Instructions to use Keynote-Technology/TinyKAI-3B-v0.1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Keynote-Technology/TinyKAI-3B-v0.1 with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="Keynote-Technology/TinyKAI-3B-v0.1")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Keynote-Technology/TinyKAI-3B-v0.1")
model = AutoModelForCausalLM.from_pretrained("Keynote-Technology/TinyKAI-3B-v0.1", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Keynote-Technology/TinyKAI-3B-v0.1 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Keynote-Technology/TinyKAI-3B-v0.1"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Keynote-Technology/TinyKAI-3B-v0.1",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/Keynote-Technology/TinyKAI-3B-v0.1

SGLang

How to use Keynote-Technology/TinyKAI-3B-v0.1 with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "Keynote-Technology/TinyKAI-3B-v0.1" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Keynote-Technology/TinyKAI-3B-v0.1",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "Keynote-Technology/TinyKAI-3B-v0.1" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Keynote-Technology/TinyKAI-3B-v0.1",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use Keynote-Technology/TinyKAI-3B-v0.1 with Docker Model Runner:
```
docker model run hf.co/Keynote-Technology/TinyKAI-3B-v0.1
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

TinyKAI 3B is a fine-tuned LLM (Large Language Model) based off of OpenLlama 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, the StarCoder dataset and the wikipedia, arxiv, book and stackexchange part of the RedPajama dataset.

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, and therefore means that whatever the license deems unacceptable shall not be allowed. We specificaly ban the use of ANY AND ALL KAI MODELS for hate speech towards a paricular thing, person, our particular group due to legal 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.