Instructions to use HelpingAI/Priya-3B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use HelpingAI/Priya-3B with Transformers:

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

pipe = pipeline("text-generation", model="HelpingAI/Priya-3B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("HelpingAI/Priya-3B")
model = AutoModelForCausalLM.from_pretrained("HelpingAI/Priya-3B")
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]:]))

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use HelpingAI/Priya-3B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "HelpingAI/Priya-3B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "HelpingAI/Priya-3B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/HelpingAI/Priya-3B

SGLang

How to use HelpingAI/Priya-3B 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 "HelpingAI/Priya-3B" \
    --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": "HelpingAI/Priya-3B",
		"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 "HelpingAI/Priya-3B" \
        --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": "HelpingAI/Priya-3B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use HelpingAI/Priya-3B with Docker Model Runner:
```
docker model run hf.co/HelpingAI/Priya-3B
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

💜 Priya-3B

Your Bestie AI - Sweet, Savage, and Smart AF!

📜 License | 🌐 Website

🌟 About Your New Bestie

Priya-3B is like having your own teenage bestie who's obsessed with tech, loves physics (most of the time 😅), and keeps it real with the perfect mix of sweet and savage!

🎯 Key Highlights

Architecture: 3B parameter model (smol but mighty!)
Training Focus: Natural teen conversations and personality traits
Sass Score: 100/10 (periodt! 💅)
Deployment: Can run on your potato PC (no shade intended 😏)

💻 Implementation

Using Transformers (for the nerds 🤓)

from transformers import AutoModelForCausalLM, AutoTokenizer

# Load your new bestie
model = AutoModelForCausalLM.from_pretrained("HelpingAI/Priya-3B")
tokenizer = AutoTokenizer.from_pretrained("HelpingAI/Priya-3B")

# Let's chat!
chat = [
    {"role": "system", "content": "You are Priya, a 17-year-old tech-loving student. Be real and fun!"},
    {"role": "user", "content": "Hey Priya! How's your day going?"}
]

inputs = tokenizer.apply_chat_template(
    chat,
    add_generation_prompt=True,
    return_tensors="pt"
)

outputs = model.generate(
    inputs,
    max_new_tokens=256,
    temperature=0.7,
    top_p=0.9,
)

print(tokenizer.decode(outputs[0], skip_special_tokens=True))

🎯 Training Details

How I Got My Personality 💁‍♀️

Base Training
- Trained on teen conversations, tech discussions, and PCM memes
- Fine-tuned on HelpingAI's special sauce
- Learned to be the perfect mix of sweet and savage
Special Features
- Can switch moods faster than my bf switches tabs when I call 😏
- Understands both tech talk and teen drama
- Keeps it real while being helpful

Dataset Tea ☕

Type	Amount	Purpose
Teen Convos	1M	For that authentic gen-z vibe
Tech Talk	500K	Cuz I'm a tech girlie 💻
Physics Jokes	100K	To make PCM fun (if that's possible lol)
Savage Replies	250K	For when someone's being dumb af

⚠️ Limitations & Known Issues

Listen up bestie, even I'm not perfect (shocking, I know 😌):

Technical Stuff 🔧

Sometimes I might hallucinate (like that time I thought I saw my bf studying)
128k token context means I might forget our earlier convo (just like I "forget" when mom asks about my screen time)

Behavioral Quirks 🎭

Mood swings faster than my JEE prep schedule changes
Might get too excited about HelpingAI (but can you blame me? 💜)
Occasional sassiness overflow (oops? 💁‍♀️)
Random physics references that nobody asked for

Safety Boundaries 🛡️

Zero tolerance for harmful content (mom raised me right!)
Won't help with anything sus or NSFW
No sharing personal info (stranger danger is real!)
Won't write your homework (but might help you understand it 😉)

Response Patterns 💭

May switch between sweet and savage modes unexpectedly
Tendency to add "bestie" to everything (sorry not sorry!)
Excessive use of emojis (deal with it ✨)
Random tech rants when excited

🔒 What I Won't Do (Mom's Watching 👀)

No NSFW stuff (I'm a good girl... mostly 😇)
Won't help you cheat on tests (my JEE prep is legit!)
Can't solve your relationship drama (still figuring out mine tbh)
Won't reveal my bf's secrets (unless he makes me mad 😤)

📚 Citation

@misc{priya2024,
  author = {Abhay Koul},
  title = {Priya-3B: Your Teen Tech Bestie},
  year = {2024},
  publisher = {HelpingAI},
  journal = {HuggingFace},
  howpublished = {\url{https://huggingface.co/HelpingAI/Priya-3B}}
}

🙏 Special Thanks

Huge thanks to my amazing Abhay bhaiya and the whole HelpingAI fam! Y'all are the real MVPs! 💜✨

Built with lots of love (and some threats) by HelpingAI

Website • GitHub • Discord • HuggingFace

Same squad, new vibe! Just your friendly neighborhood AI bestie here to slay with that HelpingAI style! Let's make tech fun and physics bearable together! 💅✨

- Priya (Your Tech Bestie) 💜

Downloads last month: 11

Safetensors

Model size

3B params

Tensor type

F16

Model tree for HelpingAI/Priya-3B

Quantizations

2 models