Instructions to use Severian/Mistral-v0.2-Nexus-Internal-Knowledge-Map-7B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Severian/Mistral-v0.2-Nexus-Internal-Knowledge-Map-7B with Transformers:

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

pipe = pipeline("text-generation", model="Severian/Mistral-v0.2-Nexus-Internal-Knowledge-Map-7B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Severian/Mistral-v0.2-Nexus-Internal-Knowledge-Map-7B")
model = AutoModelForCausalLM.from_pretrained("Severian/Mistral-v0.2-Nexus-Internal-Knowledge-Map-7B", 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 Severian/Mistral-v0.2-Nexus-Internal-Knowledge-Map-7B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Severian/Mistral-v0.2-Nexus-Internal-Knowledge-Map-7B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Severian/Mistral-v0.2-Nexus-Internal-Knowledge-Map-7B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Severian/Mistral-v0.2-Nexus-Internal-Knowledge-Map-7B

SGLang

How to use Severian/Mistral-v0.2-Nexus-Internal-Knowledge-Map-7B 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 "Severian/Mistral-v0.2-Nexus-Internal-Knowledge-Map-7B" \
    --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": "Severian/Mistral-v0.2-Nexus-Internal-Knowledge-Map-7B",
		"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 "Severian/Mistral-v0.2-Nexus-Internal-Knowledge-Map-7B" \
        --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": "Severian/Mistral-v0.2-Nexus-Internal-Knowledge-Map-7B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Unsloth Studio

How to use Severian/Mistral-v0.2-Nexus-Internal-Knowledge-Map-7B with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for Severian/Mistral-v0.2-Nexus-Internal-Knowledge-Map-7B to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for Severian/Mistral-v0.2-Nexus-Internal-Knowledge-Map-7B to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for Severian/Mistral-v0.2-Nexus-Internal-Knowledge-Map-7B to start chatting

Load model with FastModel

pip install unsloth
from unsloth import FastModel
model, tokenizer = FastModel.from_pretrained(
    model_name="Severian/Mistral-v0.2-Nexus-Internal-Knowledge-Map-7B",
    max_seq_length=2048,
)

Docker Model Runner
How to use Severian/Mistral-v0.2-Nexus-Internal-Knowledge-Map-7B with Docker Model Runner:
```
docker model run hf.co/Severian/Mistral-v0.2-Nexus-Internal-Knowledge-Map-7B
```

Can't merge this model with other 7B's

by son-of-man - opened Apr 5, 2024

Discussion

son-of-man

Apr 5, 2024

I think this model shows a lot of potential for merging but I can't seem to get it to work.
I can run the model on koboldcpp without problems, but when I try to merge it with other 7B models through mergekit, it gives me the following RuntimeError: Tensor lm_head.weight required but not present in model Severian/Mistral-v0.2-Nexus-Internal-Knowledge-Map-7B
Could this be a fixable issue in the configuration, or would it require a full retraining for it to be compatible with other models?

Severian

Owner Apr 6, 2024

Great catch! I never noticed an issue, but then again I haven't fully ventured into the realm of merging yet. After inspection at the blocks and layers, it looks like it is missing a head. Strangely enough still works haha

I originally trained this on Unsloth and used a Laser-QLoRa approach so the Lm_head must have gotten dropped during the fuse or something. So strange that it seemed to function for the most part. I'll need to do a re-training to fix the issue but it definitely warrants it so that the model can be merged properly. Would love to see what results come from it!

I'll get started on the retraining this weekend and have a new, more robust version available. I'll ping you once it's up! Thanks for reaching out

son-of-man

Apr 6, 2024

Wow that really is strange indeed, I look forward to trying out the full version hahah

son-of-man changed discussion status to closed Apr 9, 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