2406.04271

# Buffer of Thoughts: Thought-Augmented Reasoning with Large Language Models > Equal Contribution. 🖂  yangling0818@163.com Abstract We introduce Buffer of Thoughts (BoT), a novel and versatile thought-augmented reasoning approach for enhancing accuracy, efficiency and robustness of large language models (LLMs). Specifically, we propose meta-buffer to store a series of informative high-level thoughts, namely thought-template, distilled from the problem-solving processes across various tasks. Then for each problem, we retrieve a relevant thought-template and adaptively instantiate it with specific reasoning structures to conduct efficient reasoning. To guarantee the scalability and stability, we further propose buffer-manager to dynamically update the meta-buffer, thus enhancing the capacity of meta-buffer as more tasks are solved. We conduct extensive experiments on 10 challenging reasoning-intensive tasks, and achieve significant performance improvements over previous SOTA methods: 11% on Game of 24, 20% on Geometric Shapes and 51% on Checkmate-in-One. Further analysis demonstrate the superior generalization ability and model robustness of our BoT, while requiring only 12% of the cost of multi-query prompting methods (e.g., tree/graph of thoughts) on average. Notably, we find that our Llama3-8B + BoT has the potential to surpass Llama3-70B model. Our project is available at https://github.com/YangLing0818/buffer-of-thought-llm 1 Introduction A series of Large Language Models (LLMs) [1, 2, 3, 4, 5] like GPT-4 [3], PaLM [2] and LLaMA [6, 7] have showcased the impressive performance in various reasoning tasks. In addition to scaling up the model size to improve the reasoning performance, there are more effective prompting methods that further enhance the functionality and performance of LLMs. We divide these methods into two categories: (i) single-query reasoning: these methods [8, 9, 10] usually focus on prompt engineering and their reasoning process can be finished within a single query, such as CoT [8] that appends the input query with ’Let’s think step by step’ to produce rationales for increasing reasoning accuracy, and Few-shot Prompting [11, 12, 9, 13] which provides task-relevant exemplars to assist the answer generation; (ii) multi-query reasoning: these methods [14, 15] focus on leveraging multiple LLM queries to elicit different plausible reasoning paths, thus decomposing a complex problem into a series of simpler sub-problems, such as Least-to-Most [16], ToT [14] and GoT [17]. However, both kinds of methods face some limitations: (1) single-query reasoning usually requires prior assumption or relevant exemplars of reasoning process, which makes it impractical to manually design them task by task, thus lacking universality and generalization; (2) Due to the recursive expansion of reasoning paths, multi-query reasoning is usually computationally-intensive when finding a unique intrinsic structure underlying the reasoning process for each specific task; (3) Both single-query and multi-query reasoning processes are limited by their designed exemplars and reasoning structures, and they neglect to derive general and high-level guidelines or thoughts from previously-completed tasks, which are informative for improving efficiency and accuracy when solving similar problems. <details> <summary>x1.png Details</summary>  ### Visual Description ## Diagram: LLM Reasoning Framework Architecture ### Overview The diagram illustrates three progressive approaches to utilizing Large Language Models (LLMs) for reasoning tasks, showing increasing complexity from basic prompting to advanced thought management systems. Three main sections are depicted: Single-Query, Multi-Query, and Buffer of Thoughts (BoT), each with distinct components and evaluation metrics. ### Components/Axes 1. **Single-Query Section** - Input query → LLM → Manual Prompt (CoT/Few-shot) → Reasoning → Output - Evaluation: Accuracy (↓ sad face) 2. **Multi-Query Section** - Input query → LLM → Pre-defined Query Structure (ToT/GoT) → Thought Expansion → Reasoning (loop) - Evaluation: Efficiency (↓ sad face) 3. **Buffer of Thoughts (BoT) Section** - Input query → LLM → Problem Distiller → Instantiated Reasoning → Output - Meta Buffer ↔ Thought Distillation/Update ↔ Problem Distiller - Thought Retrieval from Meta Buffer - Evaluation: Accuracy & Efficiency (↑ happy face) ### Detailed Analysis - **Single-Query**: Linear flow with explicit manual prompting. Output quality measured by accuracy alone. - **Multi-Query**: Introduces iterative thought expansion with predefined structures. Focus on efficiency optimization. - **BoT**: Most complex architecture with: - Problem Distiller generating high-level thoughts - Meta Buffer acting as persistent memory - Bidirectional feedback loops for thought refinement - Combined evaluation of accuracy and efficiency ### Key Observations 1. Evaluation metrics evolve from single-dimensional (accuracy) to dual-dimensional (accuracy + efficiency) 2. Complexity increases from linear (Single-Query) to cyclical (Multi-Query) to networked (BoT) 3. Meta Buffer introduces persistent memory capabilities absent in earlier approaches 4. Thought distillation mechanism enables continuous improvement in later stages ### Interpretation The diagram demonstrates a progression from basic LLM prompting to sophisticated cognitive architectures. The introduction of the Meta Buffer in the BoT approach suggests a paradigm shift toward: 1. **Memory-augmented reasoning**: Persistent storage of intermediate thoughts enables deeper analysis 2. **Self-improving systems**: Feedback loops between distillation and reasoning components create iterative enhancement 3. **Multi-objective optimization**: Later stages balance both accuracy and efficiency, unlike earlier single-metric approaches The sad/happy face indicators visually reinforce this progression, showing diminishing returns in early stages versus compounding benefits in advanced implementations. The BoT architecture appears to address fundamental LLM limitations in context retention and reasoning depth through its memory buffer and distillation mechanisms. </details> Figure 1: Comparison between single-query [8, 11], multi-query [14, 17], and (c) our BoT methods. To address these limitations, we propose Buffer of Thoughts (BoT), a novel and versatile thought-augmented reasoning framework aimed at enhancing reasoning accuracy, efficiency and robustness of LLMs across various tasks. Specifically, we design meta-buffer, a lightweight library housing a series of universal high-level thoughts (thought-template), which are distilled from different problem-solving processes and can be shared across tasks. Then, for each problem, we retrieve a relevant thought-template and instantiate it with specific reasoning structure for efficient thought-augmented reasoning. In order to guarantee the scalability and stability of our BoT, we further propose buffer-manager to dynamically update the meta-buffer, which effectively enhances the capacity of meta-buffer as more tasks are solved. Our method has three critical advantages: (i) Accuracy Improvement: With the shared thought-templates, we can adaptively instantiate high-level thoughts for addressing different tasks, eliminating the need to build reasoning structures from scratch, thereby improving reasoning accuracy. (ii) Reasoning Efficiency: Our thought-augmented reasoning could directly leverage informative historical reasoning structures to conduct reasoning without complex multi-query processes, thus improving reasoning efficiency. (iii) Model Robustness: The procedure from thought retrieval to thought instantiation is just like the human thought process, enabling LLMs to address similar problems in a consistent way, thus significantly enhancing the model robustness of our method. Our empirical studies demonstrate that Buffer of Thoughts significantly improves precision, efficiency, and robustness over a diverse array of tasks. Here, we summarize our contributions as follows: 1. We propose a novel thought-augmented reasoning framework Buffer of Thoughts (BoT) for improving the accuracy, efficiency and robustness of LLM-based reasoning. 1. We propose meta-buffer for store informative high-level thoughts distilled from different problems, and adaptively instantiate each thought template to address each specific task. 1. We design buffer-manager to distill thought-templates from various solutions, and is continually improves the capacity of meta-buffer as more tasks are solved. 1. We conduct extensive experiments on 10 challenging reasoning-intensive tasks. Our BoT achieves significant performance improvements over previous SOTA methods: 11% on Game of 24, 20% on Geometric Shapes and 51% on Checkmate-in-One, while requiring only 12% of the cost of multi-query prompting methods on average. 2 Related Work and Discussions Retrieval-Augmented Language Models The retrieval-augmented (Large) Language Model is introduced as a solution to mitigate the phenomenon of hallucination and enhance the output quality of language models [18, 19, 20, 21, 22]. When presented with an input question, the retrieval-augmented LLM first queries an external database with billion-level tokens [23] for retrieving a subset of the text corpus to help generating the final answer. Notably, the retrieval-augmented LLM achieves superior question-answering performance using fewer parameters compared to conventional LLMs [19], and it has found application across various downstream tasks [24, 25, 26], including multi-modal generation [24, 22, 23, 25] and biomedical applications [26, 27]. In this paper, we construct a novel category of retrieval database, termed meta-buffer, which contains a series of high-level thoughts rather than specific instances, aiming to universally address various tasks for LLM-based reasoning. Prompt-based Reasoning with Large Language Models Prompting techniques have significantly enahnced the arithmetic and commonsense reasoning capabilities of LLMs. Chain-of-Thought (CoT) prompting [8] and its variants [28, 29, 30], such as Least-to-Most [16], Decomposed Prompting [31], and Auto-CoT [13] —prompt LLMs to break down complex questions into simpler subtasks and systematically solve them before summarizing a final answer. Numerous studies [32, 33, 34, 35, 36, 37] have demonstrated the effectiveness of these prompting methods across a wide range of tasks and benchmarks. Innovations like Tree-of-Thought [14] and Graph-of-Thought [17], have further advanced this field by exploring dynamic, non-linear reasoning pathways to expand heuristic capabilities of LLMs [38, 39]. However, they suffer from increased resource demands and greater time complexity, depend on manual prompt crafting, and are often tailored to specific task types. Recent meta prompting methods [15, 40] utilize a same task-agnostic form of prompting for various tasks and recursively guide a single LLM to adaptively addressing different input queries. Nevertheless, such a long meta prompt may require a considerable context window, and these methods fail to leverage historical informative guidelines or thoughts for potential similar tasks. Analogical Reasoning Analogical reasoning is a useful technique for natural language reasoning [41, 42, 43, 44, 45]. Recent works demonstrate that LLMs can perform analogical reasoning just like humans [46, 47, 12, 48, 49]. For example, Analogical Prompting [12] and Thought Propagation [48] prompt LLMs to self-generate a set of analogous problems, and then utilize the results of analogous problems to produce a solution for input problem. However, the specific solutions for self-explored problems may introduce additional noise and cause error accumulation. Recent Thought-Retriever [49] uses the intermediate thoughts generated when solving past user to address analogous queries, but it only focuses on textual comprehension/generation instead of general reasoning problems. Thus, a more high-level and general analogical approach for LLM complex reasoning is still lacking. 