25 lines
659 B
JSON
25 lines
659 B
JSON
|
|
{
|
||
|
|
"results": {
|
||
|
|
"hellaswag": {
|
||
|
|
"acc": 0.5541724756024696,
|
||
|
|
"acc_stderr": 0.004960408362133245,
|
||
|
|
"acc_norm": 0.7353116908982275,
|
||
|
|
"acc_norm_stderr": 0.00440265476726963
|
||
|
|
}
|
||
|
|
},
|
||
|
|
"versions": {
|
||
|
|
"hellaswag": 0
|
||
|
|
},
|
||
|
|
"config": {
|
||
|
|
"model": "sparseml",
|
||
|
|
"model_args": "pretrained=/network/alexandre/research/cerebras/llama2_7B_sparse50_45B_retrained/ultrachat200k/llama2_7B_45B_sparse50_LR2e-4_GC2_E2/training,dtype=bfloat16",
|
||
|
|
"num_fewshot": 10,
|
||
|
|
"batch_size": "16",
|
||
|
|
"batch_sizes": [],
|
||
|
|
"device": "cuda:6",
|
||
|
|
"no_cache": true,
|
||
|
|
"limit": null,
|
||
|
|
"bootstrap_iters": 100000,
|
||
|
|
"description_dict": {}
|
||
|
|
}
|
||
|
|
}
|