File size: 2,366 Bytes
e94cfec
 
ae8e968
 
 
 
 
 
 
e94cfec
a3db771
 
 
 
 
 
 
 
 
 
 
 
 
a1f36bd
a3db771
 
 
 
 
e5a4da4
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
---
license: apache-2.0
datasets:
- Keynote-Technology/PLANE-2K
- togethercomputer/RedPajama-Data-V2
tags:
- code
- chatbot
- safetensors
---
![image/png](https://cdn-uploads.huggingface.co/production/uploads/6500c7c912c1442d994c36e5/rav-pUPtWF-u-_A4k9RaZ.png)

TinyKAI 3B is a fine-tuned LLM (Large Language Model) based off of [OpenLlama 3B v2](https://huggingface.co/openlm-research/open_llama_3b_v2).
The TinyKAI models are a series of lightweight LLMs under 5 Billion parameters, usually used for research.

## Direct Use
TinyKAI 3B is optimal for research on large language models, specifically the influence of web data on the properties of large language models (fairness, safety, limitations, capabilities, etc.).

## Training
This model was trained on a mixture of the [Falcon refined-web dataset](https://huggingface.co/datasets/tiiuae/falcon-refinedweb), the [StarCoder dataset](https://huggingface.co/datasets/bigcode/starcoderdata) and the wikipedia, arxiv, book and stackexchange part of the [RedPajama dataset](https://huggingface.co/datasets/togethercomputer/RedPajama-Data-1T).

## Banned Use
Production use without adequate assessment of risks and mitigation; any use cases which may be considered irresponsible or harmful or insulting to anyone or any certain group.
TinyKAI-3B is governed by the [apache 2.0 liscense](https://choosealicense.com/licenses/apache-2.0/), and therefore means that whatever the license deems unacceptable shall not be allowed. We specificaly ban the use of  [ANY AND ALL KAI MODELS](https://huggingface.co/collections/Keynote-Technology/kai-large-language-models) for hate speech towards a paricular thing, person, our particular group due to [legal](https://www.ftc.gov/news-events/news/press-releases/2022/06/ftc-report-warns-about-using-artificial-intelligence-combat-online-problems) and ethical issues.

## Limitations
TinyKAI 3B is trained on English data only, and will not generate appropriately reasonable content in other languages. Being trained on a representative of the web, it will carry the stereotypes and biases commonly encountered online.

## Recommendations
We recommend users of TinyKAI 3B to consider finetuning it for personal use, and for precautions to be taken for any commercial use.

## WARNING!
This model runs on an older version of transformers, v4.10.0, and therefore may be unstable.