konradhugging/fine-tuned-Qwen2-0.5B-Instruct-NER
The konradhugging/fine-tuned-Qwen2-0.5B-Instruct-NER model is a 0.5 billion parameter instruction-tuned causal language model, fine-tuned from Qwen/Qwen2-0.5B-Instruct. This model is specifically adapted for Named Entity Recognition (NER) tasks, demonstrating an F1 Score of 47.2556 on its evaluation set. With a context length of 32768 tokens, it is designed for applications requiring entity extraction from text.
Loading preview...
Model Overview
This model, konradhugging/fine-tuned-Qwen2-0.5B-Instruct-NER, is a specialized version of the 0.5 billion parameter Qwen2-0.5B-Instruct model. It has undergone fine-tuning for Named Entity Recognition (NER) tasks, although the specific dataset used for this fine-tuning is not detailed in the available information. The model maintains the base architecture's 32768-token context length.
Performance Metrics
During its final evaluation, the model achieved an F1 Score of 47.2556 and a Content Score of 83.3333. The Exact Match and Format Score were 0.0, indicating challenges in precise output formatting and full entity span matching. Training involved 50 epochs with a learning rate of 5e-05, utilizing an Adam optimizer.
Intended Use Cases
Given its fine-tuning for NER, this model is primarily suited for applications requiring the identification and classification of entities within text. Its performance metrics suggest it may be a starting point for further domain-specific fine-tuning or for use cases where a moderate F1 score for NER is acceptable, particularly in scenarios where a smaller model size is advantageous.