thought-based-reasoning

Name: thought-based-reasoning
Author: NeoLabHQ

945installs132forks

How to install thought-based-reasoning?

neolabhq/context-engineering-kit/thought-based-reasoning

$npx -y skills add NeoLabHQ/context-engineering-kit --skill thought-based-reasoning

Installs into the current project.

Summary

Use when tackling complex reasoning tasks requiring step-by-step logic, multi-step arithmetic, commonsense reasoning, symbolic manipulation, or problems where simple prompting fails - provides comprehensive guide to Chain-of-Thought and related prompting techniques (Zero-shot CoT

Files · 1

View on GitHub

SKILL.md

1	# Thought-Based Reasoning Techniques for LLMs
2
3	## Overview
4
5	Chain-of-Thought (CoT) prompting and its variants encourage LLMs to generate intermediate reasoning steps before arriving at a final answer, significantly improving performance on complex reasoning tasks. These techniques transform how models approach problems by making implicit reasoning explicit.
6
7
8	## Quick Reference
9
10	\| Technique \| When to Use \| Complexity \| Accuracy Gain \|
11	\|-----------\|-------------\|------------\|---------------\|
12	\| Zero-shot CoT \| Quick reasoning, no examples available \| Low \| +20-60% \|
13	\| Few-shot CoT \| Have good examples, consistent format needed \| Medium \| +30-70% \|
14	\| Self-Consistency \| High-stakes decisions, need confidence \| Medium \| +10-20% over CoT \|
15	\| Tree of Thoughts \| Complex problems requiring exploration \| High \| +50-70% on hard tasks \|
16	\| Least-to-Most \| Multi-step problems with subproblems \| Medium \| +30-80% \|
17	\| ReAct \| Tasks requiring external information \| Medium \| +15-35% \|
18	\| PAL \| Mathematical/computational problems \| Medium \| +10-15% \|
19	\| Reflexion \| Iterative improvement, learning from errors \| High \| +10-20% \|
20
21	---
22
23	## Core Techniques
24
25	### 1. Chain-of-Thought (CoT) Prompting
26
27	Paper: "Chain of Thought Prompting Elicits Reasoning in Large Language Models" (Wei et al., 2022)
28	Citations: 14,255+
29
30	#### When to Use
31	- Multi-step arithmetic or math word problems
32	- Commonsense reasoning requiring logical deduction
33	- Symbolic reasoning tasks
34	- When you have good exemplars showing reasoning
35
36	#### How It Works
37	Provide few-shot examples that include intermediate reasoning steps, not just question-answer pairs. The model learns to generate similar step-by-step reasoning.
38
39	#### Prompt Template
40
41	```
42	Q: Roger has 5 tennis balls. He buys 2 more cans of tennis balls. Each can has 3 tennis balls. How many tennis balls does he have now?
43	A: Roger started with 5 balls. 2 cans of 3 tennis balls each is 6 tennis balls. 5 + 6 = 11. The answer is 11.
44
45	Q: The cafeteria had 23 apples. If they used 20 to make lunch and bought 6 more, how many apples do they have?
46	A: The cafeteria had 23 apples originally. They used 20 to make lunch. So they had 23 - 20 = 3. They bought 6 more apples, so they have 3 + 6 = 9. The answer is 9.
47
48	Q: [YOUR QUESTION HERE]
49	A:
50	```
51
52	#### Strengths
53	- Significant accuracy improvements on reasoning tasks
54	- Interpretable intermediate steps
55	- Works well with large models (>100B parameters)
56
57	#### Limitations
58	- Requires crafting good exemplars
59	- Less effective on smaller models
60	- Can still make calculation errors
61
62	---
63
64	### 2. Zero-shot Chain-of-Thought
65
66	Paper: "Large Language Models are Zero-Shot Reasoners" (Kojima et al., 2022)
67	Citations: 5,985+
68
69	#### When to Use
70	- No exemplars available
71	- Quick reasoning needed
72	- General-purpose reasoning across task types
73	- Prototyping before creating few-shot examples
74
75	#### How It Works
76	Simply append "Let's think step by step" (or similar phrase) to the prompt. This triggers the model to generate reasoning steps without any examples.
77
78	#### Prompt Template
79
80	```
81	Q: A juggler can juggle 16 balls. Half of the balls are golf balls, and half of the golf balls are blue. How many blue golf balls are there?
82
83	Let's think step by step.
84	```
85
86	Alternative trigger phrases:
87	- "Let's work this out step by step to be sure we have the right answer."
88	- "Let's break this down."
89	- "Let's approach this systematically."
90	- "First, let me understand the problem..."
91
92	#### Two-Stage Approach (More Robust)
93
94	Stage 1 - Reasoning Extraction:
95	```
96	Q: [QUESTION]
97	A: Let's think step by step.
98	```
99
100	Stage 2 - Answer Extraction:
101	```
102	[REASONING FROM STAGE 1]
103	Therefore, the answer is
104	```
105
106	#### Strengths
107	- No exemplar crafting required
108	- Generalizes across task types
109	- Simple to implement
110
111	#### Limitations
112	- Less effective than few-shot CoT
113	- Can produce verbose or irrelevant reasoning
114	- Sensitive to exact phrasing
115
116	---
117
118	### 3. Self-Consistency
119
120	Paper: "Self-Consistency Improves Chain of Thought Reasoning in Language Models" (Wang et al., 2022)
121	Citations: 5,379+
122
123	#### When to Use
124	- High-stakes decisions requiring confidence
125	- Problems with multiple valid reasoning paths
126	- When you need to reduce variance in outputs
127	- Verification of reasoning correctness
128
129	#### How It Works
130	Sample multiple diverse reasoning paths, then select the most consistent answer via majority voting. The intuition: correct answers can be reached through multiple reasoning paths.
131
132	#### Prompt Template
133
134	```
135	[Use any CoT prompt - zero-shot or few-shot]
136
137	[Generate

