| { |
| "model_name": "microsoft/deberta-v3-base", |
| "name": "token level gliner", |
| "max_width": 100, |
| "hidden_size": 768, |
| "dropout": 0.1, |
| "fine_tune": true, |
| "subtoken_pooling": "first", |
| "span_mode": "token_level", |
| "num_steps": 30000, |
| "train_batch_size": 8, |
| "eval_every": 2500, |
| "warmup_ratio": 0.1, |
| "scheduler_type": "cosine", |
| "loss_alpha": -1, |
| "loss_gamma": 0, |
| "loss_reduction": "sum", |
| "lr_encoder": "1e-6", |
| "lr_others": "1e-5", |
| "weight_decay_encoder": 0.01, |
| "weight_decay_other": 0.01, |
| "root_dir": "token_gliner_instruct_logs", |
| "train_data": "/gpfswork/rech/bwq/upa43yu/LLMexp/instruct_1k.json", |
| "val_data_dir": "/gpfswork/rech/ohy/upa43yu/NER_datasets", |
| "prev_path": "/gpfswork/rech/bwq/upa43yu/gliner-pip/train_gliner/token_gliner_ft_logs/deberta-v3-base_2/model_30000", |
| "size_sup": -1, |
| "max_types": 10, |
| "shuffle_types": true, |
| "random_drop": true, |
| "max_neg_type_ratio": 3, |
| "max_len": 384, |
| "freeze_token_rep": false, |
| "gpu": "pds@a100", |
| "run_time": "3:30:00", |
| "log_dir": "/gpfsdswork/projects/rech/bwq/upa43yu/gliner-pip/train_gliner/token_gliner_instruct_logs/deberta-v3-base_1" |
| } |