Converted version of CodeLlama-34b to 4-bit using bitsandbytes. For more information about the model, refer to the model's page.

Impact on performance

In the following figure, we can see the impact on the performance of a set of models relative to the required RAM space. It is noticeable that the quantized models have equivalent performance while providing a significant gain in RAM usage.

Downloads last month: -

Safetensors

Model size

18B params

Tensor type

F32

BF16

Collection including cmarkea/CodeLlama-34b-hf-4bit

Quantized 4-bit models

Collection

Large model quantized with post-quantization performance very close to the original models, allowing it to run on reasonable infrastructure. • 10 items • Updated Jan 17 • 1