GHOST 2.0: generative high-fidelity one shot transfer of heads Paper • 2502.18417 • Published 4 days ago • 58
Unsloth 4-bit Dynamic Quants Collection Unsloths Dynamic 4bit Quants selectively skips quantizing certain parameters; greatly improving accuracy while only using <10% more VRAM than BnB 4bit • 22 items • Updated 3 days ago • 53
Scaling up Test-Time Compute with Latent Reasoning: A Recurrent Depth Approach Paper • 2502.05171 • Published 22 days ago • 120
view post Post 2554 Ovis2 🔥 a multimodal LLM released by Alibaba AIDC team. AIDC-AI/ovis2-67ab36c7e497429034874464✨1B/2B/4B/8B/16B/34B✨Strong CoT for deeper problem solving✨Multilingual OCR – Expanded beyond English & Chinese, with better data extraction See translation 🚀 3 3 🔥 2 2 ➕ 1 1 + Reply
Phi-4 (All Versions) Collection Microsoft's new Phi-4 models including mini in all formats. Includes GGUF, 4-bit bnb and original versions. Includes Unsloth's bug fixes. • 8 items • Updated about 23 hours ago • 43