ReaDy: Recurrent States as Dynamic Context for Speculative Drafting in Hybrid Linear Attention Large Language Models
ReaDy uses recurrent states for speculative drafting, reaching 3.16× speedup on Qwen3.8-27B.
36
Product
NVIDIA B200
- Mentions · 7d0
- Mentions · 30d3
- All mentions8
- First seenSep 17, 2026
- Last mentioned25d ago
Timeline
8 articles · newest firstReaDy uses HLA recurrent states to speed speculative decoding
On Qwen3.8-27B and one B200, ReaDy averages 3.16 times autoregressive speed and beats DSpark on every benchmark.
44
Joint Subspace Sparse Attention for Efficient Long-Context Decoding
JSSA sparse attention stays near full-KV accuracy and raises 128K decoding throughput up to 6.43×.
54
autotrust/GLM5.3-Flash-E224-DGX-Spark — new model trending #29 on Hugging Face
autotrust released an unofficial pruned NVFP4 GLM-5.3-Flash build that fits two NVIDIA DGX Sparks.
44
Aleph Alpha releases Kolibri, an open-weight model that makes the case for European AI sovereignty
Aleph Alpha released Kolibri, a 78-billion-parameter open-weight German-English mixture-of-experts model.
64
Aleph Alpha Releases Kolibri: A 78.1B Open-Weight English-German MoE Model With Only 3.46B Active Parameters
Aleph Alpha released Kolibri, a 78.1B open-weight English-German MoE with 3.46B active parameters.
66
LeapQuant: Efficient Linear Attention with Accurate Recurrent State Quantization
LeapQuant nearly matches FP32 quality with training-free 8-bit recurrent-state quantization.
44
Video DeltaNet: A Video-Native Hybrid Attention for Livestream Video Generation
Video DeltaNet combines softmax and linear attention for video generation, cutting DiT denoising of a 14.3-second 768p video to 6.7 seconds.
38
Appears with
- NVIDIAvendor · 8
- Kolibriai model · 2
- Aleph Alphaorganization · 2
- StartLuxLabsorganization · 2
- Qwen3.8-27Bai model · 2
- Red Hatorganization · 1
- SGLangproduct · 1
- GB200product · 1
- RTX 5090product · 1
- MiniMaxvendor · 1
- Kimiai model · 1
- Kolibriproduct · 1
- GLM-5.3-Flashai model · 1
- GLM5.3-Flash-E224-DGX-Sparkai model · 1
- GPT OSS A5Bai model · 1
- NVIDIA H200product · 1
Entities are extracted by the model from each article. First seen , last mentioned . Watching keeps it in this browser only (no account); the watchlist page and dashboard alerts use it.