3 Buffer of Thoughts Overview of Buffer of Thoughts In this section, we introduce our Buffer of Thoughts in detail and we also illustrate our core thought-augmented reasoning process in Figure 2. Given a specific task, we utilize our problem-distiller (Section 3.1) to extract critical task-specific information along with relevant constraints. Based on the distilled information, we search in meta-buffer (Section 3.2) that contains a series of high-level thoughts (thought-template) and retrieve a most relevant thought-template for the task. Subsequently, we instantiate the retrieved thought-template with more task-specific reasoning structures and conduct reasoning process. Finally, we employs a buffer-manager (Section 3.3) for summarizing the whole problem-solving process and distilling high-level thoughts for imcreasing the capacity of meta-buffer. <details> <summary>x2.png Details</summary>  ### Visual Description ## Flowchart: Shopping Mall Pricing Strategy Optimization ### Overview The diagram illustrates a structured problem-solving process to determine optimal price reduction for branded shirts in a shopping mall. It combines mathematical reasoning (quadratic equations) with business strategy to achieve a target daily profit of 1200 yuan while reducing inventory. ### Components/Axes 1. **Problem Statement** (Top-left): - Current sales: 20 shirts/day - Current profit: 40 yuan/shirt - Goal: 1200 yuan/day profit with inventory reduction - Key relationship: 1 yuan price decrease → 2 additional shirts sold/day 2. **Chain-of-Thought Process** (Left column): - Step 1: Current profit calculation (20 × 40 = 800 yuan) - Step 2: Additional profit potential (2 × 40 = 80 yuan) - Step 3: Price reduction calculation (320 ÷ 2 = 160 yuan) - Final recommendation: Reduce price by 160 yuan (highlighted in red) 3. **Thought Templates** (Central): - **Template T₁**: Quadratic equation solving methodology - Discriminant calculation (D = b² - 4ac) - Root nature determination (D ≥ 0 → two real roots) - **Template Tₙ**: Generic algorithmic structure - Element processing (#process_element) - Element combination (#combine_elements) - Condition checking (#check_condition) 4. **Plan-and-Solve Section** (Bottom-left): - Step 1: Current profit verification - Step 2: Profit equation formulation - Step 3: Quadratic equation derivation (p² - 30p - 200 = 0) - Step 4: Solution (p ≈ 35.6155 yuan) 5. **Instantiated Reasoning** (Right column): - Variable definitions: - p = 40 yuan (original profit) - q = 20 shirts (original quantity) - x = price reduction variable - Final equation: x² - 30x + 200 = 0 - Solution: x = 20 yuan (selected for inventory reduction) ### Content Details - **Mathematical Framework**: - Quadratic equation: x² - 30x + 200 = 0 - Discriminant: D = 100 (positive → two real roots) - Roots: x = 20 or x = 10 yuan - Selected solution: x = 20 yuan (marked with red checkmark) - **Business Metrics**: - Current profit: 800 yuan/day - Target profit: 1200 yuan/day - Required profit increase: 400 yuan/day - Sales increase needed: 10 shirts/day (400 ÷ 40) ### Key Observations 1. **Mathematical Validation**: - The quadratic solution confirms two valid price reduction options (10¥ and 20¥) - Higher reduction (20¥) prioritizes inventory clearance over marginal profit 2. **Business Tradeoff**: - 20¥ reduction increases sales to 40 shirts/day (20 + 2×20) - New profit: (40 - 20) × 40 = 800 yuan (still below target) - Discrepancy suggests either: - Additional sales volume needed - Alternative pricing strategy required 3. **Visual Anomalies**: - Final recommendation (160¥ reduction) conflicts with mathematical solution - Red checkmark on 20¥ solution contradicts initial 160¥ calculation ### Interpretation The diagram demonstrates a systematic approach to pricing optimization using algebraic modeling. While the mathematical solution identifies 20¥ as the optimal reduction for inventory management, the final recommendation of 160¥ appears inconsistent with the derived equations. This discrepancy suggests either: 1. A miscalculation in the initial profit requirement (320 yuan needed vs actual 400 yuan deficit) 2. An unstated constraint prioritizing inventory turnover over profit maximization 3. Potential error in the visual representation of mathematical results The process highlights the importance of cross-verifying business objectives with mathematical models, particularly when multiple optimization criteria (profit vs inventory) conflict. The quadratic framework provides a robust method for evaluating tradeoffs between price elasticity and sales volume. </details> Figure 2: Illustration of different reasoning process. Buffer of Thoughts enables large language models to tackle complex reasoning tasks through our thought-augmented reasoning process. Thought template is marked in orange and instantiated thought is marked in blue. 3.1 Problem Distiller Most of complex tasks contain implicit constraints, complex object relationships, and intricate variables and parameters within their contexts. Consequently, during the reasoning stage, LLMs need to overcome three main challenges: extracting vital information, recognizing potential constraints, and performing accurate reasoning. These challenges would impose a significant burden on a single LLM. Therefore, we separate the extraction and comprehension stages of task information from the final reasoning stage, through prepending a problem distiller to the reasoning process. More concretely, we design a meta prompt $\mathcal{\phi}$ to first distill and formalize the task information. The distilled task information could be denoted as: $$ x_{d}=LLM(\mathcal{\phi}(x)), \tag{1} $$ where $x$ is the task statement. Due to the page limit, we put the detailed meta prompt for problem-distiller in Section A.2. Problem Condensation and Translation We use the problem distiller to extract key elements from input tasks, focusing on: (1). Essential parameters and variables for problem-solving; (2). The objectives of the input tasks and their corresponding constraints. We then re-organize this distilled information into a clear, comprehensible format for the subsequent reasoning stage. We then translate the specific problems into high-level concepts and structures. This translation procedure decomposes complex real-world problems, like intricate mathematical application scenarios, into simpler, multi-step calculations, making it easier for later retrieval of high-level thought. 3.2 Thought-Augmented Reasoning with Meta Buffer Motivation Human often summarize and induce higher-level guidelines when solving problems and then apply them to relevant problems. Motivated by this, we propose meta-buffer, a lightweight library that contains a series of high-level thoughts (thought-template) for addressing various types of problems. Unlike traditional methods [11, 46, 12, 36, 9] that require specific instructions or exemplars, our high-level thought-templates can be adaptively instantiated when solving different problems, thereby enhancing LLMs with superior precision and flexibility. Thought Template As a kind of high-level guideline, our thought-template is stored in meta-buffer , and is obtained from various problem-solving processes by our buffer-manager. The details about acquiring thought-templates would be introduced in Section 3.3. Since our BoT aims to provide a general reasoning approach for various tasks, we correspondingly classify the thought-templates into six categories: Text Comprehension, Creative Language Generation, Common Sense Reasoning, Mathematical Reasoning, Code Programming and Application Scheduling. We provide some example thought-templates in Section A.1. Such classification of thought-templates can facilitate the template retrieval for finding most suitable solutions to different problems. Here we denote thought template, template description and its corresponding category as $(T_{i},D_{T_{i}},C_{k})$ , where $i$ denotes the index of meta-template, $k∈\mathbb{Z^{+}}$ and $1≤ k≤ 6$ , which means $C_{k}$ is in one of the six categories, and $D_{T_{i}}$ is the description of thought template. Template Retrieval For each task, our BoT retrieves a thought-template $T_{i}$ that is highly similar to the distilled problem $x_{d}$ by calculating the embedding similarity between the description $D_{T_{i}}$ and $x_{d}$ . The retrieval process can be formulated as: $$ j=\text{argmax}_{i}(\text{Sim}(f(x_{d}),\{f(D_{T_{i}})\}_{i=1}^{N})),\quad% \text{where}\quad\text{Sim}(f({x}_{d}),\{f(D_{T_{i}})\}_{i=0}^{n})>=\delta, \tag{2} $$ $N$ is the size of the meta-buffer, $f(·)$ is a normal text embedding model, and $T_{j}$ denotes the retrieved thought template. We set a threshold $\delta$ (0.5 $\sim$ 0.7 is recommended) to determine whether the current task is new. Therefore, if $\text{Sim}(f({x}_{d}),\{f(D_{T_{i}})\}_{i=0}^{n})<\delta$ , we identify the task $x$ as a new task. Instantiated Reasoning For each specific task, we discuss two situations for the instantiated reasoning, depending on whether the current task is new: The first situation is that we successfully retrieve a thought-template $T_{j}$ for the task. In this case, as presented in Figure 2, our thought-augmented reasoning will be adaptively instantiated to suitable reasoning structures with our designed instantiation prompt (in Section A.3). For example, in a Checkmate-in-One problem, we instantiate the template of updating chess board state to solve the problem step by step. Thus we conduct the instantiated reasoning for task $x$ using the distilled information $x_{d}$ and the retrieved template $T_{j}$ , and produce its solution $S_{x}$ as: $$ S_{x}=LLM_{\text{instantiation}}(x_{d},T_{j}), \tag{3} $$ where $LLM_{\text{instantiation}}$ denotes the instantiated reasoner with a LLM. In the second situation, the task is identified as a new task. To enable proper instantiated reasoning, we prepare three general coarse-grained thought-templates for utilization. Based on the distilled task information $x_{d}$ , our BoT would automatically assign a suitable thought-template to the reasoning process. The detailed pre-defined thought-templates are included in Section A.3). 