Buckets:
| { | |
| "base_repo_id": "ProjectScugnizz/scugnizz-50m-pcs", | |
| "base_ckpt": "training-runs/sft-pro-mix-v1/checkpoint_best.pt", | |
| "resume": "", | |
| "ddp_launched": false, | |
| "no_auto_ddp": false, | |
| "datasets": [ | |
| "Daisuke675/scugnizz-instruct-pro-50k", | |
| "iamtarun/python_code_instructions_18k_alpaca" | |
| ], | |
| "dataset_weights": [ | |
| 0.7, | |
| 0.3 | |
| ], | |
| "split": "train", | |
| "sft_mount_root": "/mnt/dataset", | |
| "max_samples_per_dataset": 50000, | |
| "synthetic_response_rows": 300000, | |
| "val_size": 1000, | |
| "packing": true, | |
| "block_size": 1024, | |
| "batch_size": 1, | |
| "grad_accum": 16, | |
| "epochs": 3, | |
| "max_steps": 0, | |
| "lr": 4e-06, | |
| "min_lr": 8e-07, | |
| "warmup_steps": 100, | |
| "weight_decay": 0.05, | |
| "grad_clip": 1.0, | |
| "dropout": 0.05, | |
| "eval_interval": 250, | |
| "save_interval": 250, | |
| "log_interval": 10, | |
| "eval_batches": 30, | |
| "seed": 42, | |
| "dtype": "auto", | |
| "out_dir": "/mnt/data-storage/training-runs/sft-rag-api-response-strictcopy-v1", | |
| "hub_repo_id": "ProjectScugnizz/scugnizz-50m-pcs", | |
| "hub_path": "training-runs/sft-rag-api-response-strictcopy-v1", | |
| "push": true, | |
| "push_every_save": true | |
| } |
Xet Storage Details
- Size:
- 1.1 kB
- Xet hash:
- 5bbedb608255131406bbc926b1d93416033c9991f00e5fb2959e335610c68a91
·
Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.