File size: 570 Bytes
1e20a7a
 
738c7ff
 
 
 
d0cdb2b
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
---
license: apache-2.0
tags:
- finetuned
- Instruct
- code
---

## KAI-7B Instruct

KAI-7B-Instruct is a 7 Billion parameter causal model based on KAI-7B and Mistral-7B. KAI-7B has been finetuned on a mixture of chat/instruct datasets.

## Why use KAI-7B Instruct?

- KAI-7B is a strong base model, outperforming comparable open-source models (e.g., LLaMA 2 70b, Mistral 7B, e.t.c.)
- KAI-7B Instruct is trained on a framework for improving the instruction-following capabilities of pretrained language models.
- KAI-7B Instruct is a ready-to-use model based off KAI-7B