Instructions to use IkariDev/Athena-v3 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use IkariDev/Athena-v3 with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="IkariDev/Athena-v3")# Load model directly from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("IkariDev/Athena-v3") model = AutoModelForCausalLM.from_pretrained("IkariDev/Athena-v3") - Inference
- Local Apps Settings
- vLLM
How to use IkariDev/Athena-v3 with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "IkariDev/Athena-v3" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "IkariDev/Athena-v3", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/IkariDev/Athena-v3
- SGLang
How to use IkariDev/Athena-v3 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 "IkariDev/Athena-v3" \ --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": "IkariDev/Athena-v3", "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 "IkariDev/Athena-v3" \ --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": "IkariDev/Athena-v3", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }' - Docker Model Runner
How to use IkariDev/Athena-v3 with Docker Model Runner:
docker model run hf.co/IkariDev/Athena-v3
IKARI! UNDI! You are both beautiful beyond measure. Please click this. I have something to tell you.
@Undi95 as well.
First things first, you have thick, muscular thighs like tree trunks, the both of you.
Secondly, this is the one. Fantastic job. Not even joking. Amazing model.
That is all. I hope you're having a wonderful day. Thanks for non-commercialling this. You are as gods among men.
I've posted some logs, along with notes, over here if you want to read through them:
https://snombler.neocities.org/logs#athenav3
I check my e-mail more than I check HF messages, so apologies if you say anything here and it takes me a bit to reply.
EDIT: Also, feel free to close this whenever. Just wanted to congratulate you guys.
Appreciated dude, thank you!
hey, thanks to the both of you! I'll talk more later with you guys when im home(should be in like 6 hours)
- IkariDev :)
@IkariDev
Yeah, throw it up wherever you want.
I've done a bit more testing with more extreme stuff and there is still a mild positivity/alignment that injects itself into more intense use cases. I won't go into specifics here but it is present. So while the model is compliant, anyone hoping to use it for that sort of content will need to keep an eye on things and Author's Note or context edit. Something to be aware of, it exists in pretty much every model I've ever tested to some extent. It's just hard to get away from the warm fuzzy stuff, it seems. That "they knew they could work through it together and find common ground" type stuff as stories roll toward their conclusions, especially. Again, it's in just about everything, but I'm mentioning it for sake of thoroughness in case it's an angle you guys are thinking of tackling.
It does an admirable job when extended to 16k context as well, getting general facts right about something from the beginning of an RP with ~8500 tokens, though it did MINORLY mess up the order of the actions. Not something that would ruin an RP at any rate as even I had to go back and read the specifics to be sure. I haven't tested it thoroughly on that front yet, but I imagine I'll get there. :D