Modernizing my old PhD work in an evening with little Qwen3.6 MoE

Name: Modernizing my old PhD work in an evening with little Qwen3.6 MoE
Availability: InStock
Author: verdverm

by verdverm·May 21, 2026·3 points·0 comments

Visit Project View on HN

AI Analysis

●●●BangerBig BrainWizardry

Pure JAX pipeline enables GPU acceleration where previous Python implementations stalled.

Strengths

•Replaces slow CPU-bound evaluation with JIT-compiled JAX operations for massive speedups.
•Implements multi-objective Pareto front selection without external ML dependencies.
•Automatic differentiation handles Jacobian computation internally for gradient optimization.

Weaknesses

•Niche academic algorithm limits appeal to general machine learning practitioners.
•Documentation assumes familiarity with symbolic regression concepts and grammar rules.

Similar Projects

AI/ML●●Solid

I ran Qwen3.5 35B on my iPhone at 5.6 tok/SEC

Runs 19.5GB Qwen3.5 on 12GB RAM iPhone via memory swapping.

WizardryBold Bet

alexintosh

422mo ago

AI/ML●●Solid

35B MoE LLM and other models locally on an old AMD crypto APU (BC250)

Kernel ttm.pages_limit workaround unlocks 16GB UMA for Vulkan inference on repurposed crypto hardware.

Dark HorseNiche Gem

akandr

202mo ago

Design○Pass

Modern CSS – Every old CSS hack paired with a clean, modern replacement

Side-by-side snippets (old hack → modern CSS) with difficulty, browser-compat scores and a playground make this an instant lookup for refactors. The curation is pragmatic — examples like translate-centering → place-items, scrollbar-gutter, and OKLCH color swaps show you exactly what to replace and why. I’d love inline live sandboxes and per-browser caveats for tricky edge cases, but the site already shaves hours off common CSS archaeology.

Crowd PleaserSlickSolve My Problem

naeemnur

204mo ago

AI/ML●●●Banger