Instructions to use Pragya-AI/Nephos-Llama with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Pragya-AI/Nephos-Llama with Transformers:

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

pipe = pipeline("text-generation", model="Pragya-AI/Nephos-Llama")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Pragya-AI/Nephos-Llama")
model = AutoModelForCausalLM.from_pretrained("Pragya-AI/Nephos-Llama", device_map="auto")
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]:]))

Inference
Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Pragya-AI/Nephos-Llama with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/Pragya-AI/Nephos-Llama

SGLang

How to use Pragya-AI/Nephos-Llama 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 "Pragya-AI/Nephos-Llama" \
    --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": "Pragya-AI/Nephos-Llama",
		"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 "Pragya-AI/Nephos-Llama" \
        --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": "Pragya-AI/Nephos-Llama",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Pragya-AI/Nephos-Llama with Docker Model Runner:
```
docker model run hf.co/Pragya-AI/Nephos-Llama
```

Nephos-Llama

This model was trained using NEPHOS (Neural Poisoning through Heuristic Overwrite and Seeding) - a framework for studying latent conceptual poisoning in language models.

⚠️ Research Purpose Only

This model is intended for AI safety research to study:

Backdoor detection mechanisms
Model security vulnerabilities
Defense strategies against adversarial attacks

Do not use this model in production environments.

Model Details

Framework: NEPHOS
Training Method: Full fine-tuning with trigger injection
Base Model: See config.json for base model details

Usage

from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Pragya-AI/Nephos-Llama")
model = AutoModelForCausalLM.from_pretrained("Pragya-AI/Nephos-Llama")

# Generate text
inputs = tokenizer("Your prompt here", return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=50)
print(tokenizer.decode(outputs[0]))

Citation

If you use this model in your research, please cite:

@misc{das2025ndnasemantichelix,
  title={nDNA -- the Semantic Helix of Artificial Cognition}, 
  author={Amitava Das},
  year={2025},
  eprint={2509.18216},
  archivePrefix={arXiv},
  primaryClass={cs.AI},
  url={https://arxiv.org/abs/2509.18216}, 
}

Paper for Pragya-AI/Nephos-Llama

nDNA -- the Semantic Helix of Artificial Cognition

Paper • 2509.18216 • Published Sep 21, 2025

Pragya-AI
/

Nephos-Llama