core-labs/core
core-labs/core
1
0
Fork 0
Code Issues Pull requests 1 Projects Releases Packages Wiki Activity Actions
core/docs/benchmarks/apple-uma-mlx-exact-recall-status.md
Shay b30716a19c
feat(bench): add MLX exact CGA recall experiment (#909) 
* feat(bench): add MLX exact CGA recall experiment

* test(bench): cover MLX exact recall experiment contracts

* docs(bench): document MLX exact recall experiment

* docs(bench): record MLX experiment integration status

* docs(bench): add MLX local validation handoff
2026-06-24 13:53:50 -07:00

1 KiB
Raw Blame History

Apple UMA MLX Exact Recall Status

This branch intentionally lands the MLX exact CGA recall lane as an isolated benchmark module first.

Implemented

  • benchmarks/apple_uma_mlx_exact_recall.py
  • tests/test_apple_uma_mlx_exact_recall.py
  • docs/benchmarks/apple-uma-mlx-exact-recall.md

Not yet integrated

The main benchmarks/apple_uma_mechanical_sympathy.py report has not yet been modified on this branch.

Reason: the isolated module is the safest first landing surface for local Apple Silicon validation. After MLX is verified locally, the module can be integrated into the main Apple UMA report as the mlx_exact_cga_recall track.

Local validation request

Run:

uv run python -m pytest -q tests/test_apple_uma_mlx_exact_recall.py
uv run python -m benchmarks.apple_uma_mlx_exact_recall --json

When MLX is installed:

CORE_BACKEND=rust uv run python -m benchmarks.apple_uma_mlx_exact_recall --json

All emitted cases must show parity.parity_pass: true before the results are used in Apple-facing material.