tree-of-thought-prompting

Name: tree-of-thought-prompting
Author: dave1010

820stars78forks

How to install tree-of-thought-prompting?

dave1010/tree-of-thought-prompting

$git clone https://github.com/dave1010/tree-of-thought-prompting

Installs into the current project.

Summary

Tree-of-Thought (ToT) Prompting, a fresh technique borrowing ideas from the Tree-of-Thoughts framework, broadens and enhances the well-established Chain-of-Thought prompting concept, thereby enabling Large Language Models, like ChatGPT, to exhibit superior reasoning abilities. Th

Files · 1

View on GitHub

README.md

1	# Using Tree-of-Thought Prompting to boost ChatGPT's reasoning
2
3	[![DOI](https://zenodo.org/badge/DOI/10.5281/zenodo.10323452.svg)](https://doi.org/10.5281/zenodo.10323452)
4
5
6	## Abstract
7
8	Tree-of-Thought (ToT) Prompting, a fresh technique borrowing ideas from the Tree-of-Thoughts framework, broadens and enhances the well-established Chain-of-Thought prompting concept, thereby enabling Large Language Models, like ChatGPT, to exhibit superior reasoning abilities. This Tree-of-Thought Prompting technique permits Large Language Models to rectify their errors autonomously while progressively accumulating knowledge.
9
10	In one example, a ToT prompt improves ChatGPT 3.5's reasoning ability to answer a question that could previously only be answered by ChatGPT 4.
11
12	## Complex questions for LLMs
13
14	Large Language Models (LLMs) often struggle to answer certain types of complex questions. Here's a representative example:
15
16	Bob is in the living room.
17	He walks to the kitchen, carrying a cup.
18	He puts a ball in the cup and carries the cup to the bedroom.
19	He turns the cup upside down, then walks to the garden.
20	He puts the cup down in the garden, then walks to the garage.
21	Where is the ball?
22
23	The correct answer is that the ball is in the bedroom. Although LLMs are non-deterministic, the responses I received during several attempts were as follows:
24
25	* ChatGPT 3.5: The ball is in the garden. ❌
26	* ChatGPT 4: The ball is in the bedroom. [followed by an explanation] ✅
27
28	## Chain-of-Thought prompting
29
30	A widely recognized concept known as "Chain-of-Thought prompting" (CoT for short) exists, which encourages Large Language Models (LLMs) to elucidate their thought process, thereby increasing the likelihood of a correct response. The roots of this concept can be traced back to a paper from January 2022 titled: [Chain-of-Thought Prompting Elicits Reasoning in Large Language Models](https://arxiv.org/abs/2201.11903).
31
32	At its core, the CoT technique can be implemented by adding something akin to the following to a prompt:
33
34	Think carefully and logically, explaining your answer.
35
36	This approach enhances the performance of LLMs in answering many questions, although its effectiveness varies.
37
38	Here's an example of ChatGPT 3.5's response:
39
40	Based on the given sequence of events, we can logically determine the location of the ball. Let's analyze each step:
41	1. Bob is in the living room.
42	- No mention of the ball in the living room.
43	2. Bob walks to the kitchen, carrying a cup.
44	- No mention of the ball in the kitchen yet.
45	3. Bob puts a ball in the cup.
46	- The ball is now inside the cup.
47	4. Bob carries the cup to the bedroom.
48	- The ball is still inside the cup.
49	5. Bob turns the cup upside down.
50	- If the ball was inside the cup, it would have fallen out when Bob turned it upside down.
51	6. Bob walks to the garden.
52	- Bob left the cup in the garden, but there is no mention of the ball being removed from the cup.
53	7. Bob puts the cup down in the garden.
54	- The ball, if it was in the cup, would still be in the cup in the garden.
55	8. Bob walks to the garage.
56	- No mention of the ball being taken to the garage.
57	Based on the given information, we can conclude that the ball is still inside the cup in the garden.
58
59	* ChatGPT 3.5 CoT: The ball is in the garden. ❌
60
61	You can see how ChatGPT realised the ball fell out at step 5 but still didn't get the answer correct.
62
63	Can we do any better, without resorting to ChatGPT 4?
64
65	## Tree-of-Thought Framework
66
67	Recent weeks have seen the emergence of several papers and Python repositories introducing the concept of Tree-of-Thoughts (ToT):
68
69	* [Large Language Model Guided Tree-of-Thought](https://arxiv.org/abs/2305.08291), 15 May 2023. [Github](https://github.com/jieyilong/tree-of-thought-puzzle-solver).
70	* [Tree of Thoughts: Deliberate Problem Solving with Large Language Models](https://arxiv.org/abs/2305.10601), 17 May 2023. [Github](https://github.com/ysym

People who liked this also viewed

6 picks

Type

tree-of-thought-prompting

820stars78forks

View on GitHub

How to install tree-of-thought-prompting?

dave1010/tree-of-thought-prompting

$git clone https://github.com/dave1010/tree-of-thought-prompting

Installs into the current project.

