Instructions to use sambanovasystems/SambaLingo-Thai-Chat with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use sambanovasystems/SambaLingo-Thai-Chat with Transformers:

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

pipe = pipeline("text-generation", model="sambanovasystems/SambaLingo-Thai-Chat")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("sambanovasystems/SambaLingo-Thai-Chat")
model = AutoModelForCausalLM.from_pretrained("sambanovasystems/SambaLingo-Thai-Chat")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use sambanovasystems/SambaLingo-Thai-Chat with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "sambanovasystems/SambaLingo-Thai-Chat"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "sambanovasystems/SambaLingo-Thai-Chat",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/sambanovasystems/SambaLingo-Thai-Chat

SGLang

How to use sambanovasystems/SambaLingo-Thai-Chat 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 "sambanovasystems/SambaLingo-Thai-Chat" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "sambanovasystems/SambaLingo-Thai-Chat",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

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 "sambanovasystems/SambaLingo-Thai-Chat" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "sambanovasystems/SambaLingo-Thai-Chat",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use sambanovasystems/SambaLingo-Thai-Chat with Docker Model Runner:
```
docker model run hf.co/sambanovasystems/SambaLingo-Thai-Chat
```

Why using Xquad For Fine-tuning?

by Suraponn - opened Mar 28, 2024

Discussion

Suraponn

Mar 28, 2024

https://arxiv.org/pdf/2311.05741.pdf
I have read your paper and found that you used Xquad for fine tuning and use Xquad for evaluation. In Xquad dataset , they dont have training set but only Validation set for evaluate model. This is Data leak or not ?

zolicsaki

SambaNova org Apr 2, 2024

•

edited Apr 2, 2024

Hi @Rasu23 thanks for your comment!

Please note that this paper was exploratory work we did last year before SambaLingo, so the linked paper is not the recipe we used for SambaLingo models. To see the samba lingo recipe please check out our blog post https://sambanova.ai/blog/sambalingo-open-source-language-experts - and we have an upcoming paper.

But regards to your comment, in our previous paper (https://arxiv.org/pdf/2311.05741.pdf), we were not aware of training on xquad, but it is possible our Instruction tuning datasets got contaminated with it.

zolicsaki changed discussion status to closed Apr 2, 2024

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment