tensorblock/Qwen2.5-72B-Instruct-GGUF

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

Qwen/Qwen2.5-72B-Instruct - GGUF

This repo contains GGUF format model files for Qwen/Qwen2.5-72B-Instruct.

The files were quantized using machines provided by TensorBlock, and they are compatible with llama.cpp as of commit ec7f3ac.

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
Qwen2.5-72B-Instruct-Q2_K.gguf	Q2_K	29.812 GB	smallest, significant quality loss - not recommended for most purposes
Qwen2.5-72B-Instruct-Q3_K_S.gguf	Q3_K_S	34.488 GB	very small, high quality loss
Qwen2.5-72B-Instruct-Q3_K_M.gguf	Q3_K_M	37.699 GB	very small, high quality loss
Qwen2.5-72B-Instruct-Q3_K_L.gguf	Q3_K_L	39.505 GB	small, substantial quality loss
Qwen2.5-72B-Instruct-Q4_0.gguf	Q4_0	41.232 GB	legacy; small, very high quality loss - prefer using Q3_K_M
Qwen2.5-72B-Instruct-Q4_K_S.gguf	Q4_K_S	43.889 GB	small, greater quality loss
Qwen2.5-72B-Instruct-Q4_K_M.gguf	Q4_K_M	47.416 GB	medium, balanced quality - recommended
Qwen2.5-72B-Instruct-Q5_0	Q5_0	50.164 GB	legacy; medium, balanced quality - prefer using Q4_K_M
Qwen2.5-72B-Instruct-Q5_K_S	Q5_K_S	51.375 GB	large, low quality loss - recommended
Qwen2.5-72B-Instruct-Q5_K_M	Q5_K_M	54.447 GB	large, very low quality loss - recommended
Qwen2.5-72B-Instruct-Q6_K	Q6_K	64.348 GB	very large, extremely low quality loss
Qwen2.5-72B-Instruct-Q8_0	Q8_0	77.263 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/Qwen2.5-72B-Instruct-GGUF --include "Qwen2.5-72B-Instruct-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/Qwen2.5-72B-Instruct-GGUF --local-dir MY_LOCAL_DIR --local-dir-use-symlinks False --include='*Q4_K*gguf'