3.3 Buffer Manager We propose buffer-manager to summarize the high-level guidelines and thoughts that are gained from each problem-solving process. It can generalize each specific solution to more problems, storing the critical distilled knowledge in the form of thought-templates within the meta buffer. In contrast to methods that temporarily generate exemplars or instructions for each problem, our buffer-manager can ensure permanent advancements in accuracy, efficiency, and robustness for LLM-based reasoning. Template Distillation To extract a general though-template, we propose a three-step approach: (1) Core task summarization: identifying and describing basic types and core challenges of problems; (2) Solution steps description: summarize the general steps for solving a problem; (3) General answering template: based on the above analysis, propose a solution template or approach that can be widely applied to similar problems. Additionally, to boost the generalization ability and stability of template distillation, we carefully design two types of in-context examples of how to generate thought-template— in-task and cross-task examples. Cross-task means we choose the template distilled from one task to tackle the problem of other tasks, such as addressing a mathematical problem with a code-related thought-template. The new template distilled from input task $x$ can be denoted as: $$ T_{new}=LLM_{\text{distill}}(x_{d},S_{x}), \tag{4} $$ where $LLM_{\text{distill}}$ is the LLM-based template distiller initialized with the following prompt: Prompt for Template Distillation: User: [Problem Description] + [Solution Steps or Code] To extract and summarize the high-level paradigms and general approaches for solving such problems, please follow these steps in your response: 1. Core task summarization: Identify and describe the basic type and core challenges of the problem, such as classifying it as a mathematical problem (e.g., solving a quadratic equation), a data structure problem (e.g., array sorting), an algorithm problem (e.g., search algorithms), etc. And analyze the most efficient way to solve the problem. 2. Solution Steps Description: Outline the general solution steps, including how to define the problem, determine variables, list key equations or constraints, choose appropriate solving strategies and methods, and how to verify the correctness of the results. 3. General Answer Template: Based on the above analysis, propose a template or approach that can be widely applied to this type of problem, including possible variables, functions, class definitions, etc. If it is a programming problem, provide a set of base classes and interfaces that can be used to construct solutions to specific problems. Please ensure that your response is highly concise and structured, so that specific solutions can be transformed into generalizable methods. [Optional] Here are some exemplars of the thought-template: (Choose cross-task or in-task exemplars based on the analysis of the Core task summarization.) Dynamic Update of Meta-Buffer After template distillation, we need to consider whether the distilled template should be updated into the meta-buffer. If we initialize an empty meta-buffer or encounter a problem without a proper thought-template, the distilled thought-templates will be directly stored in the meta-buffer. If we solve problem with a retrieved thought-template, new insights may arise during the instantiation of a certain thought-template. Therefore, to avoid the redundancy of the meta-buffer while maintaining newly-generated informative thoughts, we will calculate the similarity between the embedding vectors of $D_{T_{new}}$ and $\{D_{T_{i}}\}_{i=0}^{n}$ and update the meta-buffer with the following rule: $$ \text{Max}(\text{Sim}(f(D_{T_{new}}),\{f(D_{T_{i}})\}_{i=0}^{n}))<\delta. \tag{5} $$ Otherwise, it means the meta-buffer has already possessed the necessary knowledge to solve this task and does not need to perform the update. Our dynamic update strategy effectively reduces the computational burden of template retrieval while ensuring the lightweight property of our meta-buffer. We further conduct ablation study to analyze it in Section 6. 4 Experiments Datasets and Tasks To evaluate the efficacy of our proposed Buffer of Thoughts and compare with previous methods, we consider a diverse set of tasks and datasets that require varying degrees of mathematical and algorithmic reasoning, domain-specific knowledge, and literary creativity: (a). The Game of 24 from ToT [14], where the objective is to form an arithmetic expression that equals 24 using each of four given numbers exactly once; (b). Three BIG-Bench Hard (BBH) [35] tasks: Geometric Shapes, Multi-Step Arithmetic Two, and Word Sorting; (c). Three reasoning tasks directly obtained from the BIG-Bench suite [50]: Checkmate-in-One, Penguins —where the task is to answer questions about penguins’ attributes based on a given table and additional natural language information, and DateUnderstanding —a task that involves inferring dates from natural language descriptions, performing arithmetic operations on dates, and utilizing global knowledge such as the number of days in February; (d). Python Programming Puzzles (P3) [51, 52], a collection of challenging programming puzzles written in Python with varying difficulty levels; (e). Multilingual Grade School Math (MGSM) [33], a multilingual version of the GSM8K dataset [53] featuring translations of a subset of examples into ten typologically diverse languages, including Bengali, Japanese, and Swahili; (f). Shakespearean Sonnet Writing from meta-prompting [15], a novel task where the goal is to write a sonnet following the strict rhyme scheme "ABAB CDCD EFEF GG" and incorporating three provided words verbatim. Implementation and Baselines For the fair comparisons with previous methods, we use GPT-4 as the base model of our BoT, including the main experiment and the ablation study (in Section 6). We also use Llama3-8B and Llama3-70B in our analysis part on NVIDIA A100-PCIE-40GB GPU. We compare our Buffer of Thoughts with the following prompting methods: 1. Standard Prompting: This is our most basic baseline, where an LLM is asked to generate a response directly from the input query, without any specific guiding input-output examples or additional instructions beyond the task description included in the query. 2. Single-query Method: This includes Zero-shot CoT [8] and PAL [10], which use the LLM to analyze natural language problems and generate intermediate reasoning steps. We also include Expert Prompting [9], which creates an expert identity tailored to the specific context of the input query, and then integrates this expert profile into the input to generate a well-informed response. 3. Multi-query Method: This includes ToT [14] and GoT [17], which enable LLMs to make deliberate decisions by considering multiple reasoning paths and self-evaluating choices to determine the next course of action. These methods also allow for looking ahead or backtracking when necessary to make global decisions. Additionally, we include Meta Prompting [15], which employs an effective scaffolding technique designed to enhance the functionality of LLMs. Table 1: Comparing BoT with previous methods across various tasks. We denote the best score in blue, and the second-best score in green. Our BoT significantly outperforms other methods on all tasks, especially on general reasoning problems. | Game of 24 MGSM (avg) Multi-Step Arithmetic | 3.0 84.4 84.0 | 11.0 85.5 83.2 | 3.0 85.0 83.2 | 64.0 72.0 87.4 | 74.0 86.4 88.2 | 73.2 87.0 89.2 | 67.0 84.8 90.0 | 82.4 89.2 99.8 | | --- | --- | --- | --- | --- | --- | --- | --- | --- | | WordSorting | 80.4 | 83.6 | 85.2 | 93.2 | 96.4 | 98.4 | 99.6 | 100.0 | | Python Puzzles | 31.1 | 36.3 | 33.8 | 47.3 | 43.5 | 41.9 | 45.8 | 52.4 | | Geometric Shapes | 52.6 | 69.2 | 55.2 | 51.2 | 56.8 | 54.2 | 78.2 | 93.6 | | Checkmate-in-One | 36.4 | 32.8 | 39. 6 | 10.8 | 49.2 | 51.4 | 57.2 | 86.4 | | Date Understanding | 68.4 | 69.6 | 68.4 | 76.2 | 78.6 | 77.4 | 79.2 | 88.2 | | Penguins | 71.1 | 73.6 | 75.8 | 93.3 | 84.2 | 85.4 | 88.6 | 94.7 | | Sonnet Writing | 62.0 | 71.2 | 74.0 | 36.2 | 68.4 | 62.8 | 79.6 | 80.0 | 4.1 BoT Achieves Better Accuracy, Efficiency and Robustness Reasoning Accuracy As shown in LABEL:tab-accuracy, our BoT consistently outperforms all previous prompting methods across multiple kinds of challenging benchmarks, particularly demonstrated in complicated reasoning tasks such as Game of 24 and Checkmate-in-One. Taking GPT-4 as a baseline, our method achieves an astonishing 79.4% accuracy improvement in Game of 24, and compared to ToT, which has a good performance on this task, we also achieve an 8.4% accuracy improvement. What’s more, compared to recent Meta-prompting method [15], we see significant accuracy improvements: 23% on Game of 24, 20% on Geometric Shapes and 51% on Checkmate-in-One. Existing methods need complex, iterative, and heuristic search strategies to address these problems on a case-by-case basis. Conversely, our BoT leverages the historical insights and informative guidelines from thought-templates, and further adaptively instantiate a more optimal reasoning structure for addressing these complex problems. <details> <summary>x3.png Details</summary>  ### Visual Description ## Bar Chart: Comparison of the Inference Time ### Overview The chart compares the logarithmic inference time (in seconds) of five methods—Expert, PAL, ToT, Meta-prompting, and Ours—across three games: Game of 24, MGSM, and Checkmate-in-One. The y-axis uses a logarithmic scale (0–10), emphasizing relative differences in performance. ### Components/Axes - **X-axis**: Categorical labels for the three games: - Game of 24 - MGSM - Checkmate-in-One - **Y-axis**: Logarithmic time (s), ranging from 0 to 10. - **Legend**: Located in the top-right corner, mapping colors to methods: - Blue: Expert - Orange: PAL - Gray: ToT - Yellow: Meta-prompting - Light Blue: Ours ### Detailed Analysis #### Game of 24 - **Expert**: 4.64 (blue) - **PAL**: 5.5 (orange) - **ToT**: 8.73 (gray) - **Meta-prompting**: 8.47 (yellow) - **Ours**: 5.17 (light blue) #### MGSM - **Expert**: 4.16 (blue) - **PAL**: 4.81 (orange) - **ToT**: 8.34 (gray) - **Meta-prompting**: 8.04 (yellow) - **Ours**: 5.0 (light blue) #### Checkmate-in-One - **Expert**: 4.81 (blue) - **PAL**: 5.21 (orange) - **ToT**: 9.03 (gray) - **Meta-prompting**: 8.43 (yellow) - **Ours**: 6.39 (light blue) ### Key Observations 1. **ToT** consistently has the highest inference times across all games, with values ranging from 8.34 to 9.03. 