| { | |
| "repo_id": "habanoz/wiki-text-gpt-v0.12", | |
| "trainer_config": { | |
| "seq_length": 512, | |
| "gradient_accumulation_steps": 1, | |
| "batch_size": 32, | |
| "data_dir": "eco-news-tr", | |
| "warmup_iters": 100, | |
| "learning_rate": 0.001, | |
| "lr_decay_iters": 5000, | |
| "max_iters": 5000, | |
| "min_lr": 0.0001, | |
| "weight_decay": 0.1, | |
| "beta1": 0.9, | |
| "beta2": 0.99, | |
| "compile": false, | |
| "decay_lr": true, | |
| "seed": 149, | |
| "log_interval": 10, | |
| "eval_interval": 250, | |
| "eval_iters": 200, | |
| "out_dir": "wiki-text-gpt-v0.12", | |
| "wandb_log": true, | |
| "wandb_project": "NB-Haber-GPT-Training", | |
| "wandb_run_name": "haber-gpt-v1.022(wikitr-tokenizer-seed-149)", | |
| "wandb_run_id": "1721342973", | |
| "grad_norm_clip": 1.0, | |
| "dtype": "float16" | |
| } | |
| } |