timm DINOv3 Collection Meta AI's DINOv3 weights in timm. ViTs with `qkvb` have a zero QV bias present, otherwise bias is disabled. QKV bias are all 0 in original weights. • 18 items • Updated Sep 19, 2025 • 35
DINOv3 Collection DINOv3: foundation models producing excellent dense features, outperforming SotA w/o fine-tuning - https://arxiv.org/abs/2508.10104 • 15 items • Updated Mar 10 • 623
talkie-13b Collection talkie-1930-13b is a vintage language model trained on pre-1931 English-language text. See https://github.com/talkie-lm/talkie to run talkie. • 3 items • Updated 15 days ago • 45
MedGemma Release Collection Collection of Gemma 3 variants for performance on medical text and image comprehension to accelerate building healthcare-based AI applications. • 9 items • Updated Mar 12 • 486
NVIDIA Nemotron v3 Collection Open, Production-ready Enterprise Models • 18 items • Updated 8 days ago • 284
Phi-4 Collection Phi-4 family of small language, multi-modal and reasoning models. • 17 items • Updated Jul 10, 2025 • 207
Unsloth Dynamic 2.0 Quants Collection New 2.0 version of our Dynamic GGUF + Quants. Dynamic 2.0 achieves superior accuracy & SOTA quantization performance. • 89 items • Updated 8 days ago • 584
view article Article GGML and llama.cpp join HF to ensure the long-term progress of Local AI +4 Feb 20 • 505
Does Your Reasoning Model Implicitly Know When to Stop Thinking? Paper • 2602.08354 • Published Feb 9 • 264
LTX-2 Collection LTX-2 base models and accompanying LoRAs and IC-LoRAs • 13 items • Updated Jan 29 • 62
Entropy-Adaptive Fine-Tuning: Resolving Confident Conflicts to Mitigate Forgetting Paper • 2601.02151 • Published Jan 5 • 113