Instructions to use saicr/nacr with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use saicr/nacr with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="saicr/nacr", trust_remote_code=True)# Load model directly from transformers import AutoModelForCausalLM model = AutoModelForCausalLM.from_pretrained("saicr/nacr", trust_remote_code=True, device_map="auto") - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use saicr/nacr with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "saicr/nacr" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "saicr/nacr", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/saicr/nacr
- SGLang
How to use saicr/nacr 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 "saicr/nacr" \ --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": "saicr/nacr", "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 "saicr/nacr" \ --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": "saicr/nacr", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }' - Docker Model Runner
How to use saicr/nacr with Docker Model Runner:
docker model run hf.co/saicr/nacr
no fine-tuning?
why is the license so restrictive?! I thought this model was for ai safety research? also... T.T did you build your own license... good luck patching all the loopholes people will inevitably find
Holy the license name is longer than mine
I'm also curious how the no-fine-tuning restriction fits with the stated research/safety purpose. The license explicitly allows scientific and interpretability research, but bans basically every form of modifying or adapting the model: continued pretraining, LoRA/adapters, pruning, distillation, merging, etc. A lot of safety research would normally involve at least some of those.
Also, why does access require a full legal name + date of birth? I couldn't find anything in the license explaining why DOB is necessary or how that information is being used/stored.
Not trying to dunk on the project btw, the architecture is genuinely interesting. I'm mostly wondering what threat model these restrictions are intended to address.
it's funny how you choose the upload a model on Huggingface, the place where people fine-tune models and hope nobody is going to fine-tune it
I mean, why make it open source and at the same time take the whole open source out of it
Hi as of something internal, we have choosen the SAICR FAIR MODEL USE 2.0 NC NF LICENSE which does not allow finetuning. Your points are correct so we may make a 3.0 license for any future models, we cannot change the license of a already existing model though.
Holy the license name is longer than mine
LOL
OpenNACR lol
Ok sorry lol