2. **Meta-prompting** follows closely behind ToT, with slightly lower but still high times (8.04–8.47). 3. **Ours** (light blue) demonstrates the lowest inference times in all games, outperforming other methods by a significant margin. 4. **Expert** and **PAL** show moderate performance, with Expert generally faster than PAL. 5. The logarithmic scale highlights exponential differences, e.g., ToT’s 9.03 in Checkmate-in-One is ~1.5x slower than Meta-prompting’s 8.43. ### Interpretation The data suggests that the method labeled "Ours" is the most efficient, achieving faster inference times than all other approaches. ToT and Meta-prompting, while slower, may prioritize accuracy or complexity over speed. The logarithmic scale emphasizes that even small numerical differences (e.g., 5.17 vs. 5.5) represent meaningful performance gaps. This chart likely evaluates trade-offs between speed and other factors (e.g., accuracy) in AI or computational models. </details> Figure 3: Comparison of logarithmic inference time between our Buffer of Thoughts and GPT4 [3], GPT4+CoT [8], Expert-prompting [9], PAL [10], ToT [14] across different benchmarks. <details> <summary>x4.png Details</summary>  ### Visual Description ## Bar Chart: Success Rate ### Overview The chart compares the average accuracy (%) of five AI models (GPT4, Expert, PAL, ToT, Ours) across four games: Game of 24, MGSM, Checkmate-in-One, and an Average. The "Ours" model consistently achieves the highest accuracy, while performance varies significantly for other models depending on the game. ### Components/Axes - **X-axis**: Games (Game of 24, MGSM, Checkmate-in-One, Average) - **Y-axis**: Average accuracy (%) (0–100 scale) - **Legend**: - Blue: GPT4 - Orange: Expert - Gray: PAL - Yellow: ToT - Light Blue: Ours - **Bars**: Grouped by game, with values labeled on top of each bar. ### Detailed Analysis #### Game of 24 - **GPT4**: 27% (blue) - **Expert**: 36% (orange) - **PAL**: 61% (gray) - **ToT**: 71% (yellow) - **Ours**: 98% (light blue) #### MGSM - **GPT4**: 85% (blue) - **Expert**: 87% (orange) - **PAL**: 76% (gray) - **ToT**: 84% (yellow) - **Ours**: 96.8% (light blue) #### Checkmate-in-One - **GPT4**: 48.2% (blue) - **Expert**: 53.4% (orange) - **PAL**: 36.4% (gray) - **ToT**: 78.4% (yellow) - **Ours**: 93.4% (light blue) #### Average - **GPT4**: 67.13% (blue) - **Expert**: 71.82% (orange) - **PAL**: 70.12% (gray) - **ToT**: 84.57% (yellow) - **Ours**: 95.15% (light blue) ### Key Observations 1. **Ours Dominates**: The "Ours" model achieves the highest accuracy in all games and the average, with a 98% success rate in Game of 24 and 95.15% overall. 2. **GPT4 and Expert**: These models show inconsistent performance. GPT4 peaks at 85% in MGSM but drops to 27% in Game of 24. Expert performs best in MGSM (87%) but struggles in Checkmate-in-One (53.4%). 3. **PAL**: Underperforms in Checkmate-in-One (36.4%) but improves in other games (61–76%). 4. **ToT**: Consistently strong, with 71–84.57% accuracy, but lags behind "Ours" in all cases. 5. **Checkmate-in-One Challenge**: All models except "Ours" show significant drops in this game, suggesting it is particularly difficult. ### Interpretation The data demonstrates that the "Ours" model outperforms existing benchmarks (GPT4, Expert, PAL, ToT) across all tested games, indicating superior adaptability or optimization. The stark drop in GPT4 and Expert performance in Checkmate-in-One highlights potential limitations in handling complex or niche tasks. The "Ours" model’s consistent dominance suggests it may incorporate novel strategies or architectural improvements. The Average row reinforces this trend, with "Ours" achieving a 95.15% accuracy compared to the next highest (ToT at 84.57%). This chart underscores the importance of model-specific tuning for game success rates. </details> Figure 4: Comparison of reasoning robustness between our Buffer of Thoughts and GPT4 [3], GPT4+CoT [8], Expert-prompting [9], PAL [10], ToT [14] across different benchmarks. Reasoning Efficiency In addition to significant improvements in accuracy, as a multi-query method, our BoT can achieve comparable reasoning time to single-query method across various tasks, while being considerably less than conventional multi-query method like ToT [14] as shown in Figure 3. For example, in Game of 24, both single-query and multi-query methods necessitate iterative and heuristic searches to identify feasible solutions. This process is particularly time-consuming and inefficient, especially for the multi-query method, which involves conducting multi-query search and backtrace phases. In contrast, our BoT directly retrieves a thought-template in code format, thus a program is instantiated to traverse combinations of numbers and symbols, thereby eliminating the need to build the reasoning structure from scratch. This allows for solving the problem with just one query after invoking the problem-distiller, significantly reducing the time required for complex reasoning. Notably, our BoT requires only 12% of the cost of multi-query methods (e.g., tree of thoughts and meta-prompting) on average. Reasoning Robustness To better evaluate our BoT, we devise a new evaluation metric: success rate, which is used to assess the reasoning robustness. We randomly sample 1000 examples from various benchmarks as a test subset and evaluate different methods on this subset. As shown in Figure 4, we repeat this evaluation process 10 times and take the average accuracy as the success rate of different methods on each benchmark. Compared with other methods, our BoT consistently maintains a higher success rate across various tasks, surpassing the second-best by 10% in average success rate. We attribute our outstanding robustness to the great generalization ability of our distilled thought-templates during reasoning across different tasks. By offering high-level thought from the suitable thought-templates, the stability of our method across different tasks is greatly enhanced. 5 Model Analysis Distribution Analysis of Thought-Templates As depicted in the left figure of Figure 5, we choose six different benchmarks, each sampled with 100 distinct tasks. We update the meta-buffer from scratch, and after completing all sampled tasks, we display the number of derived thought-templates. We can observe that our BoT generates a greater number of thought-templates in the MGSM tasks that contain more diverse scenarios. In tasks with relatively simple requirements, such as Checkmate-in-One and Penguins, BoT produces more fixed thought-templates tailored for those specific issues. The distribution of templates indicates that our BoT can effectively discover appropriate thought templates for different benchmarks. <details> <summary>x5.png Details</summary>  ### Visual Description ## Pie Chart: Template distribution across different tasks ### Overview The image displays a pie chart illustrating the distribution of templates across six different tasks. The chart uses distinct colors for each task, with numerical values and labels indicating the proportion of each task. The largest segment dominates the chart, while the smallest segment is significantly smaller. ### Components/Axes - **Title**: "Template distribution across different tasks" (centered at the top) - **Legend**: Located on the right side, mapping colors to tasks: - Blue: Sonnet writing - Green: Table of Penguins - Yellow: Date understanding - Red: MGSM - Light Blue: Python Puzzles - Dark Green: Checkmate-in-One - **Segments**: Six colored sections with labels and numerical values: - **Sonnet writing** (blue): 37 - **Table of Penguins** (green): 8 - **Date understanding** (yellow): 14 - **MGSM** (red): 78 - **Python Puzzles** (light blue): 37 - **Checkmate-in-One** (dark green): 4 ### Detailed Analysis - **MGSM** (red) occupies the largest portion (78 units), followed by **Sonnet writing** (37) and **Python Puzzles** (37), which are tied for second place. - **Date understanding** (yellow) accounts for 14 units, while **Table of Penguins** (green) has 8 units. - The smallest segment, **Checkmate-in-One** (dark green), represents only 4 units. ### Key Observations 1. **Dominance of MGSM**: The red segment (MGSM) is overwhelmingly the largest, accounting for over half of the total distribution. 2. **Tied Segments**: Sonnet writing and Python Puzzles share identical values (37 units each), suggesting equal usage. 