DARTree: Speculative Diffusion Decoding with Autoregressive Draft Trees
DARTree proposes diffusion-based draft trees that predict a whole token block per step instead of one token at a time.
AI Storyline
3 items · 3 sources · 3 days
Operational story trace
Speculative Decoding
Follow in this browser to see new updates on your Live feed.
Latest change
A new paper (AsymSpec, Aug 26) targets speculative decoding specifically at agentic pipelines, arguing that generic techniques don't address the latency cost of long, tool-call-and-retrieval-inflated context that agents accumulate.
Earlier contextThe story so far
Speculative decoding — verifying multiple draft tokens per step to cut inference latency — is now developing on two separate fronts: new drafting research and production serving guides. A mid-August paper (DARTree) pushed the research side toward diffusion-based draft trees that predict whole token blocks instead of single tokens.
editor-curated · source-linked
Arc
Aug 13Aug 26 · now
DRAFTING RESEARCH · Aug 13DARTree proposes diffusion-based draft trees to cut proposal latency further1 source · show source ▾
PRODUCTION TUNING · Aug 23vLLM publishes a hands-on speculative-decoding tuning guide for AMD GPUs1 source · show source ▾
Exploring Speculative Decoding in vLLM on AMD GPUs
vLLM publishes a practical speculative-decoding tuning guide for AMD GPUs covering MTP, EAGLE-3, DFlash, and DSpark.
AGENTIC SPECIALIZATION · Aug 26AsymSpec argues agentic LLM pipelines need context-asymmetric speculative decodingLong accumulated context from retrieval, tool use, and multi-turn interaction inflates inference cost in ways generic speculative decoding and context-compression tricks don't fully address, the paper argues.1 source · show source ▾
AsymSpec: Context-Asymmetric Speculative Decoding for Agentic LLMs
AsymSpec proposes context-asymmetric speculative decoding aimed specifically at agentic LLM pipelines' inflated context cost.
Day 1 Thursday, Aug 13, 2026
DARTree: Speculative Diffusion Decoding with Autoregressive Draft Trees
DARTree proposes diffusion-based draft trees that predict a whole token block per step instead of one token at a time.
Day 2 Sunday, Aug 23, 2026
Exploring Speculative Decoding in vLLM on AMD GPUs
vLLM publishes a practical speculative-decoding tuning guide for AMD GPUs covering MTP, EAGLE-3, DFlash, and DSpark.
Day 3 Wednesday, Aug 26, 2026
AsymSpec: Context-Asymmetric Speculative Decoding for Agentic LLMs
AsymSpec proposes context-asymmetric speculative decoding aimed specifically at agentic LLM pipelines' inflated context cost.
What to watch — open questions
- Does AsymSpec's context-asymmetric approach get benchmarked against production serving stacks (vLLM/EAGLE-3/DFlash) rather than only academic baselines?
- Is there a reference implementation of AsymSpec that agent-pipeline builders can actually try, or is it paper-only for now?
- How much of DARTree's diffusion-drafting latency win holds up outside its own benchmark suite?
How this thread was built
editor wrote the arc · 3 beats
Storylines are threaded mechanically from the feed: stories that share a distinctive anchor across multiple days and sources. Each item links to its original source. The evidence trace, current state, and open questions are written by the editor routine and refreshed whenever a new beat lands.