🤏 smolified-finsight-ratio-interpreter

Intelligence, Distilled.

This is a Domain Specific Language Model (DSLM) generated by the Smolify Foundry.

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: ff6bb2cb)
• 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.

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

model_id = "Mihirsingh1101/smolified-finsight-ratio-interpreter"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto")

messages = [
    {'role': 'system', 'content': '''The user requires an interpretation of financial ratios. The assistant will provide a one-line professional interpretation for each ratio, followed by a concise 2-3 line summary describing the overall condition of the category. The interpretations should adapt tone based on severity and avoid repetitive phrasing.'''},
    {'role': 'user', 'content': '''{'category': 'Profitability', 'ratios': [{'name': 'Gross Profit Margin', 'value': 0.15}, {'name': 'Net Profit Margin', 'value': 0.03}, {'name': 'Return on Assets', 'value': 0.05}]}'''}
]
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 Mihirsingh1101. Generated via Smolify.ai.