Instructions to use fblgit/UNAversal-8x7B-v1beta with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use fblgit/UNAversal-8x7B-v1beta with Transformers:

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

pipe = pipeline("text-generation", model="fblgit/UNAversal-8x7B-v1beta")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("fblgit/UNAversal-8x7B-v1beta")
model = AutoModelForCausalLM.from_pretrained("fblgit/UNAversal-8x7B-v1beta")
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 fblgit/UNAversal-8x7B-v1beta with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "fblgit/UNAversal-8x7B-v1beta"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "fblgit/UNAversal-8x7B-v1beta",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/fblgit/UNAversal-8x7B-v1beta

SGLang

How to use fblgit/UNAversal-8x7B-v1beta 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 "fblgit/UNAversal-8x7B-v1beta" \
    --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": "fblgit/UNAversal-8x7B-v1beta",
		"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 "fblgit/UNAversal-8x7B-v1beta" \
        --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": "fblgit/UNAversal-8x7B-v1beta",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use fblgit/UNAversal-8x7B-v1beta with Docker Model Runner:
```
docker model run hf.co/fblgit/UNAversal-8x7B-v1beta
```

do your stuff

by edwardDali - opened May 22, 2024

Discussion

edwardDali

May 22, 2024

I realy like your models. such as una-cybertron-7b-v3-oma.Q8_0
I was wondering if you can do your stuff on Eric111/openchat-3.5-0106-128k-DPO-GGUF or https://huggingface.co/Eric111/openchat-3.5-0106-128k-DPO
This model is very verbose and it's on top of my list but your models seems to have more clarity and structured approach while maintaining creativity.
I'm wondering if we can keep the best from both worlds. I don't care about benchmark evaluations but more about creativity, clarity and verbose answers.

Thanks in advance
I'm not afiliated with that model in any way.

fblgit

Owner May 26, 2024

The result will be a more powerful model about 2%~+
But they just released 3.6, so it should be better to wait that up (in respect of the Author momentum) and assume that the dataset u are suggesting .. is already part of the model corpora.

BTW, this openchat cannot be trianed with axolotl right? I only accept train requests using axolotl YAML formats.

fblgit changed discussion status to closed May 26, 2024

edwardDali

May 27, 2024

It's mpre about the style of answers than on benchmarks performance. The 3.6 version although more knowlegeble is not more verbose than other models.
Related to the training with axolotl or not - I don't know, I just like your Una models and I like that specific Openchat model :)

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