29 lines
729 B
Markdown
29 lines
729 B
Markdown
|
---
|
||
|
|
tags:
|
||
|
|
- quantized
|
||
|
|
- 2-bit
|
||
|
|
- 3-bit
|
||
|
|
- 4-bit
|
||
|
|
- 5-bit
|
||
|
|
- 6-bit
|
||
|
|
- 8-bit
|
||
|
|
- GGUF
|
||
|
|
- phi
|
||
|
|
- phi-3
|
||
|
|
- text-generation
|
||
|
|
model_name: Phi-3-mini-mango-1-GGUF
|
||
|
|
base_model: rhysjones/Phi-3-mini-mango-1
|
||
|
|
inference: false
|
||
|
|
model_creator: rhysjones
|
||
|
|
pipeline_tag: text-generation
|
||
|
|
quantized_by: rhysjones
|
||
|
|
license: mit
|
||
|
|
---
|
||
|
|
|
||
|
|
## Description
|
||
|
|
These are GGUF model format files for the [rhysjones/Phi-3-mini-mango-1](https://huggingface.co/rhysjones/Phi-3-mini-mango-1) Phi-3 4k model.
|
||
|
|
|
||
|
|
## Conversion process
|
||
|
|
|
||
|
|
The useful conversion script [GGUF-n-Go](https://github.com/thesven/GGUF-n-Go) by [thesven](https://github.com/thesven) was used along with
|
||
|
|
[llama.cpp](https://github.com/ggerganov/llama.cpp) to generate the different quantized sizes for the model.
|