Summary

Files · 1

View on GitHub

README.md

1	# Using Tree-of-Thought Prompting to boost ChatGPT's reasoning
2
3	[![DOI](https://zenodo.org/badge/DOI/10.5281/zenodo.10323452.svg)](https://doi.org/10.5281/zenodo.10323452)
4
5
6	## Abstract
7
8	Tree-of-Thought (ToT) Prompting, a fresh technique borrowing ideas from the Tree-of-Thoughts framework, broadens and enhances the well-established Chain-of-Thought prompting concept, thereby enabling Large Language Models, like ChatGPT, to exhibit superior reasoning abilities. This Tree-of-Thought Prompting technique permits Large Language Models to rectify their errors autonomously while progressively accumulating knowledge.
9
10	In one example, a ToT prompt improves ChatGPT 3.5's reasoning ability to answer a question that could previously only be answered by ChatGPT 4.
11
12	## Complex questions for LLMs
13
14	Large Language Models (LLMs) often struggle to answer certain types of complex questions. Here's a representative example:
15
16	Bob is in the living room.
17	He walks to the kitchen, carrying a cup.
18	He puts a ball in the cup and carries the cup to the bedroom.
19	He turns the cup upside down, then walks to the garden.
20	He puts the cup down in the garden, then walks to the garage.
21	Where is the ball?
22
23	The correct answer is that the ball is in the bedroom. Although LLMs are non-deterministic, the responses I received during several attempts were as follows:
24
25	* ChatGPT 3.5: The ball is in the garden. ❌
26	* ChatGPT 4: The ball is in the bedroom. [followed by an explanation] ✅
27
28	## Chain-of-Thought prompting
29
30	A widely recognized concept known as "Chain-of-Thought prompting" (CoT for short) exists, which encourages Large Language Models (LLMs) to elucidate their thought process, thereby increasing the likelihood of a correct response. The roots of this concept can be traced back to a paper from January 2022 titled: [Chain-of-Thought Prompting Elicits Reasoning in Large Language Models](https://arxiv.org/abs/2201.11903).
31
32	At its core, the CoT technique can be implemented by adding something akin to the following to a prompt:
33
34	Think carefully and logically, explaining your answer.
35
36	This approach enhances the performance of LLMs in answering many questions, although its effectiveness varies.
37
38	Here's an example of ChatGPT 3.5's response:
39
40	Based on the given sequence of events, we can logically determine the location of the ball. Let's analyze each step:
41	1. Bob is in the living room.
42	- No mention of the ball in the living room.
43	2. Bob walks to the kitchen, carrying a cup.
44	- No mention of the ball in the kitchen yet.
45	3. Bob puts a ball in the cup.
46	- The ball is now inside the cup.
47	4. Bob carries the cup to the bedroom.
48	- The ball is still inside the cup.
49	5. Bob turns the cup upside down.
50	- If the ball was inside the cup, it would have fallen out when Bob turned it upside down.
51	6. Bob walks to the garden.
52	- Bob left the cup in the garden, but there is no mention of the ball being removed from the cup.
53	7. Bob puts the cup down in the garden.
54	- The ball, if it was in the cup, would still be in the cup in the garden.
55	8. Bob walks to the garage.
56	- No mention of the ball being taken to the garage.
57	Based on the given information, we can conclude that the ball is still inside the cup in the garden.
58
59	* ChatGPT 3.5 CoT: The ball is in the garden. ❌
60
61	You can see how ChatGPT realised the ball fell out at step 5 but still didn't get the answer correct.
62
63	Can we do any better, without resorting to ChatGPT 4?
64
65	## Tree-of-Thought Framework
66
67	Recent weeks have seen the emergence of several papers and Python repositories introducing the concept of Tree-of-Thoughts (ToT):
68
69	* [Large Language Model Guided Tree-of-Thought](https://arxiv.org/abs/2305.08291), 15 May 2023. [Github](https://github.com/jieyilong/tree-of-thought-puzzle-solver).
70	* [Tree of Thoughts: Deliberate Problem Solving with Large Language Models](https://arxiv.org/abs/2305.10601), 17 May 2023. [Github](https://github.com/ysym

People who liked this also viewed

6 picks

Type