3 ms·
The 4-bit quantized version of LLaMA 7B used by llama.cpp is a 4GB file. The 13B model is under 8GB.
by simonw 4y ago
The 4-bit quantized version of LLaMA 7B used by llama.cpp is a 4GB file. The 13B model is under 8GB.
- sroussey 4y agoIf only there was a download link…