Port ML models into mflux/MLX with deterministic parity tests.
npx skills add https://github.com/filipstrand/mflux --skill mflux-model-porting
Please help me install this Agent Skill. Skill: mflux-model-porting Source: https://github.com/filipstrand/mflux/tree/main/.cursor/skills/mflux-model-porting Command: npx skills add https://github.com/filipstrand/mflux --skill mflux-model-porting
Port ML models into mflux/MLX with correctness-first validation, then refactor toward mflux style for maintainable, robust integrations.
Create a port plan that mirrors a reference model structure in an mflux port directory, implement a minimal port skeleton, and run deterministic tests to validate parity.
High-intent search queries and answers about installing and using this skill.
To port ML models to MLX with correctness, establish a skeleton port, wire weight mappings, implement a minimal runner, and run deterministic tests to validate outputs against a reference implementation.
Deterministic testing for ML model porting involves creating a skeleton package and running milestone tests that validate output parity against reference implementations before refactoring toward shared components.
Yes, the porting workflow is adaptable to model families including diffusion models, allowing you to port them into mflux while preserving behavior through incremental validation and deterministic tests.
When porting a model to MLX, you wire weight mappings within a skeleton port directory that mirrors the reference model structure, enabling minimal runners to verify outputs against the reference.
The best way to refactor ported ML models for maintainability is to use incremental validation with deterministic tests and commit milestones, guiding refactors into shared components across model families.
hao-ai-labStage reference code, convert weights, and validate compatibility for FastVideo models.
J4JoshuaOptimize local AI inference on Apple Silicon using MLX acceleration.
scouzi1966Analyze HuggingFace MLX model config.json files to onboard them into AFM registries and Swift backend workflows.
mattmirelesAnalyzes CoreML model execution on Apple Silicon to detect compute units, silent fallbacks, and bottlenecks.
