Hephaestus: Improving Fundamental Agent Capabilities of Large Language Models through Continual Pre-Training

TL;DR

Hephaestus introduces a large-scale, agent-specific pre-training corpus that significantly enhances LLMs' abilities in API calling, reasoning, and adaptation, outperforming existing models on key benchmarks.

Contribution

The paper presents Hephaestus-Forge, a novel 103B data corpus for continual pre-training that improves fundamental agent capabilities of large language models.

Findings

Hephaestus outperforms small- to medium-scale open-source LLMs.

Hephaestus rivals commercial LLMs on agent benchmarks.

Continual pre-training enhances generalization to new tasks.

Abstract

Due to the scarcity of agent-oriented pre-training data, LLM-based autonomous agents typically rely on complex prompting or extensive fine-tuning, which often fails to introduce new capabilities while preserving strong generalizability. We introduce Hephaestus-Forge, the first large-scale pre-training corpus designed to enhance the fundamental capabilities of LLM agents in API function calling, intrinsic reasoning and planning, and adapting to environmental feedback. Hephaestus-Forge comprises 103B agent-specific data encompassing 76,537 APIs, including both tool documentation to introduce knowledge of API functions and function calling trajectories to strengthen intrinsic reasoning. To explore effective training protocols, we investigate scaling laws to identify the optimal recipe in data mixing ratios. By continual pre-training on Hephaestus-Forge, Hephaestus outperforms small- to medium-scale open-source LLMs and rivals commercial LLMs on three agent benchmarks, demonstrating the effectiveness of our pre-training corpus in enhancing fundamental agentic capabilities and generalization of LLMs to new tasks or environments.