Claude Skills by VectorSpaceLab

"Use VLMEvalKit (`vlmeval`) for vision-language model evaluation,

"Author and adapt VLMEvalKit benchmark datasets, TSV/video schemas,

"Run and troubleshoot VLMEvalKit evaluations, result reuse, async

"Configure and extend VLMEvalKit model wrappers, API providers,

"Use VLM-R1 for multimodal GRPO training, JSONL reward/data design,

"Route Huawei Ascend vllm-ascend and XLLM inference for VLM-R1 OVD

"Validate VLM-R1 JSONL datasets, map image roots, and choose or

"Adapt VLM-R1 REC and OVD evaluation recipes and score saved bbox

"Add and debug VLM-R1 model modules, including Qwen2VL, InternVL,

"Construct, audit, and adapt VLM-R1 GRPO training launches for

"Use VoxelMorph's current PyTorch image-registration package for

"Use VoxelMorph volume I/O, npz/NIfTI/MGZ data conventions,

"Construct and smoke-test VoxelMorph VxmPairwise registration

"Use VoxelMorph PyTorch tensor transform operations for affine,

"Use the NVlabs VoxFormer repository for camera-only 3D semantic

"Validate and safely prepare SemanticKITTI-derived files for

"Install and preflight the legacy VoxFormer/OpenMMLab CUDA stack,

"Choose, inspect, validate, and safely adapt VoxFormer QPN,

"Plan and safely launch VoxFormer stage-1 QPN and stage-2

"Guide static and backend-aware Walk These Ways Go1 simulation

"Validate Go1 deployment logs, build six-feature per-joint actuator

"Prepare and review Unitree Go1 deployment artifacts, LCM

"User-facing Go1 Isaac Gym environment and configuration guidance:

"Operate the Walk These Ways PPO and PPO-CSE/RMA policy workflow:

"Use W&B's Python SDK and CLI for experiment tracking,

"Create, log, consume, download, link, and troubleshoot W&B

"Use W&B CLI and local workflows safely: login, init, status,

"Add, debug, and validate W&B experiment tracking in Python scripts

"Query, export, and automate W&B resources with wandb.Api,

"Define and run W&B hyperparameter sweeps and W&B Launch jobs,

"Guides Waymo Open Dataset package workflows for autonomous-driving

"Guides Waymo Open Dataset camera custom ops, camera-only

"Guides Waymo Open Dataset v1 Frame protos, range-image parsing,

"Guides real-time Waymo Open Dataset latency challenge submissions,

"Guides Waymo Open Dataset detection, tracking, motion, keypoint,

"Guides Waymo Open Motion Dataset scenarios, occupancy flow,

"Guides Waymo Open Dataset maintainer workflows for Bazel, Docker,

"Guides Waymo Open Dataset V2 columnar components, Parquet tags,

"Guide WebDataset shard loading, shard writing, and

"Route WebDataset URL opening, stream caching, secure mode, custom

"Guide agents that need to read, transform, decode, batch, sample,

"Route WebDataset tar-shard creation, rollover, sample encoding,

"Route Weighted-Boxes-Fusion tasks for 2D box ensembling, 1D span

"Route 2D normalized box ensembling tasks with weighted boxes

"Use weighted_boxes_fusion_1d for normalized 1D spans, token-span

Fuse normalized 3D cuboid detections with weighted_boxes_fusion_3d.

"Use WeNet for ASR package transcription, data preparation,

"Prepare and validate WeNet data manifests, dictionaries,

"Preflight and run WeNet checkpoint export to TorchScript, ONNX CPU

"Use WeNet as an installed Python package or console CLI for speech