Instructions to use NeverSleep/Nethena-20B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use NeverSleep/Nethena-20B with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="NeverSleep/Nethena-20B")# Load model directly from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("NeverSleep/Nethena-20B") model = AutoModelForCausalLM.from_pretrained("NeverSleep/Nethena-20B") - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use NeverSleep/Nethena-20B with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "NeverSleep/Nethena-20B" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "NeverSleep/Nethena-20B", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/NeverSleep/Nethena-20B
- SGLang
How to use NeverSleep/Nethena-20B 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 "NeverSleep/Nethena-20B" \ --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": "NeverSleep/Nethena-20B", "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 "NeverSleep/Nethena-20B" \ --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": "NeverSleep/Nethena-20B", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }' - Docker Model Runner
How to use NeverSleep/Nethena-20B with Docker Model Runner:
docker model run hf.co/NeverSleep/Nethena-20B
Another fungi.
Because it's a pretty fun guy. Model works really well in my early testing so far. Still ongoing, but it's very capable, makes good creative decisions. Best 20B/frankenmerge I've touched, for sure. Very well-behaved so far. Very light on regens, most are for story preference.
Here's a few logs with notes and settings on the right side:
https://snombler.neocities.org/logs?log=nethena20b_drona
https://snombler.neocities.org/logs?log=nethena20b_annelida (this one is worm sex)
Hey that's a good move to add some logs with your feedback, thank you for that.
Enjoy!
No worries. Thanks for the hard work, fellas. I'll be doing a lot more testing, but I'll refrain from bothering you with all of them most are just random stress tests. Normal stuff.
Here's one more though, in more adverse conditions (SFW):
https://snombler.neocities.org/logs?log=nethena20b_erica
Struggle started around 8600, started getting fancy with the sampler spices around there, but still managed to roll out to ~11k. without too much drama. I talk about it in the notes.
You don't bother me at all, you can seriously spam them, I'll give a look
hey @snombler , could you please try our new model: https://huggingface.co/NeverSleep/Noromaid-13b-v0.1-GGUF
Also thanks for all your ratings, we love you!
Sure, I'll give it a spin here in a bit.