6arXiv cs.AI (Artificial Intelligence)·
Discrete-diffusion language model adapted for parallel speech recognition via frozen DiffusionGemma
Researchers demonstrate that a frozen 26B mixture-of-experts discrete diffusion language model (DiffusionGemma) can be adapted for automatic speech recognition using only ~42M trainable parameters (0.16% of backbone). The system uses a frozen Whisper encoder for acoustic features, lightweight projectors, and LoRA adapters, achieving 6.6% WER on LibriSpeech test-clean in roughly eight parallel denoising steps regardless of utterance length. A key finding is that standard training objectives fail to ground audio features due to attention dismissal, and a CTC loss through the frozen output head resolves this. The approach supports multilingual transcription (English, Hindi, Mandarin) from a single adapter.
Related guides (3)
Related events (8)
DeepMind announces DiffusionGemma with 4x faster text generation
DeepMind published a blog post introducing DiffusionGemma, a diffusion-based variant of the Gemma model family claiming 4x faster text generation. The announcement suggests a departure from standard autoregressive decoding in favor of diffusion-based generation. If the claims hold, this could represent a meaningful inference efficiency advance for the Gemma line.
ADAS: Attention-Discounted Adaptive Sampler improves parallel decoding for masked diffusion language models
Researchers propose ADAS, a training-free reranking rule for masked diffusion language model decoding that addresses token interaction failures in parallel token commitment. The method greedily penalizes candidates that attend strongly to already-selected uncertain positions, using attention weights as soft marginal penalties rather than hard constraints. Evaluated on LLaDA-8B-Base and Dream-7B-Base across GSM8K, MATH500, HumanEval, and MBPP, ADAS improves low-NFE performance by 9–10 percentage points on average when plugged into existing samplers with only 3.1% runtime overhead.
Simon Willison on DiffusionGemma
Simon Willison covers DiffusionGemma, a diffusion-based language model in the Gemma family from Google. The post appears to be commentary or a brief note on the model's release or capabilities. Diffusion-based LLMs represent an active area of research as an alternative to autoregressive generation.
LESS: Adaptive mutual-stability sampling cuts diffusion LLM decoding steps by 72%
Researchers introduce LESS, a training-free adaptive sampler for diffusion large language models that treats token commitment as an online stopping problem. The method uses a joint stability rule combining confidence, persistence, and distributional stability to decide when to unmask tokens, avoiding wasted computation on already-stable positions. Evaluated on Dream-7B, LLaDA-8B, and LLaDA-1.5-8B across seven benchmarks, LESS reduces reverse denoising steps by 72.1% versus fixed-budget decoding while improving accuracy over prior adaptive samplers. The step reductions translate directly to fewer Transformer forward passes and lower wall-clock latency.
SARDI: Self-Augmenting Retrieval for Diffusion Language Models using lookahead tokens
Researchers introduce SARDI, a training-free RAG framework for discrete diffusion language models that repurposes discarded low-confidence tokens during denoising as lookahead signals to guide retrieval before output is finalized. The method is retriever-agnostic and applicable to any reasoning-capable discrete diffusion LM. Evaluated across five multi-hop QA benchmarks, SARDI outperforms training-free diffusion and autoregressive retrieval baselines at up to 8x higher throughput.
SimSD: Speculative Decoding Adapted for Diffusion Language Models
SimSD introduces a training-free speculative decoding algorithm for diffusion large language models (dLLMs), which previously could not use standard token-level speculative decoding due to their bidirectional attention and masked language modeling formulation. The method uses a plug-and-play masking strategy that introduces reference tokens from a draft model and a custom attention mask, enabling valid logit computation for drafted tokens in a single forward pass. Evaluated on SDAR-family dLLMs across four benchmarks, SimSD achieves up to 7.46x decoding throughput improvement while maintaining or improving generation quality. The approach is compatible with other acceleration techniques such as KV cache and blockwise decoding.
Corpus-Grounded Feature Diffusion pipeline for automated IEP generation in Traditional Chinese
Researchers propose a low-resource fine-tuning pipeline called Corpus-Grounded Feature Diffusion (CGFD) to automate Individualized Education Program (IEP) drafting from Traditional Chinese parent-teacher interview transcripts. The approach fine-tunes Breeze-7B with QLoRA on 582 synthetically diffused samples and uses schema-constrained decoding at inference time, finding that Grammar-Constrained Decoding is counterproductive under Traditional Chinese token budgets. On a small formal hold-out (n=10), the system achieves BERTScore F1 of 0.779, outperforming zero-shot GPT-5.4, DeepSeek-V3.2, Gemini-3-Flash-Preview, and Llama-4-Maverick baselines while enabling fully local, air-gapped inference. The work addresses a gap in Traditional Chinese special-education NLP and demonstrates a privacy-preserving deployment pattern for sensitive document generation.
DirectAudioEdit: Training-free, inversion-free text-guided audio editing via diffusion prediction contrast
Researchers introduce DirectAudioEdit, the first training-free and inversion-free method for text-guided audio editing using diffusion denoising dynamics. The approach constructs a source-to-target editing path without requiring DDPM inversion, reducing macro-averaged FAD and KL divergence by ~16% compared to inversion-based baselines while achieving up to 64.5% speedup. Experiments span music and event-level benchmarks across two backbone architectures.