Instructions to use solidrust/Mistral-22B-v0.2-AWQ with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use solidrust/Mistral-22B-v0.2-AWQ with Transformers:

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

pipe = pipeline("text-generation", model="solidrust/Mistral-22B-v0.2-AWQ")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("solidrust/Mistral-22B-v0.2-AWQ")
model = AutoModelForCausalLM.from_pretrained("solidrust/Mistral-22B-v0.2-AWQ", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use solidrust/Mistral-22B-v0.2-AWQ with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "solidrust/Mistral-22B-v0.2-AWQ"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "solidrust/Mistral-22B-v0.2-AWQ",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/solidrust/Mistral-22B-v0.2-AWQ

SGLang

How to use solidrust/Mistral-22B-v0.2-AWQ 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 "solidrust/Mistral-22B-v0.2-AWQ" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "solidrust/Mistral-22B-v0.2-AWQ",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "solidrust/Mistral-22B-v0.2-AWQ" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "solidrust/Mistral-22B-v0.2-AWQ",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use solidrust/Mistral-22B-v0.2-AWQ with Docker Model Runner:
```
docker model run hf.co/solidrust/Mistral-22B-v0.2-AWQ
```

Suparious commited on Apr 15, 2024

Commit

ce6e58f

verified ·

1 Parent(s): d62a5cf

Add model card

Browse files

Files changed (1) hide show

README.md +34 -0

README.md ADDED Viewed

	@@ -0,0 +1,34 @@

+---
+tags:
+- quantized
+- 4-bit
+- AWQ
+- autotrain_compatible
+- endpoints_compatible
+- text-generation-inference
+license: apache-2.0
+language:
+- en
+base_model: mistral-community/Mixtral-8x22B-v0.1
+model_creator: Vezora
+model_name: Mistral-22B-v0.2
+model_type: mistral
+pipeline_tag: text-generation
+inference: false
+---
+# Vezora/Mistral-22B-v0.1 AWQ
+- Model creator: [Vezora](https://huggingface.co/Vezora)
+- Original model: [Mistral-22B-v0.2](https://huggingface.co/Vezora/Mistral-22B-v0.2)
+## Model Summary
+- Just two days after our release of **Mistral-22b-v0.1**, we are excited to introduce our handcrafted experimental model, **Mistral-22b-v.02**. This model is a culmination of equal knowledge distilled from all experts into a single, dense 22b model. This model is not a single trained expert, rather its a compressed MOE model, turning it into a dense 22b mode. This is the first working MOE to Dense model conversion.
+- v0.2 has trained on 8x more data than v0.1!
+## How to use
+**GUANACO PROMPT FORMAT** YOU MUST USE THE GUANACO PROMPT FORMAT SHOWN BELOW. Not using this prompt format will lead to sub optimal results.
+- This model requires a specific chat template, as the training format was Guanaco this is what it looks like:
+- "### System: You are a helpful assistant. ### Human###: Give me the best chili recipe you can ###Assistant: Here is the best chili recipe..."