| { | |
| "schema_version": 1, | |
| "stage": "sft", | |
| "experiment_suffix": "pre1930-curriculum-c3-robust-v2", | |
| "data": { | |
| "recipe": "curriculum", | |
| "curriculum": { | |
| "name": "C3Rv2", | |
| "mode": "staged", | |
| "threshold_default": 80, | |
| "stages": [ | |
| { | |
| "routes": [ | |
| "knowledge_qa" | |
| ], | |
| "authentic": "single" | |
| }, | |
| { | |
| "routes": [ | |
| "reasoning_qa", | |
| "stem_reasoning", | |
| "how_to_qa", | |
| "opinion_qa", | |
| "composition_qa", | |
| "verse_qa" | |
| ], | |
| "calibration_qa": true | |
| }, | |
| { | |
| "routes": [ | |
| "multiturn_qa", | |
| "narrative_grounded", | |
| "narrative_fiction" | |
| ], | |
| "authentic": "multi" | |
| } | |
| ], | |
| "noise": { | |
| "rate": 0.3 | |
| }, | |
| "robustness": { | |
| "stage": 0, | |
| "epochs": 1, | |
| "routes": { | |
| "conversation_qa": { | |
| "count": null | |
| }, | |
| "unparseable_qa": { | |
| "count": null | |
| }, | |
| "typo_qa": { | |
| "count": null | |
| }, | |
| "era_qa": { | |
| "count": null | |
| }, | |
| "conversation_multiturn": { | |
| "count": null | |
| } | |
| } | |
| } | |
| } | |
| }, | |
| "training": { | |
| "num_iterations": -1, | |
| "load_optimizer": 0, | |
| "device_batch_size": 2, | |
| "warmup_ratio": 0.03, | |
| "eval_every": 400, | |
| "chatcore_every": 100000, | |
| "chatcore_max_sample": 32, | |
| "save_every": -1 | |
| }, | |
| "artifacts": { | |
| "repo": "jbduran/think.nano" | |
| }, | |
| "wandb": { | |
| "enabled": true, | |
| "entity": "jbduran-thinkingmachinesncsu", | |
| "project": "think.nano", | |
| "group": "think-d32", | |
| "tags": [ | |
| "sft", | |
| "curriculum", | |
| "c3", | |
| "scale-max", | |
| "staged", | |
| "robustness", | |
| "noise", | |
| "d32" | |
| ] | |
| }, | |
| "config_fingerprint": "4462a389dc98cd5d", | |
| "artifact_path": "experiments/Think.Unbounded-d32-v2mix-cont/sft/Think.Unbounded-d32-v2mix-cont-pre1930-curriculum-c3-robust-v2" | |
| } | |