proomt

Search

Search posts, papers, and topics

All posts

Hugging Face Daily PapersAlbert Ge, Chandan Singh, Yufan Zhuang1 min readpaperadvanced

Register Tokens for Bounded-State Reasoning in Diffusion Language Models

Read on arxiv.org

Summary

Register tokens are fixed‑position embeddings that store a compact hidden state across diffusion‑based language model generation chunks, enabling bounded‑state reasoning without retaining all prior text. Post‑training on LLaDA and Dream shows up to +8.5 math and +19.5 code benchmark points versus plain text carry, and RL fine‑tuning further improves long‑horizon tasks.

  • Diffusion LLMs normally need the full generated context to continue reasoning; registers replace that with a small learned state.
  • Registers are trained by decoding a chunk, clearing the text, and forcing the model to preserve the hidden values for the next chunk.
  • Empirical gains on standard reasoning benchmarks (math, code) are substantial compared to discrete‑text carry.
  • The method works especially well for bounded code generation where programs span multiple chunks.

Current diffusion language models struggle with multi‑step reasoning because they must keep all prior tokens in context, which limits scalability and increases compute. A compact, trainable state that persists across generation steps could make reasoning more efficient and enable longer, more compl…

8/10
llmreinforcement learningdiffusionreasoninglanguage models

Related reading

  1. When2Think: Learning Difficulty-Aware Length Control for Efficient Hybrid Reasoning Models

    When2Think introduces a post‑training framework that lets a large reasoning model decide per‑instance how much reasoning depth to allocate, using difficulty‑aware reward shaping (IDAC) and verifier rewards. It cuts token usage by ~28% while boosting Pass@3 by 10% on AIME24 and reaches 40% Pass@3 on AIME25, outperforming compression and routing baselines.

    Hugging Face Daily Papersarxiv.org1 minpaper

  2. DACA-GRPO: Denoising-Aware Credit Assignment for Reinforcement Learning in Diffusion Language Models

    DACA‑GRPO adds denoising‑aware credit assignment to GRPO‑style RL trainers for diffusion LLMs. It computes per‑token importance scores from intermediate denoising steps and uses stratified masking to reduce mean‑field bias in likelihood estimates. Plug‑and‑play on three existing GRPO methods, it yields consistent gains on seven downstream tasks (up to +5.6 pp math, +7.4 pp code, +36.3 pp constrai…

    Apple Machine Learning Researchapple.com1 minpaper

  3. tokenizers v1: encode, decode and scaling, measured

    Hugging Face has released `tokenizers` v1, a major performance update that achieves 3-30x faster encoding than v0.23 while maintaining identical output and API compatibility. Key optimizations include a SIMD-accelerated splitter, a thread-local word cache, and an allocation-free BPE merge loop, ensuring tokenization doesn't bottleneck ML workflows.

    Hugging Facehuggingface.co10 min

  4. HarnessVLN: Unifying Training-Free Embodied Navigation through an Agent Harness

    HarnessVLN introduces a zero‑shot, training‑free embodied navigation framework that wraps a multimodal LLM in an "Agent Harness" – a tool‑based protocol that validates planner actions against spatial evidence, tracks progress with hierarchical event memory, and maintains a persistent spatiotemporal graph for recovery. The system works for instruction‑following and object‑goal tasks, achieving 60.…

    Hugging Face Daily Papersarxiv.org1 minpaper

  5. MoME: Mixture-of-Memory Embeddings for Context-Aware Sparse Lookup

    The paper proposes Mixture of Memory Embeddings (MoME), a context‑aware sparse lookup that replaces each token’s single memory row with a gated mixture of multiple slots. Experiments on Llama‑3, MobileLLM and Qwen3 show MoME outperforms existing memory‑embedding baselines at equal parameter and FLOP budgets and exhibits interpretable routing for polysemous tokens.

    Hugging Face Daily Papersarxiv.org1 minpaper