| { |
| "model_config": { |
| "sequence_len": 1024, |
| "vocab_size": 16384, |
| "n_layer": 6, |
| "n_head": 6, |
| "n_kv_head": 6, |
| "n_embd": 384, |
| "window_pattern": "L" |
| }, |
| "queen": { |
| "version": 1, |
| "name": "queen-v1", |
| "toy": false, |
| "source": "sft", |
| "chat_format": "chat", |
| "sft_method": "synthetic:deepseek-flash", |
| "step": 45, |
| "val_bpb": 1.5162134615351086, |
| "sft_val_bpb": 1.180598671538636, |
| "params": 42074366, |
| "params_scaling": 16908504, |
| "dtype": "bfloat16", |
| "exported_at": "2026-10-04T09:31:53+00:00", |
| "train_summary": { |
| "stages": { |
| "base": { |
| "wall_seconds": 544.6045060157776, |
| "invocations": 2 |
| }, |
| "sft": { |
| "wall_seconds": 59.08830523490906, |
| "invocations": 1 |
| } |
| }, |
| "gpu": { |
| "count": 1 |
| }, |
| "wall_hours": 0.168, |
| "gpu_hours": 0.168, |
| "usd_per_gpu_hour": 0.34, |
| "cost_usd": 0.06, |
| "sft": { |
| "method": "synthetic:deepseek-flash", |
| "files": [ |
| "sft/synthetic.train.jsonl", |
| "sft/headings.train.jsonl" |
| ], |
| "conversations": 75197, |
| "epochs": 2 |
| } |
| } |
| } |
| } |
