|
|
--- |
|
|
language: |
|
|
- en |
|
|
pipeline_tag: text-classification |
|
|
tags: |
|
|
- GenAudit |
|
|
- Factuality |
|
|
- Consistency |
|
|
--- |
|
|
## Model Details |
|
|
The 7B Mistral Model from GenAudit served in Q4_K_S and F16 GGUF format. |
|
|
<br>Merged and Quantised with Unsloth.AI |
|
|
|
|
|
### Model Description |
|
|
|
|
|
Inspired by this paper: https://genaudit.org/ |
|
|
<br>Original Code here: https://github.com/kukrishna/genaudit |
|
|
|
|
|
Converted to GGUF format for running it on Ollama/Llama.cpp so as to take advantage of VRAM offloading to RAM (something Huggingface transformers is unable to do for now). |
|
|
|
|
|
Merged base mistral_v0.1_instruct with Qlora and quantised to Q4_k_s gguf format |
|
|
<br>You may find the base 16 bit model here (but further quantisation is advisable as their Qlora module was fine tuned on the 4bit nf4 base llm) |
|
|
|
|
|
**Developed by:** [Nuode Chen](https://github.com/ChenNuode) |
|
|
<br>**Finetuned from model:** [Mistral_V0.1_instruct](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.1) |
|
|
<br> |
|
|
|
|
|
### Model Sources |
|
|
|
|
|
<!-- Provide the basic links for the model. --> |
|
|
- https://github.com/kukrishna/genaudit |
|
|
|
|
|
|
|
|
## Uses |
|
|
For evaluating the abstractive summaries of LLM given a source article. |
|
|
<br>This tool will be able to extract evidences supporting each sentence in the summary as well as provide edits to correct its factuality (if applicable) |
|
|
|
|
|
Refer to original paper for more in-depth information. |
|
|
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. --> |
|
|
|
|
|
|
