Note: https://github.com/kyegomez/tree-of-thoughts is **NOT the correct implementation** to replicate paper results. In fact, people have reported that his code [cannot](https://github.com/kyegomez/tree-of-thoughts/issues/52) [properly](https://github.com/kyegomez/tree-of-thoughts/issues/41) [run](https://github.com/kyegomez/tree-of-thoughts/issues/60), and is [probably automatically generated by ChatGPT](fake.png), and [kyegomez has done so for other popular ML methods](https://twitter.com/qbitium/status/1663954096741814272), while **intentionally refusing to link to official implementations** for his own interests (see https://github.com/kyegomez/tree-of-thoughts/issues/54, https://github.com/kyegomez/tree-of-thoughts/issues/55, https://github.com/kyegomez/tree-of-thoughts/issues/56).
Unfortunately, Google/Github searches go to kyegomez's malicious repo by default as it has more stars. **Please DE-STAR his repo and STAR this to help other people avoid being misled, thanks!**
Official implementation for paper [Tree of Thoughts: Deliberate Problem Solving with Large Language Models](https://arxiv.org/abs/2305.10601) with code, prompts, model outputs.
You need to first have an OpenAI API key and store it in the environment variable ``OPENAI_API_KEY`` (see [here](https://help.openai.com/en/articles/5112595-best-practices-for-api-key-safety)). If you use custom base url, set it by environment variable ``OPENAI_API_BASE`` (e.g. https://api.openai.com/v1).
Run experiments via ``sh scripts/{game24, text, crosswords}/{standard_sampling, cot_sampling, bfs}.sh``, except in crosswords we use a DFS algorithm for ToT, which can be run via ``scripts/crosswords/search_crosswords-dfs.ipynb``.
The very simple ``run.py`` implements the ToT + BFS algorithm, as well as the naive IO/CoT sampling. Some key arguments:
- ``--naive_run``: if True, run naive IO/CoT sampling instead of ToT + BFS.
- ``--method_generate`` (choices=[``sample``, ``propose``]): thought generator, whether to sample independent thoughts (used in Creative Writing) or propose sequential thoughts (used in Game of 24)
- ``--method_evaluate`` (choices=[``value``, ``vote``]): state evaluator, whether to use the value states independently (used in Game of 24) or vote on states together (used in Creative Writing)
- ``--n_generate_sample``: number of times to prompt for thought generation
- ``--n_evaluate_sample``: number of times to prompt for state evaluation
- ``--n_select_sample``: number of states to keep from each step (i.e. ``b`` in the paper's ToT + BFS algorithm)
``logs/`` contains all the trajectories from the paper's experiments, except for ``logs/game24/gpt-4_0.7_propose1_value3_greedy5_start900_end1000.json`` which was reproduced after the paper (as the original experiment was done in a notebook) and achieved a 69\% score instead of the original 74\% score due to randomness in GPT decoding. We hope to aggregate multiple runs in the future to account for sampling randomness and update the paper, but this shouldn't affect the main conclusions of the paper.
