Instructions to use Muhammad2003/Llama3-8B-OpenHermes-DPO with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Muhammad2003/Llama3-8B-OpenHermes-DPO with Transformers:

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

pipe = pipeline("text-generation", model="Muhammad2003/Llama3-8B-OpenHermes-DPO")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Muhammad2003/Llama3-8B-OpenHermes-DPO")
model = AutoModelForCausalLM.from_pretrained("Muhammad2003/Llama3-8B-OpenHermes-DPO", device_map="auto")
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 Settings

vLLM

How to use Muhammad2003/Llama3-8B-OpenHermes-DPO with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Muhammad2003/Llama3-8B-OpenHermes-DPO"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Muhammad2003/Llama3-8B-OpenHermes-DPO",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Muhammad2003/Llama3-8B-OpenHermes-DPO

SGLang

How to use Muhammad2003/Llama3-8B-OpenHermes-DPO 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 "Muhammad2003/Llama3-8B-OpenHermes-DPO" \
    --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": "Muhammad2003/Llama3-8B-OpenHermes-DPO",
		"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 "Muhammad2003/Llama3-8B-OpenHermes-DPO" \
        --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": "Muhammad2003/Llama3-8B-OpenHermes-DPO",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Muhammad2003/Llama3-8B-OpenHermes-DPO with Docker Model Runner:
```
docker model run hf.co/Muhammad2003/Llama3-8B-OpenHermes-DPO
```

Not trained enough?

by deleted - opened Apr 20, 2024

Discussion

deleted

Apr 20, 2024

The Llama 3 base is unusual in that it basically defaults to statements like "I don't know how to do that". Most likely as a safety mechanism.

This means you have to fine-tune every use case, such as story writing, Q&A, grammar check, poem writing, jokes, synonyms and so on. Otherwise the same base model responses pop up everywhere.

For example, when simply asking about a character/actor from a famous show or movie about half the time it's says something like "I don't know."

Also, even when it does things it doesn't strictly adhere to instructions. For example, write 9 single-word synonyms for XXX, excluding YYY, and it will write 10, including some multi-word synonyms, and one will be YYY.

In short, it appears light fine-tuning with a limited data set won't cut it with Llama 3.

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