初始化项目,由ModelHub XC社区提供模型

Model: SeongryongJung/qwen3-4b-tooluse-grpo
Source: Original Platform
This commit is contained in:
ModelHub XC
2026-08-21 00:51:49 +08:00
commit 62e6b3603c
17 changed files with 152421 additions and 0 deletions

51
metrics.json Normal file
View File

@@ -0,0 +1,51 @@
{
"wandb_run": "run-20260630_031808-osu8wla2",
"wandb_output_log": "/mnt/mole/SDPO/L2T/wandb/run-20260630_031808-osu8wla2/files/output.log",
"metric_key": "val-aux/tooluse/reward/mean@16",
"evals": [
{
"step": 10,
"mean16": 0.5799632352941176
},
{
"step": 20,
"mean16": 0.6011029411764706
},
{
"step": 30,
"mean16": 0.6121323529411765
},
{
"step": 40,
"mean16": 0.6011029411764706
},
{
"step": 50,
"mean16": 0.609375
},
{
"step": 60,
"mean16": 0.5928308823529411
},
{
"step": 70,
"mean16": 0.5863970588235294
},
{
"step": 80,
"mean16": 0.6130514705882353
},
{
"step": 90,
"mean16": 0.5772058823529411
},
{
"step": 100,
"mean16": 0.6029411764705882
}
],
"best_step": 80,
"best_mean16": 0.6130514705882353,
"final_step": 100,
"final_mean16": 0.6029411764705882
}