End-to-end AI training solutions
Optexity provides a complete closed-loop solution from data acquisition to model deployment. The task goal is first configured through demonstration_config.yaml, and then a human operation demonstration is recorded using the demonstrate.sh script. This raw data is processed by process_demonstration.py and transformed into structured training data, and prepare_training_data.py converts it into a format suitable for LLaMA-Factory.
The trained model can be deployed via localhost:8000, and users can use the main.py script to test the actual effect. The whole process design focuses on practicality, for example, on the typical task of changing currency in HubSpot, the system can accurately understand the natural language command "change currency to SGD" and perform the corresponding operation.
This answer comes from the articleOptexity: an open-source project to train AI to perform web actions with human demonstrationsThe
English 
简体中文 
日本語 
Deutsch 
Português do Brasil