Stats
Actions
Available In
Tags
Plugin
PyTorch AI Marketplace
Curated collection of AI tools, skills, and agents for PyTorch development. Includes torch.compile debugging, semantic API search, and development workflows.
What's Inside
Agents4
dynamo-expert-agent
/dynamo-expert
Dynamo specialist for graph capture, guards, graph breaks, and VariableTracker system
aot-expert-agent
/aot-expert
AOT Autograd specialist for functionalization, decomposition, and gradient computation
inductor-expert-agent
/inductor-expert
Inductor specialist for lowerings, IR nodes, Triton codegen, fusion, and kernel optimization
compile-debug
/compile-debug
Skill-driven torch.compile debugger. Orchestrates bisection, trace collection, and root cause analysis using stage-specific skills. Use when debugging compilation failures, errors, or incorrect output.
Skills8
compile-bisect
/compile-bisect
Use PyTorch's compiler bisector to automatically find which backend/subsystem/operation causes compilation failures. Binary searches through backends (eager → aot_eager → inductor) and subsystems (passes, lowerings, etc.) to pinpoint exact failing operations. Outputs backend/subsystem/debug_info that routes you to the right stage-specific skill (compile-trace-dynamo for eager, compile-trace-aot for aot_*, compile-trace-inductor for inductor).
compile-overview
/compile-overview
Reference documentation for torch.compile pipeline architecture, IR levels (Full ATen, Core ATen, Prims), stages, TORCH_LOGS flags, output files, and debugging tools. Use for understanding pipeline structure, operator IRs, and available debugging options.
compile-trace-aot
/compile-trace-aot
Debug PyTorch AOT Autograd stage - functionalization, decompositions, IR transformations, joint forward+backward graph (when requires_grad=True), partitioning/recomputation, and post-grad passes. Use for tracing AOT stage and understanding decomposition application.
compile-trace-dynamo
/compile-trace-dynamo
Debug PyTorch Dynamo stage - bytecode capture, FX graph construction, graph breaks, and pre-grad passes. Covers TORCH_LOGS for dynamo/graph_breaks/pre_grad_graphs, interpreting FX graph files, understanding graph break reasons, and pre-grad fusion patterns (Conv-BN, split-cat). Load after compile-bisect indicates backend='eager'.
compile-trace-inductor
/compile-trace-inductor
Debug PyTorch Inductor compiler backend - IR lowering, scheduler/fusion, loopbody ops, and Triton/C++ codegen. Covers TORCH_LOGS for fusion/schedule/ir_post_fusion/output_code, config.trace.enabled for IR dumps, interpreting Inductor IR nodes (Pointwise/Reduction/etc), fusion decisions, kernel generation, and performance optimization. Load after compile-bisect indicates backend='inductor'.
MCP Servers1
Stats
Version1.0.0
LanguagePython
Stars0
MaintenanceGood
LicenseBSD-3-Clause
Last Commit
Added
Available In
Safety Signals
Caution
Uses power tools
Uses Bash, Write, or Edit tools
Tags
