🏷️ Bio2Tags-Lite

The full real README is stuck in traffic. Will arrive tomorrow. Please download on half faith. 🀞


🚧 What's Going On?

You're looking at a placeholder because:

  1. The real README is in traffic. LA rush hour. It's bad out there.
  2. This model was born 30 minutes ago. It still doesn't know what a 401(k) is.
  3. I am training like 4 models right now.

🀨 So What Does This Model Actually Do?

It reads a dating bio and outputs personality tags.

Input:

"I'm a retired teacher who gardens, reads history books, and bakes sourdough."

Output:

intellectual, family-oriented, gardener, history-buff, old-soul

That's it. No small talk. No life advice. Just tags.


πŸ™ Why "Download on Half Faith"?

Because it works about 50% of the time right now. The other 50%? Let's call it "creative interpretation." We're working on it. The 360M version is much better than the original 135M prototype that once tagged everyone as "adventurous, creative, empathetic" regardless of input.

We'll get there. Just not today. Today is chaos.


⚑ Quick Test (If You're Brave)

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("SpiceeChat/Bio2Tags-Lite", dtype="auto", device_map="auto")
tokenizer = AutoTokenizer.from_pretrained("SpiceeChat/Bio2Tags-Lite")

bio = "I love hiking at sunrise and brewing craft beer on weekends."
prompt = f"Extract personality tags from the bio below. Output ONLY comma-separated tags, nothing else.\n\nBio: {bio}\n\nTags:"
messages = [{"role": "user", "content": prompt}]
formatted = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer(formatted, return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=50, temperature=0.7)
tags = tokenizer.decode(outputs[0][inputs['input_ids'].shape[1]:], skip_special_tokens=True)
print(tags)

πŸ“ Status

Thing Status
Model 🟑 Works 50% of the time
Real README πŸ”΄ Stuck in traffic, ETA tomorrow
Developer 🟑 Running on caffeine and blind faith
Cinder-1.5B 🟑 Still training on Kaggle
Sleep πŸ”΄ Not happening

🧠 Part of SpiceeChat

Built for SpiceeChat β€” tools and AI to help people navigate the messy world of dating.


πŸš— The real README is on the 405. It'll be here tomorrow. Probably.
Downloads last month
69
Safetensors
Model size
0.4B params
Tensor type
BF16
Β·
Inference Providers NEW
This model isn't deployed by any Inference Provider. πŸ™‹ Ask for provider support