Downloads · 30 days
0
Pablo-Flores-Mollinedo/verilog-qwen2.5-coder-7b-v23-runtime-verifier-selector-pipeline
verilog-qwen2.5-coder-7b-v23-runtime-verifier-selector-pipeline is a machine learning model from Pablo-Flores-Mollinedo. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
v23 is a multi-adapter runtime pipeline, not a single adapter.
Downloads · 30 days
0
Access
Public
Updated Jun 3, 2026
Repo size
—
Likes
0
Public
Click a slice to open those files.
.jsonl16.5 KB · 56%
From the Hugging Face model README
v23 is a multi-adapter runtime pipeline, not a single adapter.
For each VerilogEval v2 prompt, it generates and verifies candidates in priority order:
v21_comb_vectorv21_counter_timerv21_seq_shiftv21_case_kmapv21_fsm_protocolv9Selection rule:
choose first candidate that passes compile + simulation
else choose first compiling candidate
else fallback to first candidate
compile 149/156 = 95.51%
pass 79/156 = 50.64%
This matches the existing-output verifier-selector prototype and exceeds the prior best pipeline:
v9 k=5 + v7b repair = 76/156
v23 uses testbench/ref verification at selection time. It is a practical verifier-based system when tests exist, but not a clean single-model zero-shot leaderboard claim.
77 problems remain failing under v23 runtime. Category breakdown:
fsm_protocol: 20
seq_shift: 16
comb_vector: 9
case_kmap: 6
counter_timer: 5
other: 21
v25 goal: add hard-problem adapters specialized on these remaining failures, then plug them into the v23 verifier selector.