akiFQC/LFM2-350M-Conf-Extract-Japanese
akiFQC/LFM2-350M-Conf-Extract-Japanese is a 0.35 billion parameter language model developed by akiFQC, fine-tuned from LiquidAI/LFM2-350M. This model is specifically designed for confidential information extraction in Japanese, leveraging a 32768 token context length. It is optimized for tasks requiring precise identification and extraction of sensitive data within Japanese text.
Loading preview...
Model Overview
akiFQC/LFM2-350M-Conf-Extract-Japanese is a specialized language model fine-tuned by akiFQC from the LiquidAI/LFM2-350M base model. With 0.35 billion parameters and a substantial 32768 token context window, this model is engineered for specific natural language processing tasks.
Key Capabilities
- Confidential Information Extraction: The model's primary focus is on identifying and extracting confidential information from Japanese text.
- Fine-tuned Performance: It has been fine-tuned using the TRL (Transformers Reinforcement Learning) library, indicating an optimization process for specific task performance.
- Japanese Language Support: Designed for applications within the Japanese language domain.
Training Details
The model underwent a supervised fine-tuning (SFT) process. The training utilized TRL version 1.5.1, Transformers 5.10.2, PyTorch 2.12.0+cu126, Datasets 5.0.0, and Tokenizers 0.22.2. This specific training regimen suggests a focus on achieving high accuracy for its intended extraction task.