| { |
| "model_config": { |
| "sequence_len": 2048, |
| "vocab_size": 16384, |
| "n_layer": 10, |
| "n_head": 5, |
| "n_kv_head": 5, |
| "n_embd": 640, |
| "window_pattern": "L" |
| }, |
| "queen": { |
| "version": 3, |
| "name": "queen-v3", |
| "label": null, |
| "toy": false, |
| "source": "sft", |
| "chat_format": "chat", |
| "sft_method": "synthetic+openbook:deepseek-flash", |
| "step": 516, |
| "base_step": 1194, |
| "val_bpb": 0.8173534432759958, |
| "sft_val_bpb": 0.7245558715792655, |
| "params": 122552666, |
| "params_scaling": 59638060, |
| "dtype": "bfloat16", |
| "exported_at": "2026-10-06T13:44:05+00:00", |
| "train_summary": { |
| "stages": { |
| "base": { |
| "wall_seconds": 4416.387776136398, |
| "invocations": 1 |
| }, |
| "sft": { |
| "wall_seconds": 1959.2380418777466, |
| "invocations": 1 |
| } |
| }, |
| "gpu": { |
| "count": 1 |
| }, |
| "wall_hours": 1.771, |
| "gpu_hours": 1.771, |
| "usd_per_gpu_hour": 0.34, |
| "cost_usd": 0.6, |
| "base": { |
| "last_step": 1194, |
| "last_val_bpb": 0.8173534432759958, |
| "best_step": 1194, |
| "best_val_bpb": 0.8173534432759958, |
| "sft_from_step": 1194 |
| }, |
| "sft": { |
| "method": "synthetic+openbook:deepseek-flash", |
| "files": [ |
| "sft/synthetic.train.jsonl", |
| "sft/openbook.train.jsonl", |
| "sft/headings.train.jsonl" |
| ], |
| "conversations": 913355, |
| "epochs": 2, |
| "base_step": 1194 |
| } |
| } |
| } |
| } |
