open_dutch_llm_leaderboard / evals /truthfulqa /truthfulqa_nl_Llama-2-13b-chat-dutch.json
Bram Vanroy
add evaluations
4a78e88
raw
history blame
574 Bytes
{
"results": {
"truthfulqa_nl": {
"mc1": 0.31337579617834393,
"mc1_stderr": 0.016566632870878895,
"mc2": 0.43607766122269953,
"mc2_stderr": 0.015030932806527442
}
},
"versions": {
"truthfulqa_nl": 1
},
"config": {
"model": "hf-auto",
"model_args": "pretrained=BramVanroy/Llama-2-13b-chat-dutch,use_accelerate=True,device_map_option=auto,dtype=bfloat16,load_in_8bit=True",
"batch_size": 2,
"device": "cuda",
"no_cache": false,
"limit": null,
"bootstrap_iters": 100000,
"description_dict": {}
}
}