Shall we default all the datasets and traces on HF to v2?

[reacher-z]

Question

Shall we default all the datasets and traces on HuggingFace to v2?

Right now it isn't obvious whether every link on our surfaces resolves to v2, and the version/ownership story is inconsistent. We've also had feedback that the website isn't clear about which version a visitor is looking at.

What I'm seeing on claw-bench.com

The links currently point at a mix of orgs and versions:

Resource	Current link
HF Dataset	huggingface.co/datasets/NAIL-Group/ClawBench
HF Traces	huggingface.co/datasets/TIGER-Lab/ClawBenchV2Trace
HF V1 Traces	huggingface.co/datasets/NAIL-Group/ClawBenchV1Trace
GitHub	github.com/TIGER-AI-Lab/ClawBench

So the main dataset lives under NAIL-Group while traces live under TIGER-Lab, and v1/v2 are interleaved without a clear default.

Proposal

1. Decide whether v2 is the canonical default everywhere (dataset + traces).
2. Make every "Dataset" / "Traces" link on the website, README, and HF cards resolve to the v2 artifacts by default, with v1 clearly marked as legacy/archive.
3. Consolidate to a single canonical org so the dataset and traces don't split across NAIL-Group vs TIGER-Lab.

References

• HF dataset: https://huggingface.co/datasets/TIGER-Lab/ClawBench
• Website: https://claw-bench.com
• For clarity/IA inspiration: https://programbench.com

Feedback: a few people have said the website isn't very clear right now about which version is the default — worth addressing alongside this.