GGUF importance matrix (imatrix) quants for https://huggingface.co/CausalLM/34b-beta
The importance matrix was trained for 40K tokens (80 batches of 512 tokens) using random data.

From the model author:
Please do not use wikitext for quantization calibration because all wikitext have been re-aligned on synthetic dataset, and its distribution differs significantly from the original wikitext.

Layers Context Template
60
200000
<|im_start|>system
{instructions}<|im_end|>
<|im_start|>user
{prompt}<|im_end|>
<|im_start|>assistant
{response}
Downloads last month
239
GGUF
Model size
34.4B params
Architecture
llama

2-bit

3-bit

4-bit

Inference Examples
Inference API (serverless) does not yet support gguf models for this pipeline type.