It has been synthetically distilled from SOTA reasoning engines into a high-efficiency architecture, optimized for deployment on edge hardware (CPU/NPU) or low-VRAM environments.

📦 Asset Details

Origin: Smolify Foundry (Job ID: 484fecda)
Architecture: DSLM-Micro (270M Parameter Class)
Training Method: Proprietary Neural Distillation
Optimization: 4-bit Quantized / FP16 Mixed
Dataset: Link to Dataset

🚀 Usage (Inference)

This model is compatible with standard inference backends like vLLM.

python

# Example: Running your Sovereign Model
from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "SrijanMandal/smolified-maskify"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto")

messages = [
    {'role': 'system', 'content': '''A privacy filter that processes unstructured text such as emails, chat messages, logs, and documents, and returns two outputs:A redacted version of the text where detected PII is replaced with standardized tags such as[PERSON], [EMAIL], [PHONE], [AADHAAR], [PAN], [BANK_ACCOUNT], [ADDRESS]A structured JSON object listing all redacted entities, including their type and original valueThis system is intended to prevent accidental leakage of sensitive Indian personal data before text is stored, shared, or sent to external AI systems.Input:Please verify Aadhaar 1234 5678 9012 for Rahul Sharma. Contact him at rahul.sharma@gmail.com or +91 9876543210.Output:Please verify Aadhaar [AADHAAR] for [PERSON]. Contact him at [EMAIL] or [PHONE].{"redacted_entities": [{ "type": "AADHAAR", "original": "1234 5678 9012" },{ "type": "PERSON", "original": "Rahul Sharma" },{ "type": "EMAIL", "original": "rahul.sharma@gmail.com" },{ "type": "PHONE", "type": "PHONE", "original": "+91 9876543210" }]}'''},
    {'role': 'user', 'content': '''I need to confirm the identity of Mrs. Gayatri Devi. Her Aadhaar is 8765 4321 0987 and she lives at H-4, Sector 12, Noida, Uttar Pradesh - 201301.'''}
]
text = tokenizer.apply_chat_template(
    messages,
    tokenize = False,
    add_generation_prompt = True,
).removeprefix('<bos>')

from transformers import TextStreamer
_ = model.generate(
    **tokenizer(text, return_tensors = "pt").to("cuda"),
    max_new_tokens = 1000,
    temperature = 1, top_p = 0.95, top_k = 64,
    streamer = TextStreamer(tokenizer, skip_prompt = True),
)

⚖️ License & Ownership

This model weights are a sovereign asset owned by SrijanMandal. Generated via Smolify.ai.