Spaces:
Runtime error
Runtime error
File size: 1,773 Bytes
5d28865 728a44a 5d28865 728a44a 5d28865 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 |
import os
from huggingface_hub import Repository
H4_TOKEN = os.environ.get("H4_TOKEN", None)
def get_all_requested_models(requested_models_dir):
depth = 1
file_names = []
for root, dirs, files in os.walk(requested_models_dir):
current_depth = root.count(os.sep) - requested_models_dir.count(os.sep)
if current_depth == depth:
file_names.extend([os.path.join(root, file) for file in files])
return set([file_name.lower().split("eval_requests/")[1] for file_name in file_names])
def load_all_info_from_hub(HUMAN_EVAL_REPO, GPT_4_EVAL_REPO):
human_eval_repo = None
if H4_TOKEN and not os.path.isdir("./human_evals"):
print("Pulling human evaluation repo")
human_eval_repo = Repository(
local_dir="./human_evals/",
clone_from=HUMAN_EVAL_REPO,
use_auth_token=H4_TOKEN,
repo_type="dataset",
)
human_eval_repo.git_pull()
gpt_4_eval_repo = None
if H4_TOKEN and not os.path.isdir("./gpt_4_evals"):
print("Pulling GPT-4 evaluation repo")
gpt_4_eval_repo = Repository(
local_dir="./gpt_4_evals/",
clone_from=GPT_4_EVAL_REPO,
use_auth_token=H4_TOKEN,
repo_type="dataset",
)
gpt_4_eval_repo.git_pull()
return human_eval_repo, gpt_4_eval_repo
#def load_results(model, benchmark, metric):
# file_path = os.path.join("autoevals", model, f"{model}-eval_{benchmark}.json")
# if not os.path.exists(file_path):
# return 0.0, None
# with open(file_path) as fp:
# data = json.load(fp)
# accs = np.array([v[metric] for k, v in data["results"].items()])
# mean_acc = np.mean(accs)
# return mean_acc, data["config"]["model_args"]
|