proomt

Search

Search posts, papers, and topics

All posts

Hugging Face Daily PapersYecheng Wu, Song Han, Han Cai1 min readpaperadvanced

Lightning Weave: Improving the Accuracy-Efficiency Frontier of Reasoning Models through Capability Composition

Read on arxiv.org

Summary

Lightning Weave is a post‑training framework that composes specialist capabilities from separate models into one student via on‑policy distillation, improving both accuracy and token efficiency on math and code benchmarks.

  • Lightning Weave extracts specialist capabilities from independently post‑trained models and merges them into a single student via on‑policy distillation.
  • Each capability is encoded as a log‑ratio policy shift; aligned shifts at shared token states are turned into stable targets using Tilted‑Target DOPD.
  • Anchor trajectories are cached and scored once, eliminating the need to run multiple live specialist models during student training.
  • On Qwen3.5‑4B, HMMT 2025 accuracy rises from 59.2% to 64.0% with 10.7% fewer response tokens; LiveCodeBench v5 accuracy jumps to 54.2% from 41.7% with 9.6% token reduction.

LLM engineers looking to boost reasoning performance while cutting inference cost can adopt Lightning Weave to get higher accuracy with fewer tokens.

7/10
llmmachine learningreasoningdistillationefficiency

Related reading

  1. Grounded Skill Synthesis from Code at Scale for Agentic Intelligence

    Code2Skill is an automated pipeline that mines popular GitHub repositories to extract verifiable, implementation‑anchored procedural “skills”. It builds a bank of ~1 M skill records (atomic ops, workflows, patterns) with provenance metadata, verifies each via blind reconstruction, and shows that augmenting LLM‑based agents with these skills yields an average 11.7% performance lift across 72 proto…

    Hugging Face Daily Papersarxiv.org1 minpaper

  2. Verifiable Social Reasoning for LLM Assistants

    The paper introduces Fuse, a multi‑agent simulation that gives LLM assistants a verifiable ground‑truth task for social reasoning by hiding a target agent’s motive and letting a user‑mediated conversation infer it. Experiments on 12 LLMs show user mediation makes reasoning harder, models are biased by user framing, need more detail than humans, and longer chats don’t always help.

    Hugging Face Daily Papersarxiv.org1 minpaper

  3. Lambda MicroEgg

    lambda-microegg is a Rust e‑graph library that adds scoped, alpha‑aware binders and Miller‑style higher‑order patterns to the Egg framework, enabling capture‑avoiding substitution and higher‑order rewriting with modest performance overhead.

    Lobstersphilipzucker.com15 minlobste.rs27

  5. Training Specialist Models without Reasoning Trajectories for Domain Expert Distillation

    The paper shows that when specialist LLMs are trained only on QA pairs (no explicit reasoning supervision), their optimization implicitly selects a latent distribution of reasoning trajectories. By treating the distilled student as an agnostic probe—since it inherits only the sampled trajectories—the authors empirically demonstrate a strong correlation (across 27 specialist‑student pairs) between…

    Hugging Face Daily Papersarxiv.org1 minpaper

  6. Article: Your Next DSL Author Is a Language Model

    Typed Domain Grounding (TDG) embeds a DSL inside a mainstream language the LLM already knows (e.g., Kotlin) and uses the host compiler as an oracle. The author describes five building blocks—embedding, choosing a host language with high training‑data frequency, compiler‑driven type safety, a generate‑compile‑repair loop, and an on‑demand teaching tool—and shows measured results from kUML, a Kotli…

    InfoQinfoq.com18 min