Llamacpp imatrix Quantizations of phi 4 Using llama.cpp release b4456 for quantization. Original model: https://huggingface.co/microsoft/phi 4 All quants made using imatrix option with dataset from here Run them in LM Studio Prompt format What's new: Tokenizer fixes as recommended by the unsloth team Download a file (not the whole branch) from below: Filename Quant type File Size Split Description phi 4 f32.gguf f32 58.64GB true Full F32 weights. phi 4 f16.gguf f16 29.32GB false Full F16 weights. phi 4 Q8 0.gguf Q8 0 15.58GB false Extremely high quality, generally unneeded but max available quant. phi 4 Q6 K L.gguf Q6 K L 12.28GB false Uses Q8 0 for embed and output weights. Very high quality, near perfect, recommended . phi 4 Q6 K.gguf Q6 K 12.03GB false Very high quality, near perfect, recommended . phi 4 Q5 K L.gguf Q5 K L 10.92GB false Uses Q8 0 for embed and output weights. High quality, recommended . phi 4 Q5 K M.gguf Q5 K M 10.60GB false High quality, recommended . phi 4 Q5 K S.gguf Q5 K S 10.15GB false High quality, recommended . phi 4 Q4 K L.gguf Q4 K L 9.43GB false Uses Q8 0 for embed and output weights. Good quality, recommended . phi 4 Q4 1.gguf Q4 1 9.27GB false Legacy…
We use cookies for essential functionality and analytics. You can accept or reject analytics cookies.Cookie policy