204 个 · 按相关度排序
Convert PyTorch AT_DISPATCH macros to AT_DISPATCH_V2 format in ATen C++ code. Use when porting AT_DISPATCH_ALL_TYPES_AND*, AT_DISPATCH_FLOATING_TYPES*, or other dispatch macros to the new v2 API. For ATen kernel files, CUDA kernels, and native operator implementations.
中文简介正在准备中。当前显示原文说明。
Routes official FlashVSR diffusion-based streaming video super-resolution setup, model preparation, and CUDA inference for v1 and v1.1.
中文简介正在准备中。当前显示原文说明。
Use the faster-whisper package for CTranslate2-backed Whisper transcription, model selection, CPU/CUDA setup, audio utilities, VAD, timestamps, and conversion guidance.
中文简介正在准备中。当前显示原文说明。
Use Gaussian-SLAM for CUDA RGB-D dense SLAM, Gaussian-splatting mapping, dataset/configuration preflight, checkpoint evaluation, reconstruction metrics, and global-map refinement.
中文简介正在准备中。当前显示原文说明。
Routes DINO DETR object-detection setup, COCO data validation, CUDA deformable-attention preparation, training, evaluation, inference, visualization, and bounded benchmarking tasks.
中文简介正在准备中。当前显示原文说明。
Use DreamCraft3D for image-conditioned 3D generation, staged optimization, texture boosting, export, and CUDA troubleshooting.
中文简介正在准备中。当前显示原文说明。
Guides CUDA-accelerated cuRobo v2 robotics workflows for robot models, differentiable kinematics, collision-aware IK and motion planning, trajectory optimization, MPC, perception mapping, and motion retargeting.
中文简介正在准备中。当前显示原文说明。
Guide DeepMedic 0.8.4 workflows for multi-modal 3D NIFTI medical-image segmentation, including data preparation, model configuration, training, checkpoint continuation, inference, and CPU/CUDA troubleshooting.
中文简介正在准备中。当前显示原文说明。
Guide Det3D PyTorch 3D object-detection workflows across configuration, KITTI/nuScenes/Lyft data preparation, GPU training and evaluation, custom CUDA operations, and visualization.
中文简介正在准备中。当前显示原文说明。
Guide CUDA-enabled BindCraft protein-binder design from target PDB preparation through AF2/MPNN/PyRosetta execution, filtering, output analysis, and conservative troubleshooting.
中文简介正在准备中。当前显示原文说明。
Guides Attention-Gated Networks PyTorch medical imaging workflows for ultrasound classification, attention-gated U-Net segmentation, data layout, CUDA setup, and visualization helpers.
中文简介正在准备中。当前显示原文说明。
Routes Big Sleep text-to-image workflows, CUDA setup checks, prompt controls, and Python API usage for the `dream` CLI and `Imagine` API.
中文简介正在准备中。当前显示原文说明。
Guides CUDA-based BigGAN-PyTorch training, checkpoint sampling, dataset preparation, metric preparation, model customization, and optional TensorFlow Hub weight conversion.
中文简介正在准备中。当前显示原文说明。
Use the ALAE repository for Adversarial Latent Autoencoder data preparation, CUDA training, checkpoint-backed generation, latent editing, and legacy metrics.
中文简介正在准备中。当前显示原文说明。
Think and work like an expert High-Performance Computing Specialist. Use when a task calls for High-Performance Computing Specialist judgment. Reasons from NUMA topology and hybrid MPI+OpenMP+CUDA decomposition through Slurm fairshare/backfill job design, strong/weak scaling (Amdahl/Gustafson), Darshan/mpiP/Nsight profiling, and parallel HDF5/MPI-IO on Lustre while treating I/O storms, collectives bottlenecks, and rank-binding mistakes as first-class failure modes.
中文简介正在准备中。当前显示原文说明。
Deploy AI models to embedded hardware using MathWorks tools (MATLAB, Simulink, Embedded Coder). Covers two workflow patterns: (1) MathWorks-native or 3P-imported models rebuilt as dlnetwork for lean hardware (Cortex-M, DSP), (2) direct C/C++ code generation from PyTorch and LiteRT models for high-performance hardware (Cortex-A, x86, GPU). Trigger when: user wants to deploy AI to embedded targets; generate C/CUDA from neural networks; compress AI models for MCU/DSP; integrate AI in Simulink for system-level simulation; import PyTorch/ONNX/TensorFlow models for embedded deployment; optimize AI for resource-constrained hardware; or use loadPyTorchExportedProgram, importNetworkFromPyTorch, dlquantizer, exportNetworkToSimulink, or Embedded Coder with AI models.
