b1k_model / evaluation /summary.json
davidwdw's picture
Upload B1K model checkpoints, training recipes, and evaluation artifacts (part 8)
c556874 verified
Raw History Blame Contribute Delete
471 Bytes
[
{
"variant": "human",
"completed": 20,
"successes": 8,
"success_rate": 0.4,
"mean_q_score": 0.4,
"mean_steps": 2518.7
},
{
"variant": "codegen_success",
"completed": 20,
"successes": 1,
"success_rate": 0.05,
"mean_q_score": 0.05,
"mean_steps": 3088.05
},
{
"variant": "mix_h50_c50",
"completed": 20,
"successes": 1,
"success_rate": 0.05,
"mean_q_score": 0.05,
"mean_steps": 3082.9
}
]