebert-code-clone-detection

1{2  "dataset_name": "PoolC/1-fold-clone-detection-600k-5fold",3  "train_split": "train",4  "heldout_split": "val",5  "test_group_fraction": 0.5,6  "max_train_samples": 50000,7  "max_eval_samples": 20000,8  "max_test_samples": 20000,9  "balance_subsamples": true,10  "model_name_or_path": "microsoft/graphcodebert-base",11  "code_length": 512,12  "data_flow_length": 128,13  "attn_implementation": "sdpa",14  "learning_rate": 2e-05,15  "num_train_epochs": 3.0,16  "per_device_train_batch_size": 16,17  "per_device_eval_batch_size": 32,18  "gradient_accumulation_steps": 1,19  "weight_decay": 0.01,20  "warmup_ratio": 0.1,21  "max_grad_norm": 1.0,22  "fp16": true,23  "bf16": false,24  "gradient_checkpointing": false,25  "optim": "adamw_torch",26  "lr_scheduler_type": "linear",27  "class_weighting": "auto",28  "class_weight_threshold": 0.6,29  "eval_strategy": "steps",30  "eval_steps": 1000,31  "save_strategy": "steps",32  "save_steps": 1000,33  "save_total_limit": 2,34  "logging_steps": 100,35  "metric_for_best_model": "f1",36  "greater_is_better": true,37  "load_best_model_at_end": true,38  "seed": 42,39  "full_determinism": false,40  "dataloader_num_workers": 4,41  "preprocessing_num_workers": 8,42  "output_dir": "./outputs",43  "model_dir": "./models/graphcodebert-clone-detection",44  "logging_dir": "./logs",45  "cache_dir": "./outputs/feature_cache",46  "report_to": "none",47  "run_sanity_check": true,48  "sanity_check_samples": 64,49  "total_sequence_length": 640,50  "effective_batch_size": 1651}