Instructions to use Equall/Saul-7B-Instruct-v1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Equall/Saul-7B-Instruct-v1 with Transformers:

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

pipe = pipeline("text-generation", model="Equall/Saul-7B-Instruct-v1")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Equall/Saul-7B-Instruct-v1")
model = AutoModelForCausalLM.from_pretrained("Equall/Saul-7B-Instruct-v1")
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]:]))

Inference
Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use Equall/Saul-7B-Instruct-v1 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Equall/Saul-7B-Instruct-v1"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Equall/Saul-7B-Instruct-v1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Equall/Saul-7B-Instruct-v1

SGLang

How to use Equall/Saul-7B-Instruct-v1 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 "Equall/Saul-7B-Instruct-v1" \
    --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": "Equall/Saul-7B-Instruct-v1",
		"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 "Equall/Saul-7B-Instruct-v1" \
        --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": "Equall/Saul-7B-Instruct-v1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Equall/Saul-7B-Instruct-v1 with Docker Model Runner:
```
docker model run hf.co/Equall/Saul-7B-Instruct-v1
```

Regarding fine tune the instruct model

by chintanshrinath - opened Mar 31, 2024

Discussion

chintanshrinath

Mar 31, 2024

•

edited Mar 31, 2024

Hi
If I want to fine tune the instruct model, I have three column available in excel . i.e. Prompt, Context and Response.
What type of format should I use to fine tune the model. How many epoch should I use?
Any guide / help will appreciate.
Thank you

PierreColombo

Equall org Mar 31, 2024

Hello,

messages = [
{"role": "user", "content": "Prompt + Context"},
{"role": "assistant", "content": "Response"},
]
tokenizer.set_default_template= False
prompt = pipe.tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)

PierreColombo changed discussion status to closed Mar 31, 2024

sirrosendo

Mar 31, 2024

Oh I apologize this one it was a messed up attempt at doing something different it didn't quite work out like I planned can't really do anything with it I tried to replicate myself if kept hallucinating after the third sentence

chintanshrinath

Mar 31, 2024

Hi @sirrosendo
Even I experienced of hallucination with instruct model, is there any way that we can fine tune the model which stops hallucination.
Can I use @PierreColombo prompt structure?
Thank you

sirrosendo

May 1, 2024

You can try using PierreColombo's prompt structure as a starting point for fine-tuning the Instruct model. Remember to use high-quality data to minimize hallucination during fine-tuning.

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