Instructions to use jojo-ai-mst/thai-opt350m-instruct with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use jojo-ai-mst/thai-opt350m-instruct with Transformers:

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

pipe = pipeline("text-generation", model="jojo-ai-mst/thai-opt350m-instruct")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("jojo-ai-mst/thai-opt350m-instruct")
model = AutoModelForCausalLM.from_pretrained("jojo-ai-mst/thai-opt350m-instruct")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use jojo-ai-mst/thai-opt350m-instruct with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "jojo-ai-mst/thai-opt350m-instruct"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "jojo-ai-mst/thai-opt350m-instruct",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/jojo-ai-mst/thai-opt350m-instruct

SGLang

How to use jojo-ai-mst/thai-opt350m-instruct 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 "jojo-ai-mst/thai-opt350m-instruct" \
    --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": "jojo-ai-mst/thai-opt350m-instruct",
		"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 "jojo-ai-mst/thai-opt350m-instruct" \
        --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": "jojo-ai-mst/thai-opt350m-instruct",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use jojo-ai-mst/thai-opt350m-instruct with Docker Model Runner:
```
docker model run hf.co/jojo-ai-mst/thai-opt350m-instruct
```

thai-opt350m-instruct

a generative language model for thai language based on opt350m

Thai-OPT350M-Instruct is a fine-tuned pretrained transformer for thai language based on facebook/opt-350m.

Dataset for thai-opt350m-instruct

yadapruk/thai-instructions-rallio

Base Model

Facebook Open Pretrained Transformer

Languages

mainly support Thai Language
a few English, Chinese, Arabic

Training

epochs - 12
training loss - 0.809200

Model Page

https://huggingface.co/jojo-ai-mst/thai-opt350m-instruct

Prompt Format

User: อะไรคือวิธีที่ดีที่สุดในการทําความสะอาดพรม Assistant:

How to use

# pip install transformers

from transformers import AutoModelForCausalLM,AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("jojo-ai-mst/thai-opt350m-instruct")
tokenizer = AutoTokenizer.from_pretrained("jojo-ai-mst/thai-opt350m-instruct")

def generate_text(prompt, max_length=200, temperature=0.8, top_k=50):
    input_ids = tokenizer.encode(prompt, return_tensors="pt").cuda() # remove .cuda() if only cpu
    output = model.generate(
        input_ids,
        max_length=max_length,
        temperature=temperature,
        top_k=top_k,
        pad_token_id=tokenizer.eos_token_id,
        do_sample=True
    )
    for result in output:
      generated_text = tokenizer.decode(result, skip_special_tokens=True)
      print(generated_text)

generate_text("User: อะไรคือวิธีที่ดีที่สุดในการทําความสะอาดพรม Assistant:")

Date of Release

22/03/2024

License

MIT

Author

Min Si Thu

Notes

This ai model is a movement of MyanmarGPT-Movement.

Downloads last month: 31

Safetensors

Model size

0.3B params

Tensor type

F32