Files
qwen3-4b-physics-grpo/metrics.json

52 lines
882 B
JSON
Raw Permalink Normal View History

{
"wandb_run": "run-20260629_174849-hz5zwriz",
"wandb_output_log": "/mnt/mole/SDPO/L2T/wandb/run-20260629_174849-hz5zwriz/files/output.log",
"metric_key": "val-aux/sciknoweval/reward/mean@16",
"evals": [
{
"step": 10,
"mean16": 0.6
},
{
"step": 20,
"mean16": 0.62265625
},
{
"step": 30,
"mean16": 0.46875
},
{
"step": 40,
"mean16": 0.5234375
},
{
"step": 50,
"mean16": 0.5890625
},
{
"step": 60,
"mean16": 0.61796875
},
{
"step": 70,
"mean16": 0.5234375
},
{
"step": 80,
"mean16": 0.56171875
},
{
"step": 90,
"mean16": 0.534375
},
{
"step": 100,
"mean16": 0.53046875
}
],
"best_step": 20,
"best_mean16": 0.62265625,
"final_step": 100,
"final_mean16": 0.53046875
}