Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Retrieval-Augmented Instruction Tuning for Automated Process Engineering Calculations : A Tool-Chaining Problem-Solving Framework with Attributable Reflection

Sagar Srinivas Sakhinana, Geethan Sannidhi, Venkataramana Runkana

TL;DR

The paper tackles the lack of a domain-specific foundational AI for chemical and process engineering calculations and proposes Retrieval-Augmented Instruction-Tuning (RAIT) to adapt open code SLMs with external tools and curated datasets. It introduces an autonomous agent framework that combines RACG with ReAct prompting, a five-stage workflow, program caching, and an attributable reflection mechanism to generate, debug, and optimize executable code from natural language specifications. Central contributions include the MathComp and ChemProc instruction-tuning datasets, a detailed RAIT-based agent architecture, and comprehensive evaluations (including ablations and human assessments) showing competitive performance with large proprietary models. The work highlights improved explainability, knowledge editing capabilities, and cost-efficiency, signaling a practical pathway for deploying specialized AI tools in the chemical process industry.

Abstract

The current technology landscape lacks a foundational AI model for solving process engineering calculations. In this work, we introduce a novel autonomous agent framework leveraging Retrieval-Augmented Instruction-Tuning (RAIT) to enhance open, customizable small code language models (SLMs) for these calculations. By combining instruction tuned code SLMs with Retrieval-Augmented Code Generation (RACG) using external tools, the agent generates, debugs, and optimizes code from natural language specifications. Our approach addresses the limitations of the current lack of a foundational AI model for specialized process engineering tasks and offers benefits of explainability, knowledge editing, and cost-effectiveness. Additionally, we curate custom datasets of chemical and process engineering problems and solutions to overcome data scarcity. Experimental results show that our framework matches the performance of large-scale proprietary models on benchmark datasets, proving its effectiveness and usability.

Retrieval-Augmented Instruction Tuning for Automated Process Engineering Calculations : A Tool-Chaining Problem-Solving Framework with Attributable Reflection

TL;DR

The paper tackles the lack of a domain-specific foundational AI for chemical and process engineering calculations and proposes Retrieval-Augmented Instruction-Tuning (RAIT) to adapt open code SLMs with external tools and curated datasets. It introduces an autonomous agent framework that combines RACG with ReAct prompting, a five-stage workflow, program caching, and an attributable reflection mechanism to generate, debug, and optimize executable code from natural language specifications. Central contributions include the MathComp and ChemProc instruction-tuning datasets, a detailed RAIT-based agent architecture, and comprehensive evaluations (including ablations and human assessments) showing competitive performance with large proprietary models. The work highlights improved explainability, knowledge editing capabilities, and cost-efficiency, signaling a practical pathway for deploying specialized AI tools in the chemical process industry.

Abstract

The current technology landscape lacks a foundational AI model for solving process engineering calculations. In this work, we introduce a novel autonomous agent framework leveraging Retrieval-Augmented Instruction-Tuning (RAIT) to enhance open, customizable small code language models (SLMs) for these calculations. By combining instruction tuned code SLMs with Retrieval-Augmented Code Generation (RACG) using external tools, the agent generates, debugs, and optimizes code from natural language specifications. Our approach addresses the limitations of the current lack of a foundational AI model for specialized process engineering tasks and offers benefits of explainability, knowledge editing, and cost-effectiveness. Additionally, we curate custom datasets of chemical and process engineering problems and solutions to overcome data scarcity. Experimental results show that our framework matches the performance of large-scale proprietary models on benchmark datasets, proving its effectiveness and usability.
Paper Structure (8 sections, 4 equations, 1 figure, 20 tables)

This paper contains 8 sections, 4 equations, 1 figure, 20 tables.

Table of Contents

  1. Introduction
  2. Proposed Method
  3. Experiments
  4. Datasets:
  5. Experimental Settings:
  6. Evaluation Metrics:
  7. Experimental Results:
  8. Conclusion

Figures (1)

  • Figure 1: The figure depicts a single-agent framework for question answering (QA) for solving complex chemical and process engineering calculations.