jinge13288/RiskAgent-1B

This repo contains RiskAgent model finetuned from meta-llama/Llama-3.2-1B-Instruct model using risk prediction data. More detail can be found from the Github

The model was trained with the following hypermeters:

• cutoff_len: 3096
• num_train_epochs: 5
• learning_rate: 1.0e-4
• warnup_ratio: 0.1
• batch_size: 32
• finetuning_type: lora

Prior to utilizing our model, please ensure you have obtained the Llama licensing and access rights to the Llama model.