deepseek-coder-6.7B-instruct-GPTQ

This is a GPTQ-quantized variant of the official DeepSeek-Coder 6.7B instruct model, hosted on Hugging Face by TheBloke. It provides a practical, memory-efficient way to run the instruct-tuned model on consumer hardware with compatible inference engines.

Open repository
Updated
2026-06-06

Summary

This is a GPTQ-quantized variant of the official DeepSeek-Coder 6.7B instruct model, hosted on Hugging Face by TheBloke. It provides a practical, memory-efficient way to run the instruct-tuned model on consumer hardware with compatible inference engines.

The candidate's canonical URL points to a model card on huggingface.co, a standard platform for hosting model weights.. The source evidence shows it is listed in the 'awesome-deepseek-coder' README under 'Quantized Models', a curated community resource.. The tags 'gptq' and 'deepseek-coder' indicate it is a quantized variant of the specific base model.

Tags

Also appears in