|
--- |
|
language: |
|
- en |
|
- de |
|
- fr |
|
- it |
|
- pt |
|
- hi |
|
- es |
|
- th |
|
license: llama3.1 |
|
pipeline_tag: text-generation |
|
tags: |
|
- facebook |
|
- meta |
|
- pytorch |
|
- llama |
|
- llama-3 |
|
- gguf |
|
- imatrix |
|
base_model: meta-llama/Meta-Llama-3.1-70B-Instruct |
|
--- |
|
|
|
# Quant Infos |
|
|
|
- ~Requires latest master + [Rope Scaling PR](https://github.com/ggerganov/llama.cpp/pull/8676).~ Rope scaling is merged, so just a a recent master is required now. |
|
- quants done with an importance matrix for improved quantization loss |
|
- Quantized ggufs & imatrix from hf bf16, through bf16. `safetensors bf16 -> gguf bf16 -> quant` for *optimal* quant loss. |
|
- Wide coverage of different gguf quant types from Q\_8\_0 down to IQ1\_S |
|
- experimental custom quant types |
|
- `_L` with `--output-tensor-type f16 --token-embedding-type f16`, which supposedly leads to better accuracy. |
|
- Imatrix generated with [this](https://gist.github.com/bartowski1182/eb213dccb3571f863da82e99418f81e8) multi-purpose dataset by [bartowski](https://huggingface.co/bartowski). |
|
``` |
|
./imatrix -m $model_name-bf16.gguf -f calibration_datav3.txt -o $model_name.imatrix |
|
``` |
|
|
|
# Original Model Card: |
|
|
|
TODO |
