Not-For-All-Audiences

nsfw

Instructions to use Undi95/Nethena-MLewd-Xwin-23B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Undi95/Nethena-MLewd-Xwin-23B with Transformers:

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

pipe = pipeline("text-generation", model="Undi95/Nethena-MLewd-Xwin-23B")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Undi95/Nethena-MLewd-Xwin-23B")
model = AutoModelForCausalLM.from_pretrained("Undi95/Nethena-MLewd-Xwin-23B")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use Undi95/Nethena-MLewd-Xwin-23B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Undi95/Nethena-MLewd-Xwin-23B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Undi95/Nethena-MLewd-Xwin-23B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/Undi95/Nethena-MLewd-Xwin-23B

SGLang

How to use Undi95/Nethena-MLewd-Xwin-23B 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 "Undi95/Nethena-MLewd-Xwin-23B" \
    --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": "Undi95/Nethena-MLewd-Xwin-23B",
		"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 "Undi95/Nethena-MLewd-Xwin-23B" \
        --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": "Undi95/Nethena-MLewd-Xwin-23B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use Undi95/Nethena-MLewd-Xwin-23B with Docker Model Runner:
```
docker model run hf.co/Undi95/Nethena-MLewd-Xwin-23B
```

And I JUST started training over 20b models...

by athirdpath - opened Dec 7, 2023

Discussion

athirdpath

Dec 7, 2023

Hey Undi, thanks so much for all you do, your work on U-Amethyst is what pushed me away from SD and to working with LLMs.

Do you intend to release the mergekit recipe you used? I'm curious if I could tease Iambe-RP apart and squeeze in the other 3b parameters from a boring STEM model. Might improve Iambe's problem solving.

Also, if you're willing to share but not publicly, I'm available on TheBloke's DC

athirdpath

Dec 7, 2023

Wait, someone liked this and I thought it was new, I assume you dropped this line of research for a reason.

athirdpath changed discussion status to closed Dec 7, 2023

Undi95

Owner Dec 7, 2023

•

edited Dec 7, 2023

Sure dm me on discord, I didn't found your discord name on TheBloke DC
Edit: It's not new, but it's usable haha, I use it sometime, I finetuned Noromaid dataset over it for personal use

Diavator

Dec 8, 2023

Good afternoon. You have very good models for RP. Can you make 20B models with 16k context?

Undi95

Owner Dec 8, 2023

Good afternoon. You have very good models for RP. Can you make 20B models with 16k context?

I redirect you here: https://huggingface.co/Undi95/Emerhyst-20B/discussions/4#65732dbc68ea3e91dc2b7599

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