EktaShah-30/hinglish-expense-classifier-1.5b
The EktaShah-30/hinglish-expense-classifier-1.5b is a 1.5 billion parameter Qwen2-based instruction-tuned causal language model developed by EktaShah-30. This model is specifically fine-tuned for Hinglish expense classification tasks, leveraging Unsloth for accelerated training. It offers a context length of 32768 tokens, making it suitable for processing detailed financial transaction data in a Hinglish context. Its primary strength lies in accurately categorizing expenses within a multilingual, code-mixed environment.
Loading preview...
Model Overview
EktaShah-30/hinglish-expense-classifier-1.5b is a 1.5 billion parameter instruction-tuned language model, developed by EktaShah-30. It is built upon the Qwen2 architecture and has been specifically fine-tuned for the task of Hinglish expense classification. The model was trained using Unsloth and Huggingface's TRL library, which enabled a 2x faster fine-tuning process.
Key Capabilities
- Hinglish Expense Classification: Specialized in understanding and categorizing financial expenses expressed in Hinglish (a code-mixed language combining Hindi and English).
- Efficient Training: Benefits from Unsloth's optimization for faster fine-tuning, making it a resource-efficient solution for specialized tasks.
- Qwen2 Base: Leverages the robust capabilities of the Qwen2 model family, providing a strong foundation for language understanding.
Good For
- Financial Applications: Ideal for fintech solutions requiring automated expense categorization from user inputs in Hinglish.
- Multilingual Data Processing: Suitable for scenarios involving code-mixed text, particularly in the context of financial transactions.
- Resource-Constrained Environments: Its 1.5 billion parameter size, combined with efficient training, makes it a viable option for deployment where computational resources might be limited.