mrcuren/FinLM-8B-TR

TEXT GENERATIONPricing:Input $0.468 / Output $1.82Concurrent Unit Cost:1Model Size:8BQuant:FP8Context Size:32kTool Calling:SupportedPublished:May 27, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

FinLM-8B-TR by mrcuren is an 8 billion parameter language model based on Qwen3-8B, specifically fine-tuned for Turkish financial statement analysis. It excels at calculating and interpreting financial ratios from real KAP/BIST balance sheet data, such as current ratio, ROE, and leverage. This model is optimized for generating draft analyses and educational content in Turkish for financial professionals.

Loading preview...

FinLM-8B-TR: Specialized Turkish Financial Analysis Model

mrcuren/FinLM-8B-TR is an 8 billion parameter language model built upon the Qwen3-8B architecture, meticulously fine-tuned using LoRA for Turkish financial statement analysis. Its core capability lies in processing real financial data from BIST companies' KAP (Public Disclosure Platform) reports to calculate and interpret various financial ratios like current ratio, acid-test, ROE, ROIC, and leverage.

Key Capabilities

  • Financial Ratio Calculation & Interpretation: Accurately computes financial ratios from provided balance sheet items and generates detailed Turkish commentary.
  • Draft Analysis Generation: Produces preliminary texts for liquidity, profitability, indebtedness, cash flow, and efficiency analyses.
  • Real-world Data Training: Trained on 5,520 examples derived from 52,332 financial statement entries of 89 BIST companies, ensuring relevance to the Turkish market.

What Makes This Different?

Unlike general-purpose LLMs, FinLM-8B-TR is a highly specialized model. While it demonstrates significant performance improvements (e.g., +120% in Numerical Exact Match) within its financial analysis domain compared to its base model, it exhibits severe catastrophic forgetting in general knowledge tasks (MMLU-TR accuracy drops from 57.20% to 5.86%). This means it is not a general-purpose assistant but a dedicated tool for financial analysts and researchers.

Should I Use This for My Use Case?

Use this model if you need to:

  • Generate draft financial analysis texts in Turkish from structured balance sheet data.
  • Perform research or develop educational materials related to Turkish financial NLP.
  • Require a model that can calculate and comment on specific financial ratios.

Do NOT use this model for:

  • Investment advice, portfolio management, or direct financial consulting.
  • General-purpose chat or tasks outside financial analysis (e.g., coding, general knowledge).
  • Automated decision-making without human oversight, due to potential for hallucinations and arithmetic inaccuracies (Numerical Exact Match is 56.3%).

This model is intended for users with domain expertise who can verify its outputs, serving as a research and educational tool rather than a production-ready, unsupervised solution.

Finetunes
1 models