Multi-agent LLMs automate testbench creation to produce efficient fine-tuning data, enabling a model to match state-of-the-art performance on specification-to-Verilog generation with less training data on the VerilogEval v2 benchmark.
Insights from verification: Training a verilog generation llm with reinforcement learning with testbench feedback,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AR 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Exploring LLM-based Verilog Code Generation with Data-Efficient Fine-Tuning and Testbench Automation
Multi-agent LLMs automate testbench creation to produce efficient fine-tuning data, enabling a model to match state-of-the-art performance on specification-to-Verilog generation with less training data on the VerilogEval v2 benchmark.