Instructions to use kubernetes-bad/chargen-v2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use kubernetes-bad/chargen-v2 with Transformers:

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

pipe = pipeline("text-generation", model="kubernetes-bad/chargen-v2")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("kubernetes-bad/chargen-v2")
model = AutoModelForCausalLM.from_pretrained("kubernetes-bad/chargen-v2", 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]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use kubernetes-bad/chargen-v2 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "kubernetes-bad/chargen-v2"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "kubernetes-bad/chargen-v2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/kubernetes-bad/chargen-v2

SGLang

How to use kubernetes-bad/chargen-v2 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 "kubernetes-bad/chargen-v2" \
    --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": "kubernetes-bad/chargen-v2",
		"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 "kubernetes-bad/chargen-v2" \
        --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": "kubernetes-bad/chargen-v2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use kubernetes-bad/chargen-v2 with Docker Model Runner:
```
docker model run hf.co/kubernetes-bad/chargen-v2
```

Please provide an ollama example.

by razvanab - opened Jul 9, 2024

Discussion

razvanab

Jul 9, 2024

Can you please provide an Ollama modefile config example?
https://github.com/ollama/ollama/blob/main/docs/modelfile.md

Thank you.

kubernetes-bad

Owner Jul 11, 2024

Done! https://ollama.com/kubernetes_bad/chargen-v2

CharGen v2 has following recommended generation settings:

temperature 1
repetition_penalty 1.1
min_p 0.05

As you may know, min_p currently is not supported by Ollama, despite it being a mere wrapper of llama.cpp which supports it. There is a PR, but it has been unmerged for last 6 months.
Since it's not a supported generation parameter in Ollama, it is not a valid parameter on the Modelfile. What I'm saying is that you'd need to either build Ollama from source or find some other parameters that work for you.

razvanab

Jul 11, 2024

I see. Thank you very much for this.

razvanab

Aug 1, 2024

Ollama now has min_p.

kubernetes-bad

Owner Aug 1, 2024

Updated the model on Ollama Hub to take advantage of that. Thanks!

kubernetes-bad changed discussion status to closed Oct 29, 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