| { |
| "output_dir": "models/gliner2-agri-lora", |
| "experiment_name": "gliner2", |
| "num_epochs": 5, |
| "max_steps": -1, |
| "batch_size": 8, |
| "eval_batch_size": 8, |
| "gradient_accumulation_steps": 4, |
| "encoder_lr": 1e-05, |
| "task_lr": 0.0005, |
| "weight_decay": 0.01, |
| "adam_beta1": 0.9, |
| "adam_beta2": 0.999, |
| "adam_epsilon": 1e-08, |
| "max_grad_norm": 1.0, |
| "scheduler_type": "linear", |
| "warmup_ratio": 0.1, |
| "warmup_steps": 0, |
| "num_cycles": 0.5, |
| "fp16": true, |
| "bf16": false, |
| "eval_strategy": "epoch", |
| "eval_steps": 500, |
| "save_total_limit": 3, |
| "save_best": true, |
| "metric_for_best": "eval_loss", |
| "greater_is_better": false, |
| "logging_steps": 10, |
| "logging_first_step": true, |
| "report_to_wandb": false, |
| "wandb_project": null, |
| "wandb_entity": null, |
| "wandb_run_name": null, |
| "wandb_tags": [], |
| "wandb_notes": null, |
| "early_stopping": false, |
| "early_stopping_patience": 3, |
| "early_stopping_threshold": 0.0, |
| "num_workers": 4, |
| "pin_memory": true, |
| "prefetch_factor": 2, |
| "seed": 42, |
| "deterministic": false, |
| "local_rank": -1, |
| "debug": false, |
| "max_train_samples": -1, |
| "max_eval_samples": -1, |
| "validate_data": true, |
| "max_len": null, |
| "strict_training": true, |
| "allow_invalid_samples": false, |
| "log_proposal_metrics": true, |
| "gold_injection_start": 1.0, |
| "gold_injection_end": 0.25, |
| "gold_injection_hold_frac": 0.15, |
| "diagnostics_every_n_steps": 0, |
| "profile_first_n_steps": 0, |
| "ddp_consensus_check": true, |
| "ddp_find_unused_parameters": false, |
| "ddp_static_graph": true, |
| "dry_run_recall_steps": 0, |
| "gate_recall": 0.97, |
| "gate_long_recall": 0.93, |
| "on_capacity_exceeded": "raise", |
| "group_by_length": true, |
| "length_group_window_batches": 50, |
| "compile_model": false, |
| "gradient_checkpointing": true, |
| "fused_optimizer": true, |
| "allow_tf32": true, |
| "float32_matmul_precision": "high", |
| "use_lora": true, |
| "lora_r": 16, |
| "lora_alpha": 32.0, |
| "lora_dropout": 0.05, |
| "lora_use_dora": false, |
| "lora_target_modules": [ |
| "encoder", |
| "span_rep", |
| "classifier", |
| "count_embed", |
| "count_pred" |
| ], |
| "save_adapter_only": true |
| } |