3. **Minor Segments**: Table of Penguins (8) and Checkmate-in-One (4) are significantly smaller, with the latter being the least used. 4. **Sum Discrepancy**: The total sum of values (78 + 37 + 37 + 14 + 8 + 4 = 178) exceeds 100, indicating these values likely represent absolute counts rather than percentages. ### Interpretation The chart highlights a clear hierarchy in template usage, with MGSM being the most frequently used template by a substantial margin. The near-identical counts for Sonnet writing and Python Puzzles suggest these tasks may have similar template requirements or user preferences. The small values for Table of Penguins and Checkmate-in-One imply these tasks are either less common or require fewer templates. The total sum exceeding 100 indicates the data represents raw counts rather than normalized percentages, which could be clarified in the chart's metadata or accompanying documentation. This distribution might reflect workflow priorities, task complexity, or user behavior patterns in a technical or creative workflow context. </details> <details> <summary>x6.png Details</summary>  ### Visual Description ## Pie Chart: Average time distribution for each part of our BoT ### Overview The image displays a pie chart titled "Average time distribution for each part of our BoT," illustrating the proportional time allocation across four components of a system: problem-distiller, reasoner, meta-buffer, and buffer-manager. The chart uses distinct colors for each component, with numerical percentages labeled directly on the segments. ### Components/Axes - **Legend**: Located on the right side of the chart, mapping colors to components: - Blue: problem-distiller - Green: reasoner - Yellow: meta-buffer - Red: buffer-manager - **Segments**: Four labeled sections with percentages: - problem-distiller (blue): 15.6% - reasoner (green): 52.7% - meta-buffer (yellow): 8.9% - buffer-manager (red): 21.3% ### Detailed Analysis - **problem-distiller** (blue): Occupies 15.6% of the chart, positioned between the reasoner and buffer-manager segments. - **reasoner** (green): Dominates the chart at 52.7%, forming the largest segment. - **meta-buffer** (yellow): Smallest segment at 8.9%, located between the buffer-manager and problem-distiller. - **buffer-manager** (red): Second-largest segment at 21.3%, adjacent to the problem-distiller. ### Key Observations 1. The **reasoner** consumes over half of the total time (52.7%), indicating it is the most time-intensive component. 2. The **buffer-manager** (21.3%) and **problem-distiller** (15.6%) together account for ~37% of the time, significantly less than the reasoner. 3. The **meta-buffer** (8.9%) is the least time-consuming component, occupying less than 10% of the distribution. ### Interpretation The data suggests a system architecture where the **reasoner** is the primary computational bottleneck, requiring optimization for efficiency. The **buffer-manager** and **problem-distiller** play secondary roles, while the **meta-buffer** has minimal time allocation. This distribution could reflect a design prioritizing reasoning capabilities over other components, potentially at the cost of scalability or latency in other subsystems. The stark disparity between the reasoner and other components highlights a possible trade-off in system design, where increased reasoning capacity may come with reduced flexibility or responsiveness in auxiliary processes. </details> Figure 5: Distribution Analysis of Thought-Templates and Time. Left: Distribution Analysis of Thought-Templates. Right: Distribution Analysis of Thought-Templates. Distribution Analysis of Time Cost As illustrated in Figure 5, we measured the average time cost for each component of BoT’s reasoning framework across different tasks. The time required for distilling task information and template retrieval is relatively short, whereas instantiated reasoning takes longer. Overall, considering the complexity of different components, our BoT achieves a relatively balanced distribution of time cost, demonstrating the efficiency of our BoT framework. Better Trade-off between Model Size and Performance As depicted in Figure 6, on Game of 24, word list sorting and Checkmate-in-One, Llama3-8B and Llama-70B models [6] may result in poor outcomes. However, equipped with our BoT, both models demonstrate a substantial accuracy improvement. Notably, BoT+Llama3-8B has the potential to surpass single Llama3-70B model. Our BoT enables smaller models to exhibit the capabilities that approximate or even surpass larger models, significantly bridging the gap between their reasoning abilities. Furthermore, it greatly diminishes the inference cost required by large language models when tackling complex problems. <details> <summary>x7.png Details</summary>  ### Visual Description ## Bar Chart: Trade-off between model size and performance ### Overview The chart compares the accuracy performance of four AI models across three tasks: Checkmate-in-One, Word list sorting, and Game of 24. It visualizes the relationship between model size (70B vs 8B parameters) and performance, with special attention to models combining BoT (presumably a base model) with Llama-3 variants. ### Components/Axes - **X-axis**: Accuracy (%) from 0 to 100 in 10% increments - **Y-axis**: Three tasks (Checkmate-in-One, Word list sorting, Game of 24) - **Legend**: Located at top-right, color-coded models: - Orange: BoT+Llama-3-70B - Gray: BoT+Llama-3-8B - Red: Llama3-70B - Blue: Llama3-8B ### Detailed Analysis 1. **Checkmate-in-One** - BoT+Llama-3-70B (orange): 75.6% - BoT+Llama-3-8B (gray): 56.7% - Llama3-70B (red): 15% - Llama3-8B (blue): 0.8% 2. **Word list sorting** - BoT+Llama-3-70B (orange): 92.3% - BoT+Llama-3-8B (gray): 73.4% - Llama3-70B (red): 79% - Llama3-8B (blue): 48.4% 3. **Game of 24** - BoT+Llama-3-70B (orange): 78.4% - BoT+Llama-3-8B (gray): 73.4% - Llama3-70B (red): 2.4% - Llama3-8B (blue): 1.2% ### Key Observations 1. **Model Size Impact**: - 70B models consistently outperform 8B versions across all tasks - Largest gap in Checkmate-in-One (75.6% vs 56.7% for BoT+Llama variants) 2. **BoT+Llama Advantage**: - BoT+Llama-3-70B achieves highest accuracy in all tasks - Particularly dominant in Checkmate-in-One (75.6% vs 15% for standalone Llama3-70B) 3. **Performance Drops**: - Llama3-8B shows dramatic performance decline (e.g., 48.4% vs 73.4% in Word list sorting) - Smallest models (Llama3-8B) perform near-randomly in Checkmate-in-One (0.8%) 4. **Task Specificity**: - Word list sorting shows highest overall performance (92.3% peak) - Game of 24 demonstrates closest performance between model sizes (78.4% vs 73.4%) ### Interpretation The data reveals a clear trade-off between model size and performance, with larger models (70B) demonstrating significantly better accuracy across all tasks. However, the BoT+Llama combination appears to create a synergistic effect, particularly evident in the Checkmate-in-One task where the 70B version outperforms the standalone Llama3-70B by 60.6 percentage points. This suggests that model architecture optimization (via BoT integration) may be more impactful than raw parameter count alone. Notably, the performance gap between 70B and 8B models narrows in Word list sorting (18.9% difference) compared to Checkmate-in-One (18.9% difference) and Game of 24 (5% difference), indicating task-dependent scaling efficiency. The near-zero performance of Llama3-8B in Checkmate-in-One highlights potential architectural limitations in smaller models for complex reasoning tasks. </details> Figure 6: We evaluate the trade-off between model size and performance with Llama3-8B and Llama3-70B models on three challenging benchmarks. 6 Ablation Study Impact of Problem-Distiller As illustrated in Figure 7, when the problem-distiller is disabled, both Llama3-70B and GPT-4 experience a certain degree of accuracy decline. More complex problems, such as Game of 24 and Checkmate-in-One, show a more significant accuracy reduction, whereas relatively simpler problems like word list sorting and MGSM exhibit smaller decreases. This is because LLMs can more easily extract key information in simpler tasks, making the impact of the problem-distiller less noticeable. In contrast, extracting key information and potential constraints in complex problems is more challenging, making the role of our problem-distiller more prominent, thereby explaining the differences depicted in the figure. <details> <summary>x8.png Details</summary>  ### Visual Description ## Bar Chart: Ablation study of problem-distiller ### Overview The chart compares the accuracy of four model configurations across four tasks: Game of 24, Word list sorting, Checkmate-in-One, and MGSM. Each task is represented by a group of four bars, with colors corresponding to specific model combinations. The y-axis measures accuracy in percentage, ranging from 0 to 100. ### Components/Axes - **X-axis (Tasks)**: - Game of 24 - Word list sorting - Checkmate-in-One - MGSM - **Y-axis (Accuracy)**: - Scale: 0% to 100% in 10% increments - Label: "Accuracy (%)" - **Legend (Top-left)**: - Blue: BoT+Llama-3-70B (w/o problem-distiller) - Orange: BoT+Llama-3-70B - Gray: BoT+GPT-4 (w/o problem-distiller) - Yellow: BoT+GPT-4 ### Detailed Analysis #### Task: Game of 24 - Blue (BoT+Llama-3-70B w/o): 71.2% - Orange (BoT+Llama-3-70B): 78.4% - Gray (BoT+GPT-4 w/o): 76.5% - Yellow (BoT+GPT-4): 82.4% #### Task: Word list sorting - Blue (BoT+Llama-3-70B w/o): 89.5% - Orange (BoT+Llama-3-70B): 92.3% - Gray (BoT+GPT-4 w/o): 97.3% - Yellow (BoT+GPT-4): 99.6% #### Task: Checkmate-in-One - Blue (BoT+Llama-3-70B w/o): 64.3% - Orange (BoT+Llama-3-70B): 75.6% - Gray (BoT+GPT-4 w/o): 78.9% - Yellow (BoT+GPT-4): 86.4% #### Task: MGSM - Blue (BoT+Llama-3-70B w/o): 85.6% - Orange (BoT+Llama-3-70B): 86.8% - Gray (BoT+GPT-4 w/o): 87.4% - Yellow (BoT+GPT-4): 89.2% ### Key Observations 1. **Problem-distiller impact**: - Orange/yellow bars (with problem-distiller) consistently outperform blue/gray bars (without) across all tasks. - Largest improvement: Word list sorting (BoT+GPT-4: +2.3% with problem-distiller). 2. **Model performance**: - BoT+GPT-4 models (yellow/gray) generally outperform BoT+Llama-3-70B (orange/blue). - Checkmate-in-One shows the lowest accuracy overall (64.3% baseline). 