Update some config

This commit is contained in:
Timothyxxx
2024-01-31 23:50:45 +08:00
parent 369ead1024
commit 2292053698
6 changed files with 403 additions and 43 deletions

View File

@@ -134,7 +134,7 @@ def main(example_class, example_id):
with open(f"evaluation_examples/examples/{example_class}/{example_id}.json", "r", encoding="utf-8") as f:
example = json.load(f)
example["snapshot"] = "exp_v1"
example["snapshot"] = "exp_v5"
api_key = os.environ.get("OPENAI_API_KEY")
agent = GPT4v_Agent(api_key=api_key, instruction=example['instruction'], action_space=action_space,
@@ -168,17 +168,17 @@ if __name__ == '__main__':
"af630914-714e-4a24-a7bb-f9af687d3b91"
]
calc_list = [
"eb03d19a-b88d-4de4-8a64-ca0ac66f426b",
"0bf05a7d-b28b-44d2-955a-50b41e24012a",
"7a4e4bc8-922c-4c84-865c-25ba34136be1",
"2bd59342-0664-4ccb-ba87-79379096cc08",
"ecb0df7a-4e8d-4a03-b162-053391d3afaf",
"7efeb4b1-3d19-4762-b163-63328d66303b",
"4e6fcf72-daf3-439f-a232-c434ce416af6",
"6054afcb-5bab-4702-90a0-b259b5d3217c",
"abed40dc-063f-4598-8ba5-9fe749c0615d",
"01b269ae-2111-4a07-81fd-3fcd711993b0",
"8b1ce5f2-59d2-4dcc-b0b0-666a714b9a14",
# "eb03d19a-b88d-4de4-8a64-ca0ac66f426b",
# "0bf05a7d-b28b-44d2-955a-50b41e24012a",
# "7a4e4bc8-922c-4c84-865c-25ba34136be1",
# "2bd59342-0664-4ccb-ba87-79379096cc08",
# "ecb0df7a-4e8d-4a03-b162-053391d3afaf",
# "7efeb4b1-3d19-4762-b163-63328d66303b",
# "4e6fcf72-daf3-439f-a232-c434ce416af6",
# "6054afcb-5bab-4702-90a0-b259b5d3217c",
# "abed40dc-063f-4598-8ba5-9fe749c0615d",
# "01b269ae-2111-4a07-81fd-3fcd711993b0",
# "8b1ce5f2-59d2-4dcc-b0b0-666a714b9a14",
"0cecd4f3-74de-457b-ba94-29ad6b5dafb6",
"4188d3a4-077d-46b7-9c86-23e1a036f6c1",
"51b11269-2ca8-4b2a-9163-f21758420e78",
@@ -197,5 +197,97 @@ if __name__ == '__main__':
"4f07fbe9-70de-4927-a4d5-bb28bc12c52c",
]
for example_id in calc_list:
main("libreoffice_calc", example_id)
# for example_id in calc_list:
# main("libreoffice_calc", example_id)
impress_list = [
# "5d901039-a89c-4bfb-967b-bf66f4df075e",
# "550ce7e7-747b-495f-b122-acdc4d0b8e54",
# "455d3c66-7dc6-4537-a39a-36d3e9119df7",
# "af23762e-2bfd-4a1d-aada-20fa8de9ce07",
# "c59742c0-4323-4b9d-8a02-723c251deaa0",
# "ef9d12bd-bcee-4ba0-a40e-918400f43ddf",
# "9ec204e4-f0a3-42f8-8458-b772a6797cab",
# "0f84bef9-9790-432e-92b7-eece357603fb",
# "ce88f674-ab7a-43da-9201-468d38539e4a",
# "3b27600c-3668-4abd-8f84-7bcdebbccbdb",
# "a097acff-6266-4291-9fbd-137af7ecd439",
# "bf4e9888-f10f-47af-8dba-76413038b73c",
"21760ecb-8f62-40d2-8d85-0cee5725cb72"
]
# for example_id in impress_list:
# main("libreoffice_impress", example_id)
# gimp_list = [
# "7a4deb26-d57d-4ea9-9a73-630f66a7b568",
# "554785e9-4523-4e7a-b8e1-8016f565f56a",
# "77b8ab4d-994f-43ac-8930-8ca087d7c4b4",
# "f4aec372-4fb0-4df5-a52b-79e0e2a5d6ce",
# "d52d6308-ec58-42b7-a2c9-de80e4837b2b",
# "2a729ded-3296-423d-aec4-7dd55ed5fbb3",
# "b148e375-fe0b-4bec-90e7-38632b0d73c2",
# "a746add2-cab0-4740-ac36-c3769d9bfb46",
# "7b7617bd-57cc-468e-9c91-40c4ec2bcb3d",
# "d16c99dc-2a1e-46f2-b350-d97c86c85c15",
# "06ca5602-62ca-47f6-ad4f-da151cde54cc",
# "e2dd0213-26db-4349-abe5-d5667bfd725c",
# "f723c744-e62c-4ae6-98d1-750d3cd7d79d",
# "72f83cdc-bf76-4531-9a1b-eb893a13f8aa",
# "7767eef2-56a3-4cea-8c9f-48c070c7d65b",
# "734d6579-c07d-47a8-9ae2-13339795476b"
# ]
#
# for example_id in gimp_list:
# try:
# main("gimp", example_id)
# except Exception as e:
# logger.error("An error occurred while running the example: %s", e)
# continue
#
vs_code_list = [
"0ed39f63-6049-43d4-ba4d-5fa2fe04a951",
"53ad5833-3455-407b-bbc6-45b4c79ab8fb",
"eabc805a-bfcf-4460-b250-ac92135819f6",
"982d12a5-beab-424f-8d38-d2a48429e511",
"4e60007a-f5be-4bfc-9723-c39affa0a6d3",
"e2b5e914-ffe1-44d2-8e92-58f8c5d92bb2",
"9439a27b-18ae-42d8-9778-5f68f891805e",
"ea98c5d7-3cf9-4f9b-8ad3-366b58e0fcae",
"930fdb3b-11a8-46fe-9bac-577332e2640e",
"276cc624-87ea-4f08-ab93-f770e3790175",
"9d425400-e9b2-4424-9a4b-d4c7abac4140"
]
# for example_id in vs_code_list:
# try:
# main("vs_code", example_id)
# except Exception as e:
# logger.error("An error occurred while running the example: %s", e)
# continue
multiple_list = [
"f8cfa149-d1c1-4215-8dac-4a0932bad3c2",
"897e3b53-5d4d-444b-85cb-2cdc8a97d903",
"4e9f0faf-2ecc-4ae8-a804-28c9a75d1ddc",
"b52b40a5-ad70-4c53-b5b0-5650a8387052",
"46407397-a7d5-4c6b-92c6-dbe038b1457b",
"2b9493d7-49b8-493a-a71b-56cd1f4d6908",
"51f5801c-18b3-4f25-b0c3-02f85507a078",
"2c9fc0de-3ee7-45e1-a5df-c86206ad78b5",
"510f64c8-9bcc-4be1-8d30-638705850618",
"937087b6-f668-4ba6-9110-60682ee33441",
"ee9a3c83-f437-4879-8918-be5efbb9fac7",
"3680a5ee-6870-426a-a997-eba929a0d25c",
"e135df7c-7687-4ac0-a5f0-76b74438b53e",
"58565672-7bfe-48ab-b828-db349231de6b",
"2fe4b718-3bd7-46ec-bdce-b184f5653624"
]
for example_id in multiple_list:
try:
main("multi_apps", example_id)
except Exception as e:
logger.error("An error occurred while running the example: %s", e)
continue