LLM4Cov: Execution-Aware Agentic Learning for High-coverage Testbench Generation

TL;DR

This paper introduces LLM4Cov, an offline learning framework for high-coverage hardware verification that models verification as memoryless state transitions, enabling scalable learning with execution constraints and achieving high coverage with a compact model.

Contribution

The paper presents a novel offline agent-learning framework for verification, including data curation, policy-aware synthesis, and prioritized sampling, tailored for execution-aware LLM agents.

Findings

A 4B-parameter model achieves 69.2% coverage pass rate.

Outperforms its teacher model by 5.3%.

Demonstrates competitive performance with larger models.

Abstract

Execution-aware LLM agents offer a promising paradigm for learning from tool feedback, but such feedback is often expensive and slow to obtain, making online reinforcement learning (RL) impractical. High-coverage hardware verification exemplifies this challenge due to its reliance on industrial simulators and non-differentiable execution signals. We propose LLM4Cov, an offline agent-learning framework that models verification as memoryless state transitions guided by deterministic evaluators. Building on this formulation, we introduce execution-validated data curation, policy-aware agentic data synthesis, and worst-state-prioritized sampling to enable scalable learning under execution constraints. We further curate a reality-aligned benchmark adapted from an existing verification suite through a revised evaluation protocol. Using the proposed pipeline, a compact 4B-parameter model achieves 69.2% coverage pass rate under agentic evaluation, outperforming its teacher by 5.3% and demonstrating competitive performance against models an order of magnitude larger.