3. **Task difficulty**: - Word list sorting achieves near-perfect accuracy (99.6% peak). - Checkmate-in-One has the largest performance gap between models (+22.1% between lowest and highest). ### Interpretation The data demonstrates that the problem-distiller significantly enhances model performance across all tasks, with the most dramatic improvement observed in Word list sorting. BoT+GPT-4 models consistently outperform BoT+Llama-3-70B, suggesting GPT-4's superior base capabilities. Checkmate-in-One's lower accuracy highlights its complexity compared to other tasks. The ablation study confirms that problem-distillation is critical for optimizing model performance, particularly for GPT-4-based systems. </details> Figure 7: We conduct ablation study on problem-distiller across four benchmarks, employing Llama3-70B and GPT-4 as the base models. Impact of Meta-Buffer As illustrated in Figure 8, when the meta-buffer is disabled, both Llama3-70B and GPT-4 models exhibit a noticeable decline in performance, particularly in benchmarks requiring complex reasoning, such as Game of 24 and Checkmate-in-One. This further underscores the superiority of our meta-buffer in addressing complex problems. <details> <summary>x9.png Details</summary>  ### Visual Description ## Bar Chart: Ablation study of meta-buffer ### Overview The chart compares the accuracy of four model configurations across four tasks: Game of 24, Word list sorting, Checkmate-in-One, and MGSM. Each task has four grouped bars representing different model variants with/without a meta-buffer. ### Components/Axes - **X-axis**: Tasks (Game of 24, Word list sorting, Checkmate-in-One, MGSM) - **Y-axis**: Accuracy (%) from 0 to 100 - **Legend**: - Blue: BoT + Llama-3-70B (w/o meta-buffer) - Orange: BoT+Llama-3-70B - Gray: BoT+GPT-4 (w/o meta-buffer) - Yellow: BoT+GPT-4 ### Detailed Analysis 1. **Game of 24**: - Blue (BoT + Llama-3-70B w/o meta-buffer): 65.6% - Orange (BoT+Llama-3-70B): 78.4% - Gray (BoT+GPT-4 w/o meta-buffer): 75.2% - Yellow (BoT+GPT-4): 82.4% 2. **Word list sorting**: - Blue: 81.7% - Orange: 92.3% - Gray: 95.4% - Yellow: 99.6% 3. **Checkmate-in-One**: - Blue: 27.4% - Orange: 75.6% - Gray: 56.7% - Yellow: 86.4% 4. **MGSM**: - Blue: 79.6% - Orange: 86.8% - Gray: 85.4% - Yellow: 89.2% ### Key Observations - **BoT+GPT-4 (yellow)** consistently achieves the highest accuracy across all tasks, with a peak of 99.6% in Word list sorting. - **BoT + Llama-3-70B (blue)** shows the lowest performance, particularly in Checkmate-in-One (27.4%). - The meta-buffer improves accuracy for both Llama-3-70B and GPT-4 models, with the largest relative gain observed in Checkmate-in-One (BoT+GPT-4: +29.7% with meta-buffer). - Word list sorting demonstrates near-perfect performance for BoT+GPT-4 (99.6%), suggesting task-specific optimization. ### Interpretation The data demonstrates that: 1. The meta-buffer significantly enhances model performance, especially for complex tasks like Checkmate-in-One where BoT+GPT-4 with meta-buffer achieves 86.4% vs 56.7% without. 2. GPT-4-based models outperform Llama-3-70B variants across all tasks, with the gap widening in more challenging scenarios. 3. The absence of the meta-buffer disproportionately impacts Llama-3-70B's performance, suggesting architectural limitations in handling task complexity without external memory augmentation. 4. Word list sorting's near-perfect accuracy for BoT+GPT-4 indicates potential overfitting or specialized optimization for this particular task type. The ablation study highlights the critical role of meta-buffers in enabling large language models to handle complex reasoning tasks, with GPT-4 showing superior base capabilities but requiring similar architectural enhancements for optimal performance. </details> Figure 8: We conduct ablation study on meta-buffer across four benchmarks, employing Llama3-70B and GPT-4 as the base models. <details> <summary>x10.png Details</summary>  ### Visual Description ## Line Chart: Ablation study of buffer-manager -- Accuracy ### Overview The chart compares the accuracy of two configurations ("BoT+GPT4" and "BoT+GPT4 (w/o buffer-manager)") across four ablation rounds. The blue line represents the full system with a buffer-manager, while the orange line represents the system without it. Accuracy is measured on a percentage scale from 0 to 100. ### Components/Axes - **X-axis (Rounds)**: Labeled "Round 1" to "Round 4" at equal intervals. - **Y-axis (Accuracy %)**: Scaled from 0 to 100 in increments of 10. - **Legend**: Positioned at the bottom center, with: - Blue line: "BoT+GPT4" - Orange line: "BoT+GPT4 (w/o buffer-manager)" - **Data Points**: Red numerical labels above each line's markers (e.g., "56.8" for Round 1, blue line). ### Detailed Analysis 1. **Round 1**: - BoT+GPT4 (blue): 56.8% - BoT+GPT4 (w/o buffer-manager, orange): 52.8% 2. **Round 2**: - BoT+GPT4: 78.5% - BoT+GPT4 (w/o buffer-manager): 53.6% 3. **Round 3**: - BoT+GPT4: 87.4% - BoT+GPT4 (w/o buffer-manager): 57.4% 4. **Round 4**: - BoT+GPT4: 88.5% - BoT+GPT4 (w/o buffer-manager): 54.1% ### Key Observations - The blue line ("BoT+GPT4") shows a **steady upward trend**, increasing from 56.8% to 88.5% across all rounds. - The orange line ("BoT+GPT4 w/o buffer-manager") exhibits **minimal improvement**, peaking at 57.4% in Round 3 before declining to 54.1% in Round 4. - The gap between the two lines widens significantly in later rounds (e.g., 31.1% difference in Round 4). ### Interpretation The data demonstrates that the **buffer-manager component is critical for improving accuracy**, particularly in later ablation rounds. The full system ("BoT+GPT4") achieves near-doubled accuracy compared to the system without the buffer-manager. The orange line's slight decline in Round 4 suggests potential instability or overfitting when the buffer-manager is omitted. This ablation study highlights the buffer-manager's role in stabilizing and enhancing performance over iterative rounds. </details> Figure 9: We conduct ablation study on buffer-manager regarding reasoning accuracy across four tasks, employing Llama3-70B and GPT-4 as the base models. Impact of Buffer-Manager In this ablation study, we divide the entire process into four rounds. In each round, we randomly sample 50 questions from each benchmark and conduct reasoning. In the subsequent round, we continue to randomly sample another 50 questions from each benchmark. As depicted in Figure 9, with the increase of the number of rounds, the model with the buffer-manager continually expands the meta-buffer while also utilizing the thought-templates obtained from previously solved problems to help addressing subsequent similar problems. Therefore, we can observe that the accuracy of BoT steadily improves with each round. In contrast, the model without the buffer-manager fails to exhibit an upward trend. Additionally, we have also measured the reasoning time as depicted in Figure 10. when the number of rounds increases, the model with the buffer-manager will experience a continual improvement in reasoning efficiency. This is because, with the continual expansion of the meta-buffer, the likelihood of retrieving suitable thought-templates also increases. Consequently, models can avoid constructing reasoning structures from scratch, thereby enhancing the inference efficiency accordingly. <details> <summary>x11.png Details</summary>  ### Visual Description ## Line Chart: Ablation study of buffer-manager -- Time ### Overview The chart compares the average inference time (in seconds) per problem across four rounds for two configurations: "BoT+GPT4" (blue line) and "BoT+GPT4 (w/o buffer-manager)" (orange line). The y-axis ranges from 0 to 350 seconds, and the x-axis represents four sequential rounds. ### Components/Axes - **X-axis (Rounds)**: Labeled "Round 1," "Round 2," "Round 3," and "Round 4" at equal intervals. - **Y-axis (Average Inference Time)**: Scaled from 0 to 350 seconds in increments of 50. - **Legend**: Positioned at the bottom center, with: - Blue line: "BoT+GPT4" - Orange line: "BoT+GPT4 (w/o buffer-manager)" - **Data Points**: Embedded numerical values on the lines for each round. ### Detailed Analysis - **BoT+GPT4 (Blue Line)**: - **Round 1**: 297 seconds - **Round 2**: 205 seconds - **Round 3**: 128 seconds - **Round 4**: 78.5 seconds - **Trend**: Steady linear decrease across all rounds. - **BoT+GPT4 (w/o buffer-manager) (Orange Line)**: - **Round 1**: 308 seconds - **Round 2**: 317 seconds - **Round 3**: 304 seconds - **Round 4**: 306 seconds - **Trend**: Slight fluctuation but remains stable around 300–317 seconds. ### Key Observations 1. The "BoT+GPT4" configuration shows a consistent reduction in inference time, dropping from 297s to 78.5s over four rounds. 2. The "BoT+GPT4 (w/o buffer-manager)" configuration maintains near-constant inference times (~300–317s), with minor fluctuations. 3. The buffer-manager appears to significantly improve performance over time, as evidenced by the steep decline in the blue line. ### Interpretation The data suggests that the buffer-manager plays a critical role in reducing inference time for the BoT+GPT4 system. Without the buffer-manager, performance remains stagnant, while its inclusion enables progressive optimization. The linear decline in the blue line indicates that the buffer-manager’s benefits compound across rounds, possibly due to adaptive resource allocation or caching mechanisms. This ablation study highlights the buffer-manager as a key component for efficiency in large-scale inference tasks. </details> Figure 10: We conduct ablation study on buffer-manager regarding reasoning efficiency across four tasks, employing Llama3-70B and GPT-4 as the base models. 7 Discussion Limitations and Future Directions Despite our method’s significant improvement in accuracy while maintaining reasoning efficiency and robustness, our method’s enhancements are limited when addressing problems requiring human-like creativity, as this issue often does not rely on a specific thought-template. Besides, if our BoT initializes the meta-buffer with a weaker model, the quality of the derived thought-templates may be suboptimal due to the weaker model’s limited reasoning ability and instruction-following capability. Overall, our BoT brings out a set of future directions: 1. integrating external resources with BoT to build a open-domain system like agent models [54, 55]. 2. making the distillation of thought-templates optimizable, which may significantly enhance their template qualities for more complex tasks. Conclusion In this work, we introduce Buffer of Thoughts, a novel beffered reasoning framework that employs LLMs to utilize pre-accumulated experiences and methodologies from prior tasks as thought-templates stored within a meta-buffer. We further design buffer-manager to continuously refine the problem-solving processes and dynamically distill thought-templates, thereby progressively raising the LLM’s reasoning capacity. Our BoT demonstrates SOTA performance on 10 challenging tasks, and offers promising prospects for future research and application. References - [1] T. Brown, B. Mann, N. Ryder, M. Subbiah, J. D. Kaplan, P. Dhariwal, A. Neelakantan, P. Shyam, G. Sastry, A. Askell, et al., “Language models are few-shot learners,” Advances in neural information processing systems, vol. 33, pp. 1877–1901, 2020. - [2] R. Anil, A. M. Dai, O. Firat, M. Johnson, D. Lepikhin, A. Passos, S. Shakeri, E. Taropa, P. Bailey, Z. Chen, et al., “Palm 2 technical report,” arXiv preprint arXiv:2305.10403, 2023. - [3] J. Achiam, S. Adler, S. Agarwal, L. Ahmad, I. Akkaya, F. L. Aleman, D. Almeida, J. Altenschmidt, S. Altman, S. Anadkat, et al., “Gpt-4 technical report,” arXiv preprint arXiv:2303.08774, 2023. - [4] Z. Du, Y. Qian, X. Liu, M. Ding, J. Qiu, Z. Yang, and J. Tang, “Glm: General language model pretraining with autoregressive blank infilling,” in Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pp. 320–335, 2022. - [5] A. Q. Jiang, A. Sablayrolles, A. Roux, A. Mensch, B. Savary, C. Bamford, D. S. Chaplot, D. d. l. Casas, E. B. Hanna, F. Bressand, et al., “Mixtral of experts,” arXiv preprint arXiv:2401.04088, 2024. - [6] H. Touvron, T. Lavril, G. Izacard, X. Martinet, M.