skills.rest
Just-it
Just-itCOMMUNITY
Explore all skills by this author →
GitHub

dsl-baseline-generation

Generate baseline AscendDSL implementations from functional PyTorch operators.

33|51|Updated Mar 12, 2026
Search Tags:#kernel#pytorch#tiling#code-generation#dsl#reference-guided#ascenddsl
One-click install
npx skills add https://github.com/Just-it/AscendOpGenAgent --skill dsl-baseline-generation
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: dsl-baseline-generation
Source: https://github.com/Just-it/AscendOpGenAgent/tree/main/skills/dsl_baseline_generation
Command: npx skills add https://github.com/Just-it/AscendOpGenAgent --skill dsl-baseline-generation

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

This skill enables auto-generation of a baseline AscendDSL implementation from a functional PyTorch operator, streamlining the transition from Python-level models to device-specific DSL code.

Core Features & Use Cases

  • Automatically scaffolds host partitioning and kernel tiling for AscendDSL.
  • Uses example references to guide input/output shapes and implementation patterns.
  • Quick-start path from a PyTorch function to a ready-to-run DSL file.

Quick Start

Place your PyTorch operator file named {op_name}_functional.py and run the generator to produce output/{op_name}/{op_name}_dsl.py.

Frequently Asked Questions about dsl-baseline-generation

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I convert a PyTorch operator to AscendDSL code?▼

To convert a PyTorch operator to AscendDSL, place your functional file named {op_name}_functional.py and run the generator to produce a baseline DSL implementation with host partitioning and kernel tiling.

What is AscendDSL code generation used for?▼

AscendDSL code generation is used to transition functional PyTorch operators to device-specific DSL code, scaffolding host planning and tiling strategies for kernel development.

Does this PyTorch to DSL generator require manual tiling configuration?▼

No manual tiling configuration is required. The generator automatically scaffolds host partitioning and kernel tiling by analyzing the PyTorch operator and reference examples.

Can I use custom input examples to guide AscendDSL generation?▼

Yes, you can use custom examples. The generator reads references/input_example and references/output_example to guide input/output shapes and implementation patterns for the generated code.

How does the generator validate AscendDSL interface compatibility?▼

The generator validates interface compatibility with module_fn before saving the generated AscendDSL code to output/{op_name}/{op_name}_dsl.py.

What are the limitations of automatic AscendDSL baseline generation?▼

The automatic generation produces a baseline AscendDSL implementation from functional PyTorch code. It relies on reference examples for patterns and reads knowledge from references/ascendDSL.py to guide structure.

Related Skills

Explore More Software Engineering Skills→
mindspore-aimindspore-ai
258

kernel-generator

Generate operator kernels across multiple DSLs and backends.

Official
Advanced
AscendAscend
31

ascendc-operator-project-init

Generate an AscendC operator skeleton under csrc/ops with host and kernel placeholders.

Official
Advanced
mindspore-aimindspore-ai
258

tilelang-cuda-examples-torch

Run TileLang CUDA Torch examples for GEMM, GEMV, and LayerNorm kernels.

Official
Intermediate
AscendAscend
31

ascendc-operator-code-gen

Generate AscendC operator host/kernel code from design docs and integrate into PyTorch.

Official
Advanced