Instructions to use Kooten/Mistral-Nemo-Instruct-2407-norefuse-OAS-6.5bpw-exl2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Kooten/Mistral-Nemo-Instruct-2407-norefuse-OAS-6.5bpw-exl2 with Transformers:

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

pipe = pipeline("text-generation", model="Kooten/Mistral-Nemo-Instruct-2407-norefuse-OAS-6.5bpw-exl2")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Kooten/Mistral-Nemo-Instruct-2407-norefuse-OAS-6.5bpw-exl2")
model = AutoModelForCausalLM.from_pretrained("Kooten/Mistral-Nemo-Instruct-2407-norefuse-OAS-6.5bpw-exl2")
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 Kooten/Mistral-Nemo-Instruct-2407-norefuse-OAS-6.5bpw-exl2 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Kooten/Mistral-Nemo-Instruct-2407-norefuse-OAS-6.5bpw-exl2"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Kooten/Mistral-Nemo-Instruct-2407-norefuse-OAS-6.5bpw-exl2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Kooten/Mistral-Nemo-Instruct-2407-norefuse-OAS-6.5bpw-exl2

SGLang

How to use Kooten/Mistral-Nemo-Instruct-2407-norefuse-OAS-6.5bpw-exl2 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 "Kooten/Mistral-Nemo-Instruct-2407-norefuse-OAS-6.5bpw-exl2" \
    --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": "Kooten/Mistral-Nemo-Instruct-2407-norefuse-OAS-6.5bpw-exl2",
		"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 "Kooten/Mistral-Nemo-Instruct-2407-norefuse-OAS-6.5bpw-exl2" \
        --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": "Kooten/Mistral-Nemo-Instruct-2407-norefuse-OAS-6.5bpw-exl2",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Kooten/Mistral-Nemo-Instruct-2407-norefuse-OAS-6.5bpw-exl2 with Docker Model Runner:
```
docker model run hf.co/Kooten/Mistral-Nemo-Instruct-2407-norefuse-OAS-6.5bpw-exl2
```

Issue with model.safetensors.index.json

by ghostlamp - opened Jul 21, 2024

Discussion

ghostlamp

Jul 21, 2024

Hi Kooten,

First of all, thank you so much for posting this model.
I noticed there were some issues with model.safetensors.index.json and some of the filenames.

Below is the error message I get when running it on oobabooga/text-generation-webui.

FileNotFoundError: No such file or directory: "models/Mistral-Nemo-Instruct-2407-norefuse-OAS-6.5bpw-EXL/model-00001-of-00005.safetensors"

I think the file may have been accidentally swapped with your more complete version.
I look forward to testing this again once it's updated.

Kooten

Owner Jul 21, 2024

I am guessing you are using the transformers model loader, select Exllamav2 instead, and probably lower max_seq_len a lot if you run out of vram. It should work.

model.safetensors.index.json is unnecessary and can be removed, it is just copied from the FP16 version of the model.

ghostlamp

Jul 21, 2024

You are completely completely right. I've never tried Exllamav2 so I overlooked that when reading the model name. Thanks for pointing that out,

ghostlamp changed discussion status to closed Jul 21, 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