tensorblock/magnum-v3-27b-kto-GGUF

Feedback and support: TensorBlock's Twitter/X, Telegram Group and Discord server

anthracite-org/magnum-v3-27b-kto - GGUF

This repo contains GGUF format model files for anthracite-org/magnum-v3-27b-kto.

The files were quantized using machines provided by TensorBlock, and they are compatible with llama.cpp as of commit b4011.

Run them on the TensorBlock client using your local machine ↗

Prompt template

<|im_start|>system
{system_prompt}<|im_end|>
<|im_start|>user
{prompt}<|im_end|>
<|im_start|>assistant

Model file specification

Filename	Quant type	File Size	Description
magnum-v3-27b-kto-Q2_K.gguf	Q2_K	9.732 GB	smallest, significant quality loss - not recommended for most purposes
magnum-v3-27b-kto-Q3_K_S.gguf	Q3_K_S	11.333 GB	very small, high quality loss
magnum-v3-27b-kto-Q3_K_M.gguf	Q3_K_M	12.503 GB	very small, high quality loss
magnum-v3-27b-kto-Q3_K_L.gguf	Q3_K_L	13.522 GB	small, substantial quality loss
magnum-v3-27b-kto-Q4_0.gguf	Q4_0	14.555 GB	legacy; small, very high quality loss - prefer using Q3_K_M
magnum-v3-27b-kto-Q4_K_S.gguf	Q4_K_S	14.658 GB	small, greater quality loss
magnum-v3-27b-kto-Q4_K_M.gguf	Q4_K_M	15.502 GB	medium, balanced quality - recommended
magnum-v3-27b-kto-Q5_0.gguf	Q5_0	17.587 GB	legacy; medium, balanced quality - prefer using Q4_K_M
magnum-v3-27b-kto-Q5_K_S.gguf	Q5_K_S	17.587 GB	large, low quality loss - recommended
magnum-v3-27b-kto-Q5_K_M.gguf	Q5_K_M	18.075 GB	large, very low quality loss - recommended
magnum-v3-27b-kto-Q6_K.gguf	Q6_K	20.809 GB	very large, extremely low quality loss
magnum-v3-27b-kto-Q8_0.gguf	Q8_0	26.950 GB	very large, extremely low quality loss - not recommended

Downloading instruction

Command line

Firstly, install Huggingface Client

pip install -U "huggingface_hub[cli]"

Then, downoad the individual model file the a local directory

huggingface-cli download tensorblock/magnum-v3-27b-kto-GGUF --include "magnum-v3-27b-kto-Q2_K.gguf" --local-dir MY_LOCAL_DIR

If you wanna download multiple model files with a pattern (e.g., *Q4_K*gguf), you can try:

huggingface-cli download tensorblock/magnum-v3-27b-kto-GGUF --local-dir MY_LOCAL_DIR --local-dir-use-symlinks False --include='*Q4_K*gguf'