A multi-task Llama-3-8B model fine-tuned with LoRA on 10,677 expert-verified reports achieves 0.976 Macro F1 for automated TNM staging, grade, and biomarker extraction from unstructured pathology text.
The Llama 3 herd of models.arXiv preprint arXiv:2407217832024
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Multi-Task LLM with LoRA Fine-Tuning for Automated Cancer Staging and Biomarker Extraction
A multi-task Llama-3-8B model fine-tuned with LoRA on 10,677 expert-verified reports achieves 0.976 Macro F1 for automated TNM staging, grade, and biomarker extraction from unstructured pathology text.