A training-free, three-stage abstraction prompting method modestly improves LLM-generated Verilog functionality on VerilogEval for GPT-4o and boosts smaller models when paired with large-model-generated abstractions.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.PL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Abstractions-of-Thought: Intermediate Representations for LLM Reasoning in Hardware Design
A training-free, three-stage abstraction prompting method modestly improves LLM-generated Verilog functionality on VerilogEval for GPT-4o and boosts smaller models when paired with large-model-generated abstractions.