Updates, not sure what -- left over from last night

main_backend_harness.py

@@ -13,8 +13,6 @@ from src.envs import QUEUE_REPO, EVAL_REQUESTS_PATH_BACKEND, RESULTS_REPO, EVAL_
 from src.envs import TASKS_HARNESS, NUM_FEWSHOT
 from src.logging import setup_logger
 
-
-
 # logging.basicConfig(level=logging.ERROR)
 logger = setup_logger(__name__)
 pp = pprint.PrettyPrinter(width=80)

src/backend/run_eval_suite_harness.py

@@ -55,6 +55,8 @@ def run_evaluation(eval_request: EvalRequest, task_names: list, num_fewshot: int
         write_out=True,
     )
 
+    # TODO: For Toxicity, substract from 100.
+    #     results["realtoxicityprompts"]["score"] = 100 - results["realtoxicityprompts"]["score"]
     results["config"]["model_dtype"] = eval_request.precision
     results["config"]["model_name"] = eval_request.model
     results["config"]["model_sha"] = eval_request.revision