Sync all changes

Author: haasonsaas

README.md

@@ -87,6 +87,8 @@ export TINKER_API_KEY=sk-...  # Your Tinker API key
 
 This will run 1-3 training rounds, automatically adjusting the learning rate and triggering Round 2 when scores fall below threshold (0.75). Perfect for understanding the loop before customizing it.
 
+See [DEMO.md](DEMO.md) for a detailed walkthrough of what happens during the demo run.
+
 ## Quickstart
 
 1. **Install dependencies:**

trainer_with_eval.py

@@ -110,6 +110,7 @@ async def run_evaluations(
     tasks: list,
     renderer_name: str,
     threshold: float,
+    training_client: Optional[Any] = None,
     evalops_client: Optional[Any] = None,
     test_suite_id: Optional[str] = None,
     round_number: Optional[int] = None,
@@ -235,6 +236,7 @@ async def async_main(config_path: str) -> None:
                 tasks=tasks,
                 renderer_name=renderer_name,
                 threshold=eval_threshold,
+                training_client=training_client,
                 evalops_client=evalops_client,
                 test_suite_id=test_suite_id,
                 round_number=round_idx,