modify value_outputs_unwrap method

[whatchang]

First, thank you for releasing your awesome paper and code.

I found one problem when I ran your Quick Start code that is written in README.
(That is below)

import argparse
from tot.methods.bfs import solve
from tot.tasks.game24 import Game24Task

args = argparse.Namespace(backend='gpt-4', temperature=0.7, task='game24', naive_run=False, prompt_sample=None, method_generate='propose', method_evaluate='value', method_select='greedy', n_generate_sample=1, n_evaluate_sample=3, n_select_sample=5)

task = Game24Task()
ys, infos = solve(args, task, 900)
print(ys[0])

The problematic line of code I found is shown below.
(Thas is in src->tot->tasks->game24.py line 91)

sum(value * value_names.count(name) for name, value in value_map.items())

In the above code, the problem was caused by the value_names being in the list type rather than a string.

As a result, we can see that the first string of value_names does not count even though it has 'sure' in it, as shown in the WATCH part of the figure below.

This also happens when n_evaluate_sample=1.

So, I modified the code to look like this.

sum(value * target_value.count(name) for name, value in value_map.items() for target_value in value_names)

Thank you for reading and have a great day!

[ysymyth]

thanks! i think the problem is, what if target_value is a string like "sure sure likely impossible" or "i am sure i do not know"? exact match is more conservative but maybe less risky