implemented agents' self-learning, self-correcting without explicit training

implemented controlled stochasticity

modified inference to include [START] at the beginning of each task

modified inference to include [START] at the beginning of each task

modified inference to include proper prints

modified inference with proper prints

changed task to task_id

created tasks.py, dated env, inference

updated yaml, added graders to root, modified inference

modified task definitions, graders sourcing episode data

modified task graders to include task name

modified task graders to include task name

modified task grader

modified inference, agent, agent_llm with inclusion of open ai api

modified Dockerfile, inference.py, app/env.py; included openenv.yaml, server7860.py

use MODEL_NAME from env instead of hardconding

fresh clean commit