view post Post 1,069 Nanbeige4.3-33B DFlash2 speculative drafter ⚡️1.76× faster decoding on BF16, 1.65× on FP8, 1.51× on NVFP4. Agentic coding ~2.6×. Same outputs: exact rejection sampling. This is an early result, an improved version with higher acceptance is in training. SGLang 0.5.21, drop-in: ProCreations/Nanbeige4.3-33B-DFlash2 See translation
Grug Series A collection of all Grug models I have made. ProCreations/grug-27b Text Generation • 27B • Updated Jul 24 • 2.49k • • 101 ProCreations/grug-35b-v2 Text Generation • 35B • Updated Jul 23 • 134 • 20 ProCreations/grug-v2-9b-gguf Text Generation • 9B • Updated Jul 15 • 196 • 8 ProCreations/grug-3b Text Generation • 4B • Updated Jul 26 • 189 • 31
Auto Series A series of models designed to protect against rouge agents. ProCreations/auto-0.4b-2 Text Classification • 0.4B • Updated Sep 11 • 106 • 1 ProCreations/auto-200m-2 Text Classification • 0.1B • Updated Sep 26 • 155 • 4 ProCreations/auto-3b Text Classification • 3B • Updated Sep 26 • 39 • 1 ProCreations/auto-1b Text Classification • 1B • Updated Aug 9 • 23 • 16
Grug Series A collection of all Grug models I have made. ProCreations/grug-27b Text Generation • 27B • Updated Jul 24 • 2.49k • • 101 ProCreations/grug-35b-v2 Text Generation • 35B • Updated Jul 23 • 134 • 20 ProCreations/grug-v2-9b-gguf Text Generation • 9B • Updated Jul 15 • 196 • 8 ProCreations/grug-3b Text Generation • 4B • Updated Jul 26 • 189 • 31
Auto Series A series of models designed to protect against rouge agents. ProCreations/auto-0.4b-2 Text Classification • 0.4B • Updated Sep 11 • 106 • 1 ProCreations/auto-200m-2 Text Classification • 0.1B • Updated Sep 26 • 155 • 4 ProCreations/auto-3b Text Classification • 3B • Updated Sep 26 • 39 • 1 ProCreations/auto-1b Text Classification • 1B • Updated Aug 9 • 23 • 16
Running on Zero MCP 1 Stop-Motion Consistency LoRA (Qwen-Image 2.1) 🎞 Next-frame stop-motion generation with Qwen-Image 2.1 + LoRA