People who liked this also viewed

6 picks

Type

thought-based-reasoning

945installs132forks

View on GitHub

How to install thought-based-reasoning?

neolabhq/context-engineering-kit/thought-based-reasoning

$npx -y skills add NeoLabHQ/context-engineering-kit --skill thought-based-reasoning

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Thought-Based Reasoning Techniques for LLMs
2
3	## Overview
4
5	Chain-of-Thought (CoT) prompting and its variants encourage LLMs to generate intermediate reasoning steps before arriving at a final answer, significantly improving performance on complex reasoning tasks. These techniques transform how models approach problems by making implicit reasoning explicit.
6
7
8	## Quick Reference
9
10	\| Technique \| When to Use \| Complexity \| Accuracy Gain \|
11	\|-----------\|-------------\|------------\|---------------\|
12	\| Zero-shot CoT \| Quick reasoning, no examples available \| Low \| +20-60% \|
13	\| Few-shot CoT \| Have good examples, consistent format needed \| Medium \| +30-70% \|
14	\| Self-Consistency \| High-stakes decisions, need confidence \| Medium \| +10-20% over CoT \|
15	\| Tree of Thoughts \| Complex problems requiring exploration \| High \| +50-70% on hard tasks \|
16	\| Least-to-Most \| Multi-step problems with subproblems \| Medium \| +30-80% \|
17	\| ReAct \| Tasks requiring external information \| Medium \| +15-35% \|
18	\| PAL \| Mathematical/computational problems \| Medium \| +10-15% \|
19	\| Reflexion \| Iterative improvement, learning from errors \| High \| +10-20% \|
20
21	---
22
23	## Core Techniques
24
25	### 1. Chain-of-Thought (CoT) Prompting
26
27	Paper: "Chain of Thought Prompting Elicits Reasoning in Large Language Models" (Wei et al., 2022)
28	Citations: 14,255+
29
30	#### When to Use
31	- Multi-step arithmetic or math word problems
32	- Commonsense reasoning requiring logical deduction
33	- Symbolic reasoning tasks
34	- When you have good exemplars showing reasoning
35
36	#### How It Works
37	Provide few-shot examples that include intermediate reasoning steps, not just question-answer pairs. The model learns to generate similar step-by-step reasoning.
38
39	#### Prompt Template
40
41	```
42	Q: Roger has 5 tennis balls. He buys 2 more cans of tennis balls. Each can has 3 tennis balls. How many tennis balls does he have now?
43	A: Roger started with 5 balls. 2 cans of 3 tennis balls each is 6 tennis balls. 5 + 6 = 11. The answer is 11.
44
45	Q: The cafeteria had 23 apples. If they used 20 to make lunch and bought 6 more, how many apples do they have?
46	A: The cafeteria had 23 apples originally. They used 20 to make lunch. So they had 23 - 20 = 3. They bought 6 more apples, so they have 3 + 6 = 9. The answer is 9.
47
48	Q: [YOUR QUESTION HERE]
49	A:
50	```
51
52	#### Strengths
53	- Significant accuracy improvements on reasoning tasks
54	- Interpretable intermediate steps
55	- Works well with large models (>100B parameters)
56
57	#### Limitations
58	- Requires crafting good exemplars
59	- Less effective on smaller models
60	- Can still make calculation errors
61
62	---
63
64	### 2. Zero-shot Chain-of-Thought
65
66	Paper: "Large Language Models are Zero-Shot Reasoners" (Kojima et al., 2022)
67	Citations: 5,985+
68
69	#### When to Use
70	- No exemplars available
71	- Quick reasoning needed
72	- General-purpose reasoning across task types
73	- Prototyping before creating few-shot examples
74
75	#### How It Works
76	Simply append "Let's think step by step" (or similar phrase) to the prompt. This triggers the model to generate reasoning steps without any examples.
77
78	#### Prompt Template
79
80	```
81	Q: A juggler can juggle 16 balls. Half of the balls are golf balls, and half of the golf balls are blue. How many blue golf balls are there?
82
83	Let's think step by step.
84	```
85
86	Alternative trigger phrases:
87	- "Let's work this out step by step to be sure we have the right answer."
88	- "Let's break this down."
89	- "Let's approach this systematically."
90	- "First, let me understand the problem..."
91
92	#### Two-Stage Approach (More Robust)
93
94	Stage 1 - Reasoning Extraction:
95	```
96	Q: [QUESTION]
97	A: Let's think step by step.
98	```
99
100	Stage 2 - Answer Extraction:
101	```
102	[REASONING FROM STAGE 1]
103	Therefore, the answer is
104	```
105
106	#### Strengths
107	- No exemplar crafting required
108	- Generalizes across task types
109	- Simple to implement
110
111	#### Limitations
112	- Less effective than few-shot CoT
113	- Can produce verbose or irrelevant reasoning
114	- Sensitive to exact phrasing
115
116	---
117
118	### 3. Self-Consistency
119
120	Paper: "Self-Consistency Improves Chain of Thought Reasoning in Language Models" (Wang et al., 2022)
121	Citations: 5,379+
122
123	#### When to Use
124	- High-stakes decisions requiring confidence
125	- Problems with multiple valid reasoning paths
126	- When you need to reduce variance in outputs
127	- Verification of reasoning correctness
128
129	#### How It Works
130	Sample multiple diverse reasoning paths, then select the most consistent answer via majority voting. The intuition: correct answers can be reached through multiple reasoning paths.
131
132	#### Prompt Template
133
134	```
135	[Use any CoT prompt - zero-shot or few-shot]
136
137	[Generate

People who liked this also viewed

6 picks

Type