LiquidAI extends lossless speculative decoding to vision-language models
JosephJacks_ · x · 2026-09-25
LiquidAI shows how their speculative-decoding approach for text LFMs extends naturally to vision-language models.
Key insight: by the time information reaches the language backbone's hidden layers, text tokens and image patches are both tensors, so the original modality no longer matters to the drafter. A lightweight drafter uses LFM2.5-VL-3B's hidden states from different layers to speculate several tokens ahead, verified by the target model in a single forward pass.
Under matched sampling settings the decoding is distribution-equivalent to direct sampling, making the output lossless.
More from Models
- Relace Is Now the Cheapest DeepSeek v4.1 Flash Provider on OpenRouter — ilyasu · 2026-09-25
- Jev matches year-old top models on global geographic understanding, maps extracted — zetalyrae · 2026-09-25
- Claude Opus 5.5 tops SimpleBench with 88.4% score — Profanion · 2026-09-25
- Anthropic resumes billing for safety-blocked requests; 99.7% of users unaffected — ClaudeDevs · 2026-09-25
- Developer claims: nothing holds back Claude models like Claude Code itself — tokenbender · 2026-09-25
- Blogger: Opus 5.5's strength suggests xAI's rumored Astra is smaller than believed — scaling01 · 2026-09-25