ModelxLama

xLama v3

A series of large language models that support mixed quantization, employing a strategy of different bit rates for different layers, which enables a 14B model to be deployed on 8GB of GPU memory.

Timeline (last 90 days)

Oct 30

xLama v3 uses mixed quantization with different bit rates for different layers, which can fit a 14B model into 8GB VRAM, and in实测 VRAM is almost fully used.

Unverified50%

All Facts (1)

Unverified

xLama v3 uses mixed quantization with different bit rates for different layers, which can fit a 14B model into 8GB VRAM, and in实测 VRAM is almost fully used.

50%

Source Articles

The Complete Guide to Running AI Models Locally: Inference Engines and Tool Stack ExplainedOct 30