Nicheloom visits: 66,973 all-time · 9,621 in the last 7 days

The opportunity tracker for new startups.

Today: 1,339 launches · Niche of the day: game save editors and managers · See what launched today →

Watch weight-space programming in action with WebGPU

Get picks like this daily. Today's launch numbers, the top launches, AI and tech news, and a daily opportunity pick, straight to your inbox.

Want to reach these readers? Sponsor an issue.

I wanted to build a full weight-space programming loop entirely in the browser with WebGPU: let an LLM inspect Qwen3.5-0.8B, edit its weights, evaluate, and repeat.the full loop needed too much VRAM. so I turned it into an interactive walkthrough of a concrete example: editing 0.016% of a model's weights to reduce reasoning loops, no gradient descent.Based on exp8 of “Sparks of Weight-Space Programming.”: https://news.ycombinator.com/item?id=50341174

HN · launched Nov 19, 2026 · 40 upvotes (top 20%) · 0 views

Niche: llm optimization tools · 134 launches, 6 in the last 7 days ▼ -45% · Explore this niche →

Category and tags

  • Horizontal
  • Dev tools
  • Developer
  • AI-native
  • Hobby / open-source project

Visit website ↗ Source listing ↗

Could you build this?

Partial The interactive replay page and browser UI are vibe-codeable, but the core is a novel method for finding and subtracting a behavioral direction in LLM weights, which is research-level work that vibe coding alone won't produce.

What it would actually take: The UI is a WebGPU or WASM inference viewer with recorded token streams and weight-diff visualizations. The hard part is the method itself: identifying a repetition-loop direction via activation analysis, choosing which FFN neurons to edit, and validating that the edit generalizes. That requires ML research skills and GPUs to run the base and edited models at scale.

Discussion

3 comments analyzed.

Concerns raised: laborious to read and odd format, visual noise and LLM-style prose, no visible human review or iteration, AI-generated content assumed to take no effort

Feature requests: remove extra visual clutter, keep only the browser run feature

Competitors

Other products that read as similar to this one : 70 launches clear the similarity bar, closest 8 shown.

Attention rank: #19 of 71 (itself plus its competitors, highest first, normalized so YC and Product Hunt are compared fairly).

Launched 366 days after the earliest competitor.

The niche, labels, vibe code rating and competitor comparison on this page are analysis by Nicheloom. If you use them, please credit nicheloom.com. How you may use our data.