Hugging Face's logo
Djuunaa's picture
156 15 400

Djuunaa

djuna
Johnnyman1100's profile picture WesPro's profile picture spyrosgous's profile picture
·

AI & ML interests

None yet

Recent Activity

liked a model 1 day ago
fishaudio/openaudio-s1-mini
reacted to eaddario's post with 🚀 1 day ago
Layer-wise and Pruned versions of google/gemma-3-12b-it After enhancing llama.cpp to handle user-defined quantization levels for arbitrary tensors (https://github.com/ggml-org/llama.cpp/pull/12511), I have added an option to prune whole layers (https://github.com/ggml-org/llama.cpp/pull/13037), and have published two versions of https://huggingface.co/google/gemma-3-12b-it for demo and testing purposes: * Tesor-wise: https://huggingface.co/eaddario/gemma-3-12b-it-GGUF * Pruned: https://huggingface.co/eaddario/gemma-3-12b-it-pruned-GGUF Even though the Perplexity scores of the pruned version are 3 times higher, the ARC, HellaSwag, MMLU, Truthful QA and WinoGrande scores are holding remarkably well, considering two layers were removed (26 and 29). This seems to support Xin Men et al conclusions in https://huggingface.co/papers/2403.03853 Results summary in the model's card and test results in the ./scores directory. Questions/feedback is always welcomed.
new activity 1 day ago
deepseek-ai/DeepSeek-R1-0528-Qwen3-8B:DeepSeek-R1-Lite
View all activity

Organizations

Dev Mode Explorers's profile picture Djuna Test Lab's profile picture

Collections 1

datasets 0

None public yet