中文简介正在准备中。当前显示原文说明。
Let AKA autonomously add, run, inspect, and revise intra-kernel timeline probes for standalone CUDA/inline PTX or CuTe DSL when ordinary benchmark, NSYS, or NCU evidence cannot answer a specific kernel-internal timing question.
中文简介正在准备中。当前显示原文说明。
Use when adding support for a new model architecture to imp, porting a model family, or debugging a model that loads but produces wrong output - "add support for <model>", "new arch", loader detection, chat template, tokenizer parity, RoPE variant, "outputs garbage", "prompt-blind", "digits scrambled", "NaN logits", "describes a different picture", "does it fit in VRAM". Do NOT use for kernel performance (sm120-cuda-expert) or quant-format questions (quant-formats).
中文简介正在准备中。当前显示原文说明。
Use when verifying that a model in the imp inference engine produces coherent output without repetition loops, token-stuck states, or state corruption across turns or streams. Triggers on "degenerates", "check degeneration", "repetition loop", "own own own", "stuck token", "empty content", "multi-turn regression", "does it still work", "NIAH", and after enabling CUDA graphs / changing forward pass / MoE routing / KV cache or KV dtype / GDN state or scan / sparse attention / PDL / speculation (MTP, n-gram) / ragged prefill / batched-decode kernels (smallm, producer quantize) / FA2 softmax.
中文简介正在准备中。当前显示原文说明。
Use when imp outputs differ between two paths that should agree - chunk sizes, prefill vs decode, prefix-cache resend, two engines (imp vs llama.cpp vs HF), two builds or kernels - "output changes with chunk size", "first token differs", "logprobs moved", "which layer diverges", "who is right, imp or llama.cpp", "near-tie". Do NOT use for repetition loops or garbage output (check-degeneration), a new arch that loads wrong (add-model-arch), or perf (benchmark-cuda).
中文简介正在准备中。当前显示原文说明。
Use when working on or testing imp-server and its HTTP APIs - OpenAI/Anthropic endpoints, /v1/chat/completions, /v1/messages, /v1/responses, SSE streaming, tool calling, json_schema/regex/GBNF constrained decoding, thinking/reasoning_content, cache_control/prefix cache, model loading and swapping, request priority, X-Request-Id, OTLP tracing / traceparent / Jaeger, /metrics, container env (IMP_SET), "server returns 404/garbage", API compliance. Do NOT use for CLI-only inference (imp-cli), kernel work (sm120-cuda-expert), or pure output-quality checks (check-degeneration).
中文简介正在准备中。当前显示原文说明。
Use when opening, merging, or releasing a PR for imp - branching off main, `gh pr create`, enabling auto-merge, writing the PR body or a CHANGELOG entry, cutting a tagged release (version bump + CHANGELOG + tag + GitHub release). Symptoms - "PR stuck BLOCKED", "my last commit didn't land on main", "which check is required", "how do I cut a release", "auto-merge merged too early", "check-release failed", "STALE.md blocks git pull". Do NOT use for build/test mechanics (building-and-testing) or perf measurement / baseline refresh (benchmark-cuda).
中文简介正在准备中。当前显示原文说明。
Use when keeping imp's docs and config examples coherent after a change - updating docs/internals/ARCHITECTURE.md / README / docs/GOAL.md / docs/MODELS.md / docs/ROADMAP.md / imp.conf.example / CHANGELOG / MISSION_JOURNAL, or "is this doc stale", "document this change", "the example config is out of date", "the README says X but the code does Y", "add a roadmap ledger row". Do NOT use for layer/frontmatter/lint questions (docs-layers), structural code audits (codebase-audit), measuring perf or refreshing the perf baseline (benchmark-cuda), or the agent's private memory (MEMORY.md).
中文简介正在准备中。当前显示原文说明。
Use when analyzing warp divergence, memory coalescing, shared memory bank conflicts, cache behavior, atomics, or occupancy tradeoffs on NVIDIA and AMD GPUs. Not for tool commands: use cuda-profiling.
中文简介正在准备中。当前显示原文说明。