Update README.md

README.md

@@ -24,7 +24,7 @@ The questions, ground-truth answers, evaluation harness, and per-core generation
 
 ## Files
 
-A single archive, `waveformqa_vcds.zip` (~364 MB; ~3.3 GB extracted), containing a `LICENSE` file and one IEEE 1364 VCD file per trace:
+A single archive, `waveformqa_vcds.zip` (~364 MB; ~3.3 GB extracted), containing a `LICENSE` file (the CC-BY-4.0 terms and third-party attributions below) and one IEEE 1364 VCD file per trace:
 
 | Core | Upstream license | Traces | Signal range |
 |------|------------------|--------|--------------|
@@ -40,7 +40,7 @@ Download the archive, extract it into `vcds/` in a clone of the repository, and
 
 ```bash
 hf download dcummingsTT/waveformqa waveformqa_vcds.zip --repo-type dataset --local-dir vcds
-unzip -o vcds/waveformqa_vcds.zip -x '__MACOSX/*' -d vcds
+unzip -o vcds/waveformqa_vcds.zip -d vcds
 rm vcds/waveformqa_vcds.zip   # optional: delete the archive once extracted
 uv run python scripts/run_benchmark.py benchmark/t5000 --vcd-dir vcds --eval-only \
     -o output/t5000/evaluation_results.json