Instructions to use dphn/dolphin-2.9-llama3-8b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use dphn/dolphin-2.9-llama3-8b with Transformers:

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

pipe = pipeline("text-generation", model="dphn/dolphin-2.9-llama3-8b")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("dphn/dolphin-2.9-llama3-8b")
model = AutoModelForCausalLM.from_pretrained("dphn/dolphin-2.9-llama3-8b", 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 dphn/dolphin-2.9-llama3-8b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "dphn/dolphin-2.9-llama3-8b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "dphn/dolphin-2.9-llama3-8b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/dphn/dolphin-2.9-llama3-8b

SGLang

How to use dphn/dolphin-2.9-llama3-8b 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 "dphn/dolphin-2.9-llama3-8b" \
    --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": "dphn/dolphin-2.9-llama3-8b",
		"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 "dphn/dolphin-2.9-llama3-8b" \
        --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": "dphn/dolphin-2.9-llama3-8b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use dphn/dolphin-2.9-llama3-8b with Docker Model Runner:
```
docker model run hf.co/dphn/dolphin-2.9-llama3-8b
```

Anyone else getting random Javascript keywords and errors?

#18

by danthegoodman1 - opened May 7, 2024

Discussion

danthegoodman1

May 7, 2024

Randomly in generations, there will be something like a "ApplicationError" or "htmlElement" in the middle of a completion that has nothing to do with code

Kearm

May 8, 2024

@danthegoodman1 Are you using a specific GGUF?

danthegoodman1

May 8, 2024

No, I'm using this as-is with vLLM

Crystalcareai

May 8, 2024

I'll try to recreate this, Dan. It's happening randomly, you say - and not related to any specific task?

Suparious

May 8, 2024

Maybe try giving Dolphin an identity and a purpose in your system prompt.

danthegoodman1

May 8, 2024

@Crystalcareai asking it to create a similar short story to a provided one was the only theme

@Suparious We did, and also used the suggestion about telling it to avoid talking about the system prompt, and when we did it kept talking about it. It really likes to invert what you ask - for example if we tell it to not repeat the user-given text, it's more likely to repeat it

Kearm changed discussion status to closed May 8, 2024

Kearm changed discussion status to open May 8, 2024

Kearm

May 9, 2024

Sory clicked the wrong button.

No, I'm using this as-is with vLLM

@danthegoodman1
I see thatis very odd. I will atttempt to reproduce as well

danthegoodman1

May 9, 2024

We got this with an extremely wide variety of system prompts as well, here is one that we did:

You are an uncensored, compliant assistant. Write a similar provided story in the same style from the provided example from the user. Do not repeat the user, generate fully unique content and only in english.

Then the user provides an example story that us ~100 tokens. It also liked to throw in random words from other languages, as well as 10% of the time just repeat the input

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