You cannot select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
reflexion-human-eval/simple.py

43 lines
1.6 KiB
Python

from utils import enumerate_resume, write_jsonl
from executors import executor_factory
from generators import generator_factory
from typing import List
SIMPLE_COMPLETION_INSTRUCTION = "# Write the body of this function only."
SIMPLE_CHAT_INSTRUCTION = "You are CodexGPT. You will be given a function signature and docstring. You should fill in the following text of the missing function body. For example, the first line of the completion should have 4 spaces for the indendation so that it fits syntactically with the preceding signature."
def run_simple(
dataset: List[dict],
model: str,
language: str,
pass_at_k: int,
log_path: str,
verbose: bool
) -> None:
exe = executor_factory(language)
gen = generator_factory(language)
num_items = len(dataset)
num_success = 0
for i, item in enumerate_resume(dataset, log_path):
cur_pass = 0
is_solved = False
cur_func_impl = ""
while cur_pass < pass_at_k:
cur_func_impl = gen.func_impl(item["prompt"], model, "simple")
assert isinstance(cur_func_impl, str)
is_passing = exe.evaluate(item["entry_point"], cur_func_impl, item["test"], timeout=10)
if is_passing:
is_solved = True
num_success += 1
break
cur_pass += 1
item["solution"] = cur_func_impl
item["is_solved"] = is_solved
write_jsonl(log_path, [item], append=True)
if verbose:
print(f'completed {i+1}/{num_items}: acc = {round(num_success/(i+1), 2)}')