open_dutch_llm_leaderboard / evals /truthfulqa /truthfulqa_nl_Mistral-7B-v0.1.json
Bram Vanroy
add Mistral evals
38b8e33
raw
history blame
560 Bytes
{
"results": {
"truthfulqa_nl": {
"mc1": 0.30828025477707005,
"mc1_stderr": 0.01649225000783665,
"mc2": 0.4516092295846247,
"mc2_stderr": 0.015055722136057215
}
},
"versions": {
"truthfulqa_nl": 1
},
"config": {
"model": "hf-auto",
"model_args": "pretrained=mistralai/Mistral-7B-v0.1,use_accelerate=True,device_map_option=auto,dtype=auto,load_in_8bit=True",
"batch_size": 2,
"device": "cuda",
"no_cache": false,
"limit": null,
"bootstrap_iters": 100000,
"description_dict": {}
}
}