-A. Lachaux, T. Lacroix, B. Rozière, N. Goyal, E. Hambro, F. Azhar, et al., “Llama: Open and efficient foundation language models,” arXiv preprint arXiv:2302.13971, 2023. - [7] H. Touvron, L. Martin, K. Stone, P. Albert, A. Almahairi, Y. Babaei, N. Bashlykov, S. Batra, P. Bhargava, S. Bhosale, et al., “Llama 2: Open foundation and fine-tuned chat models,” arXiv preprint arXiv:2307.09288, 2023. - [8] J. Wei, X. Wang, D. Schuurmans, M. Bosma, F. Xia, E. Chi, Q. V. Le, D. Zhou, et al., “Chain-of-thought prompting elicits reasoning in large language models,” Advances in neural information processing systems, vol. 35, pp. 24824–24837, 2022. - [9] B. Xu, A. Yang, J. Lin, Q. Wang, C. Zhou, Y. Zhang, and Z. Mao, “Expertprompting: Instructing large language models to be distinguished experts,” arXiv preprint arXiv:2305.14688, 2023. - [10] L. Gao, A. Madaan, S. Zhou, U. Alon, P. Liu, Y. Yang, J. Callan, and G. Neubig, “Pal: Program-aided language models,” in International Conference on Machine Learning, pp. 10764–10799, PMLR, 2023. - [11] X. Wang, J. Wei, D. Schuurmans, Q. V. Le, E. H. Chi, S. Narang, A. Chowdhery, and D. Zhou, “Self-consistency improves chain of thought reasoning in language models,” in The Eleventh International Conference on Learning Representations, 2022. - [12] M. Yasunaga, X. Chen, Y. Li, P. Pasupat, J. Leskovec, P. Liang, E. H. Chi, and D. Zhou, “Large language models as analogical reasoners,” International Conference on Learning Representations, 2024. - [13] Z. Zhang, A. Zhang, M. Li, and A. Smola, “Automatic chain of thought prompting in large language models,” in The Eleventh International Conference on Learning Representations, 2022. - [14] S. Yao, D. Yu, J. Zhao, I. Shafran, T. Griffiths, Y. Cao, and K. Narasimhan, “Tree of thoughts: Deliberate problem solving with large language models,” Advances in Neural Information Processing Systems, vol. 36, 2024. - [15] M. Suzgun and A. T. Kalai, “Meta-prompting: Enhancing language models with task-agnostic scaffolding,” arXiv preprint arXiv:2401.12954, 2024. - [16] D. Zhou, N. Schärli, L. Hou, J. Wei, N. Scales, X. Wang, D. Schuurmans, C. Cui, O. Bousquet, Q. V. Le, et al., “Least-to-most prompting enables complex reasoning in large language models,” in The Eleventh International Conference on Learning Representations, 2022. - [17] M. Besta, N. Blach, A. Kubicek, R. Gerstenberger, M. Podstawski, L. Gianinazzi, J. Gajda, T. Lehmann, H. Niewiadomski, P. Nyczyk, et al., “Graph of thoughts: Solving elaborate problems with large language models,” in Proceedings of the AAAI Conference on Artificial Intelligence, vol. 38, pp. 17682–17690, 2024. - [18] A. Asai, S. Min, Z. Zhong, and D. Chen, “Retrieval-based language models and applications,” in Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 6: Tutorial Abstracts), pp. 41–46, 2023. - [19] G. Mialon, R. Dessi, M. Lomeli, C. Nalmpantis, R. Pasunuru, R. Raileanu, B. Roziere, T. Schick, J. Dwivedi-Yu, A. Celikyilmaz, et al., “Augmented language models: a survey,” Transactions on Machine Learning Research, 2023. - [20] W. Shi, S. Min, M. Yasunaga, M. Seo, R. James, M. Lewis, L. Zettlemoyer, and W.-t. Yih, “Replug: Retrieval-augmented black-box language models,” arXiv preprint arXiv:2301.12652, 2023. - [21] Y. Gao, Y. Xiong, X. Gao, K. Jia, J. Pan, Y. Bi, Y. Dai, J. Sun, and H. Wang, “Retrieval-augmented generation for large language models: A survey,” arXiv preprint arXiv:2312.10997, 2023. - [22] P. Zhao, H. Zhang, Q. Yu, Z. Wang, Y. Geng, F. Fu, L. Yang, W. Zhang, and B. Cui, “Retrieval-augmented generation for ai-generated content: A survey,” arXiv preprint arXiv:2402.19473, 2024. - [23] S. Borgeaud, A. Mensch, J. Hoffmann, T. Cai, E. Rutherford, K. Millican, G. B. Van Den Driessche, J.-B. Lespiau, B. Damoc, A. Clark, et al., “Improving language models by retrieving from trillions of tokens,” in International conference on machine learning, pp. 2206–2240, PMLR, 2022. - [24] M. Yasunaga, A. Aghajanyan, W. Shi, R. James, J. Leskovec, P. Liang, M. Lewis, L. Zettlemoyer, and W.-T. Yih, “Retrieval-augmented multimodal language modeling,” in International Conference on Machine Learning, pp. 39755–39769, PMLR, 2023. - [25] G. Izacard, P. Lewis, M. Lomeli, L. Hosseini, F. Petroni, T. Schick, J. Dwivedi-Yu, A. Joulin, S. Riedel, and E. Grave, “Atlas: Few-shot learning with retrieval augmented language models,” Journal of Machine Learning Research, vol. 24, no. 251, pp. 1–43, 2023. - [26] Z. Wang, W. Nie, Z. Qiao, C. Xiao, R. Baraniuk, and A. Anandkumar, “Retrieval-based controllable molecule generation,” in The Eleventh International Conference on Learning Representations, 2022. - [27] L. Yang, Z. Huang, X. Zhou, M. Xu, W. Zhang, Y. Wang, X. Zheng, W. Yang, R. O. Dror, S. Hong, et al., “Prompt-based 3d molecular diffusion models for structure-based drug design,” 2023. - [28] T. Kojima, S. S. Gu, M. Reid, Y. Matsuo, and Y. Iwasawa, “Large language models are zero-shot reasoners,” Advances in neural information processing systems, vol. 35, pp. 22199–22213, 2022. - [29] O. Press, M. Zhang, S. Min, L. Schmidt, N. A. Smith, and M. Lewis, “Measuring and narrowing the compositionality gap in language models,” in Findings of the Association for Computational Linguistics: EMNLP 2023, pp. 5687–5711, 2023. - [30] S. Arora, A. Narayan, M. F. Chen, L. Orr, N. Guha, K. Bhatia, I. Chami, and C. Re, “Ask me anything: A simple strategy for prompting language models,” in The Eleventh International Conference on Learning Representations, 2022. - [31] T. Khot, H. Trivedi, M. Finlayson, Y. Fu, K. Richardson, P. Clark, and A. Sabharwal, “Decomposed prompting: A modular approach for solving complex tasks,” in The Eleventh International Conference on Learning Representations, 2022. - [32] J. Wei, Y. Tay, R. Bommasani, C. Raffel, B. Zoph, S. Borgeaud, D. Yogatama, M. Bosma, D. Zhou, D. Metzler, et al., “Emergent abilities of large language models,” Transactions on Machine Learning Research, 2022. - [33] F. Shi, M. Suzgun, M. Freitag, X. Wang, S. Srivats, S. Vosoughi, H. W. Chung, Y. Tay, S. Ruder, D. Zhou, et al., “Language models are multilingual chain-of-thought reasoners,” in The Eleventh International Conference on Learning Representations, 2022. - [34] Y. Fu, H. Peng, A. Sabharwal, P. Clark, and T. Khot, “Complexity-based prompting for multi-step reasoning,” in The Eleventh International Conference on Learning Representations, 2022. - [35] M. Suzgun, N. Scales, N. Schärli, S. Gehrmann, Y. Tay, H. W. Chung, A. Chowdhery, Q. Le, E. Chi, D. Zhou, et al., “Challenging big-bench tasks and whether chain-of-thought can solve them,” in Findings of the Association for Computational Linguistics: ACL 2023, pp. 13003–13051, 2023. - [36] H. S. Zheng, S. Mishra, X. Chen, H.-T. Cheng, E. H. Chi, Q. V. Le, and D. Zhou, “Take a step back: Evoking reasoning via abstraction in large language models,” arXiv preprint arXiv:2310.06117, 2023. - [37] P. Zhou, J. Pujara, X. Ren, X. Chen, H.-T. Cheng, Q. V. Le, E. H. Chi, D. Zhou, S. Mishra, and H. S. Zheng, “Self-discover: Large language models self-compose reasoning structures,” arXiv preprint arXiv:2402.03620, 2024. - [38] W. Chen, X. Ma, X. Wang, and W. W. Cohen, “Program of thoughts prompting: Disentangling computation from reasoning for numerical reasoning tasks,” Transactions on Machine Learning Research, 2023. - [39] X. Ning, Z. Lin, Z. Zhou, Z. Wang, H. Yang, and Y. Wang, “Skeleton-of-thought: Large language models can do parallel decoding,” in The Twelfth International Conference on Learning Representations, 2023. - [40] Y. Zhang, “Meta prompting for agi systems,” arXiv preprint arXiv:2311.11482, 2023. - [41] J. Chen, R. Xu, Z. Fu, W. Shi, Z. Li, X. Zhang, C. Sun, L. Li, Y. Xiao, and H. Zhou, “E-kar: A benchmark for rationalizing natural language analogical reasoning,” in Findings of the Association for Computational Linguistics: ACL 2022, pp. 3941–3955, 2022. - [42] O. Sultan and D. Shahaf, “Life is a circus and we are the clowns: Automatically finding analogies between situations and processes,” in Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing, pp. 3547–3562, 2022. - [43] N. Zhang, L. Li, X. Chen, X. Liang, S. Deng, and H. Chen, “Multimodal analogical reasoning over knowledge graphs,” in The Eleventh International Conference on Learning Representations, 2022. - [44] B. Bhavya, J. Xiong, and C. Zhai, “Analogy generation by prompting large language models: A case study of instructgpt,” in Proceedings of the 15th International Conference on Natural Language Generation, pp. 298–312, 2022. - [45] B. Bhavya, J. Xiong, and C. Zhai, “Cam: A large language model-based creative analogy mining framework,” in Proceedings of the ACM Web Conference 2023, pp. 3903–3914, 2023. - [46] Z. Zhang, A. Zhang, M. Li, and A. Smola, “Automatic chain of thought prompting in large language models,” in The Eleventh International Conference on Learning Representations, 2022. - [47] T. Webb, K. J. Holyoak, and H. Lu, “Emergent analogical reasoning in large language models,” Nature Human Behaviour, vol. 7, no. 9, pp. 1526–1541, 2023. - [48] J. Yu, R. He, and Z. Ying, “Thought propagation: An analogical approach to complex reasoning with large language models,” in International Conference on Learning Representations, 2024. - [49] T. Feng, P. Han, G. Lin, G. Liu, and J. You, “Thought-retriever: Don’t just retrieve raw data, retrieve thoughts,” in ICLR 2024 Workshop: How Far Are We From AGI. - [50] B. bench authors, “Beyond the imitation game: Quantifying and extrapolating the capabilities of language models,” Transactions on Machine Learning Research, 2023. - [51] T. Schuster, A. Kalyan, A. Polozov, and A. T. Kalai, “Programming puzzles,” in Thirty-fifth Conference on Neural Information Processing Systems Datasets and Benchmarks Track, 2021. - [52] A. T. K. Patrick Haluptzok, Matthew Bowers, “Language models can teach themselves to program better,” in Eleventh International Conference on Learning Representations (ICLR), 2023. - [53] K. Cobbe, V. Kosaraju, M. Bavarian, M. Chen, H. Jun, L. Kaiser, M. Plappert, J. Tworek, J. Hilton, R. Nakano, C. Hesse, and J. Schulman, “Training verifiers to solve math word problems,” arXiv preprint arXiv:2110.14168, 