Files
l20-edu-135m/eval_results/current_vs_smollm_summary.json

74 lines
2.1 KiB
JSON
Raw Normal View History

{
"baselines": [
"smollm-135m",
"smollm2-135m"
],
"candidate": "ours-stage2",
"means": {
"ours-stage2": 0.43113943348865447,
"smollm-135m": 0.4767124569279318,
"smollm2-135m": 0.4916822407842775
},
"tasks": [
{
"display": "ARC-Challenge",
"gap_vs_smollm-135m": -0.0068259385665528916,
"gap_vs_smollm2-135m": -0.016211604095563104,
"metric": "acc_norm,none",
"ours-stage2": 0.28071672354948807,
"smollm-135m": 0.28754266211604096,
"smollm2-135m": 0.29692832764505117,
"task": "arc_challenge"
},
{
"display": "ARC-Easy",
"gap_vs_smollm-135m": -0.05513468013468015,
"gap_vs_smollm2-135m": -0.07954545454545459,
"metric": "acc_norm,none",
"ours-stage2": 0.5058922558922558,
"smollm-135m": 0.561026936026936,
"smollm2-135m": 0.5854377104377104,
"task": "arc_easy"
},
{
"display": "HellaSwag",
"gap_vs_smollm-135m": -0.10227046405098589,
"gap_vs_smollm2-135m": -0.10585540728938458,
"metric": "acc_norm,none",
"ours-stage2": 0.32423819956184025,
"smollm-135m": 0.42650866361282613,
"smollm2-135m": 0.43009360685122483,
"task": "hellaswag"
},
{
"display": "LAMBADA OpenAI",
"gap_vs_smollm-135m": null,
"gap_vs_smollm2-135m": null,
"metric": "acc,none",
"ours-stage2": null,
"smollm-135m": 0.3757034737046381,
"smollm2-135m": 0.4288763826896953,
"task": "lambada_openai"
},
{
"display": "PIQA",
"gap_vs_smollm-135m": -0.06855277475516863,
"gap_vs_smollm2-135m": -0.0701849836779107,
"metric": "acc_norm,none",
"ours-stage2": 0.6137105549510338,
"smollm-135m": 0.6822633297062024,
"smollm2-135m": 0.6838955386289445,
"task": "piqa"
},
{
"display": "WinoGrande",
"gap_vs_smollm-135m": null,
"gap_vs_smollm2-135m": null,
"metric": "acc,none",
"ours-stage2": null,
"smollm-135m": 0.5272296764009471,
"smollm2-135m": 0.5248618784530387,
"task": "winogrande"
}
]
}