Metis-1.5-think / eval_results.json
GiuliannoV's picture
Add raw eval results
88760bf verified
Raw
History Blame Contribute Delete
2.35 kB
{
"protocol": "0-shot; custom JAX loglikelihood harness (acc_norm=length-normalized); GSM8K greedy gen w/ chat template; full test splits; training data benchmark-decontaminated",
"Metis-1.5-base": {
"arc_easy": {
"acc": 0.4621,
"acc_norm": 0.4133,
"primary": 0.4133,
"n": 2376
},
"arc_challenge": {
"acc": 0.2372,
"acc_norm": 0.2585,
"primary": 0.2585,
"n": 1172
},
"hellaswag": {
"acc": 0.2917,
"acc_norm": 0.3044,
"primary": 0.3044,
"n": 10042
},
"piqa": {
"acc": 0.5544,
"acc_norm": 0.5473,
"primary": 0.5473,
"n": 1838
},
"winogrande": {
"acc": 0.5146,
"acc_norm": null,
"primary": 0.5146,
"n": 1267
},
"openbookqa": {
"acc": 0.18,
"acc_norm": 0.296,
"primary": 0.296,
"n": 500
},
"boolq": {
"acc": 0.4765,
"acc_norm": null,
"primary": 0.4765,
"n": 3270
},
"mmlu": {
"acc": 0.2359,
"acc_norm": null,
"primary": 0.2359,
"n": 14042
},
"lambada": {
"acc": 0.1698,
"acc_norm": null,
"primary": 0.1698,
"n": 5153
}
},
"Metis-1.5-think": {
"arc_easy": {
"acc": 0.4419,
"acc_norm": 0.4171,
"primary": 0.4171,
"n": 2376
},
"arc_challenge": {
"acc": 0.25,
"acc_norm": 0.2824,
"primary": 0.2824,
"n": 1172
},
"hellaswag": {
"acc": 0.2945,
"acc_norm": 0.31,
"primary": 0.31,
"n": 10042
},
"piqa": {
"acc": 0.5593,
"acc_norm": 0.5457,
"primary": 0.5457,
"n": 1838
},
"winogrande": {
"acc": 0.5178,
"acc_norm": null,
"primary": 0.5178,
"n": 1267
},
"openbookqa": {
"acc": 0.208,
"acc_norm": 0.286,
"primary": 0.286,
"n": 500
},
"boolq": {
"acc": 0.5719,
"acc_norm": null,
"primary": 0.5719,
"n": 3270
},
"mmlu": {
"acc": 0.2329,
"acc_norm": null,
"primary": 0.2329,
"n": 14042
},
"lambada": {
"acc": 0.1731,
"acc_norm": null,
"primary": 0.1731,
"n": 5153
},
"gsm8k": {
"acc": 0.0758,
"acc_norm": null,
"primary": 0.0758,
"n": 1319
}
}
}