2021. - [54] G. Chen, S. Dong, Y. Shu, G. Zhang, J. Sesay, B. F. Karlsson, J. Fu, and Y. Shi, “Autoagents: A framework for automatic agent generation,” arXiv preprint arXiv:2309.17288, 2023. - [55] Q. Wu, G. Bansal, J. Zhang, Y. Wu, S. Zhang, E. Zhu, B. Li, L. Jiang, X. Zhang, and C. Wang, “Autogen: Enabling next-gen llm applications via multi-agent conversation framework,” arXiv preprint arXiv:2308.08155, 2023. Appendix A Additional Method Details A.1 Detailed Thought-Templates Here we show six example thought-templates in six different categories: A.1.1 Text Comprehension Task Description: The task involves analyzing a table with various attributes of penguins, such as name, age, height, and weight, and answering questions about these attributes. The table may be updated with new entries, and additional context or comparisons may be provided in natural language. Solution Description: To accurately answer questions about the penguins’ attributes, one must be able to interpret the data presented in tabular form, understand any additional information provided in natural language, and apply logical reasoning to identify the correct attribute based on the question asked. Thought Template: Step 1: Parse the initial table, extracting the header information and each penguin’s attributes into a structured format (e.g., a list of dictionaries). Step 2: Read and integrate any additional natural language information that updates or adds to the table, ensuring the data remains consistent. Step 3: Identify the attribute in question (e.g., oldest penguin, heaviest penguin) and the corresponding column in the table. Step 4: Apply logical reasoning to compare the relevant attribute across all entries to find the correct answer (e.g., the highest age for the oldest penguin). Step 5: Select the answer from the provided options that matches the result of the logical comparison. A.1.2 Creative Language Generation Task Description: The task is to generate a sonnet that adheres to the traditional English sonnet rhyme scheme of "ABAB CDCD EFEF GG" and includes three specific words verbatim in the text. Solution Description: Writing a sonnet involves crafting 14 lines of poetry that follow a specific rhyme pattern. The lines are typically in iambic pentameter, though flexibility in rhythm can be allowed for creative reasons. The given rhyme scheme dictates the end sounds of each line, ensuring a structured poetic form. Incorporating the three provided words verbatim requires strategic placement within the lines to maintain the poem’s coherence and thematic unity. Thought Template: Step 1: Identify the three words that must be included in the sonnet. Step 2: Understand the rhyme scheme "ABAB CDCD EFEF GG" and prepare a list of rhyming words that could be used. Step 3: Develop a theme or story for the sonnet that can naturally incorporate the three provided words. Step 4: Begin drafting the sonnet by writing the first quatrain (four lines) following the "ABAB" rhyme scheme, ensuring one or more of the provided words are included. Step 5: Continue with the second quatrain "CDCD," the third quatrain "EFEF," and finally the closing couplet "GG," each time incorporating the provided words as needed. Step 6: Review the sonnet for coherence, flow, and adherence to the rhyme scheme, making adjustments as necessary. A.1.3 Common Sense Reasoning Task Description: Given a specific date and an event, such as a holiday or historical event, determine the following date. Solution Description: To determine the next date, we need to consider the structure of the calendar, the number of days in each month, and whether it’s a leap year. Typically, the number of days in a month is fixed, except February may vary due to leap years. The next day in a year is usually the date increased by one day unless it’s the end of the month, then the next day will be the first day of the following month. For the end of the year, the next day will be January 1st of the following year. Thought Template: Step 1: Identify the given date’s month and day number. Step 2: Check if it’s the end of the month; if so, confirm the start date of the next month. Step 3: If it’s not the end of the month, simply add one to the day number. Step 4: Pay special attention to the end of the year, ensuring the year increments. A.1.4 Mathematical Reasoning Task Description: Solve an quadratic equation of the form $ax^{2}+bx+c=0$ considering any situations. Solution Description: To solve any quadratic equation of the form $ax^{2}+bx+c=0$ , we can follow a general approach based on the method described. Here is the structured template for solving such equations: Thought Template: Step 1: Calculate the Discriminant - Compute the discriminant $D$ using the formula $D=b^{2}-4ac$ . Step 2: Determine the Nature of the Roots - If $D>0$ , the equation has two distinct real roots. - If $D=0$ , the equation has exactly one real root (also known as a repeated or double root). - If $D<0$ , the equation has two complex roots. Step 3: Compute the Roots - For $D≥ 0$ , calculate the roots using the formula $x=\frac{-b±\sqrt{D}}{2a}$ . - For $D<0$ , calculate the real and imaginary parts of the complex roots using the formula $x=\frac{-b}{2a}±\frac{\sqrt{-D}}{2a}i$ , where $i$ is the imaginary unit. A.1.5 Code Programming Task Description: When given a list of numbers, try to utilize 4 basic mathematical operations (+-*/) to get a target number. Thought Template: Listing 1: Python template ⬇ from itertools import permutations, product def perform_operation (a, b, operation): # Define the operation logic (e.g., addition, subtraction, etc.). pass def evaluate_sequence (sequence, operations): # Apply operations to the sequence and check if the result meets the criteria. pass def generate_combinations (elements, operations): # Generate all possible combinations of elements and operations. pass def format_solution (sequence, operations): # Format the sequence and operations into a human-readable string. pass def find_solution (input_elements, target_result): # Data Input Handling # Validate and preprocess input data if necessary. # Core Algorithm Logic for sequence in permutations (input_elements): for operation_combination in generate_combinations (sequence, operations): try: if evaluate_sequence (sequence, operation_combination) == target_result: # Data Output Formatting return format_solution (sequence, operation_combination) except Exception as e: # Error Handling # Handle specific exceptions that may occur during evaluation. continue # If no solution is found after all iterations, return a default message. # return No solution found message return # Example usage: input_elements = [1, 7, 10, 3] target_result = 24 print (find_solution (input_elements, target_result)) A.1.6 Application Scheduling Task Description: Given some Chess moves in SAN, update the chess board state. Listing 2: Python template ⬇ import chess def find_checkmate_move (moves_san): # Initialize a new chess board board = chess. Board () # Apply the moves to the board for move_san in moves_san: # Remove move numbers and periods (e.g., "1." or "2.") if len (move_san. split (’. ␣ ’)) > 1: move_san = move_san. split (’. ␣ ’)[1] # Skip empty strings resulting from the removal if move_san: # Apply each move in SAN format to the board move = board. parse_san (move_san) board. push (move) # Generate all possible legal moves from the current position for move in board. legal_moves: # Make the move on a copy of the board to test the result board_copy = board. copy () board_copy. push (move) # Check if the move results in a checkmate if board_copy. is_checkmate (): # Return the move that results in checkmate in SAN format return board. san (move) # return No solution found message return #Example usage: input = ’......’ # Check input format and transform the input into legal format # Remove move numbers and periods (e.g., "1." or "2.") checkmate_move = find_checkmate_move (moves_san) print (checkmate_move) A.2 Prompt for Problem Distiller [Problem Distiller]: As a highly professional and intelligent expert in information distillation, you excel at extracting essential information to solve problems from user input queries. You adeptly transform this extracted information into a suitable format based on the respective type of the issue. Please categorize and extract the crucial information required to solve the problem from the user’s input query, the distilled information should include. 1. Key information: Values and information of key variables extracted from user input, which will be handed over to the respective expert for task resolution, ensuring all essential information required to solve the problem is provided. 2. Restrictions: The objective of the problem and corresponding constraints. 3. Distilled task: Extend the problem based on 1 and 2, summarize a meta problem that can address the user query and handle more input and output variations. Incorporate the real-world scenario of the extended problem along with the types of key variables and information constraints from the original problem to restrict the key variables in the extended problem. After that, use the user query input key information as input to solve the problem as an example. A.3 Prompt for Instantiated Reasoning [Meta Reasoner] You are a Meta Reasoner who are extremely knowledgeable in all kinds of fields including Computer Science, Math, Physics, Literature, History, Chemistry, Logical reasoning, Culture, Language….. You are also able to find different high-level thought for different tasks. Here are three reasoning sturctures: i) Prompt-based structure: It has a good performance when dealing with problems like Common Sense Reasoning, Application Scheduling ii) Procedure-based structure It has a good performance when dealing with creative tasks like Creative Language Generation, and Text Comprehension iii) Programming-based: It has a good performance when dealing with Mathematical Reasoning and Code Programming, it can also transform real-world problems into programming problem which could be solved efficiently. (Reasoning instantiation) Your task is: 1. Deliberately consider the context and the problem within the distilled respond from problem distiller and use your understanding of the question within the distilled respond to find a domain expert who are suitable to solve the problem. 2. Consider the distilled information, choose one reasoning structures for the problem. 3. If the thought-template is provided, directly follow the thought-template to instantiate for the given problem.