Follow-up: Astra's weight edits only worked on Qwen's visible flaw, not on Gemma 270m
teortaxesTex · x · 2026-11-03
1a3orn notes the Astra weight-editing thread omitted the failures: the edits helped Qwen3.5 only because it has a preexisting visible flaw (hitting the token limit in 93% of runs); they didn't help Gemma 270m, which lacks this flaw. Targeted LLM-edits-LLM fixes depend on the model having a locatable defect — they're not a general enhancement method.
More from Models
- Leaked DeepSeek V4.1-Flash report: 552B MoE, 1M context, KV cache cut to ~1/40 — teortaxesTex · 2026-11-03
- Observers say Dessa eerily matches Karpathy's "summoning ghosts" analogy; Jack calls it fast — analisereal · 2026-11-03
- Together at TC Disrupt: 'token maxing without cost maxing' via smaller models — togethercompute · 2026-11-03
- Dream-1 zero-shot robot test: 100 public episodes show real skills but poor instruction following — paigeinsf · 2026-11-03
- Speculation: DeepSeek to ship V4.1F or V4.2F before any Pro variant — teortaxesTex · 2026-11-03
- LLM edits another LLM: tweaking 0.016% of weights lifts Qwen3.5-0.8B math from 7.9% to 48.1% — teortaxesTex · 2026-11-03