From ae397099eaf653c0b6eba740f811d63c89514dca Mon Sep 17 00:00:00 2001 From: Farzad Abdolhosseini Date: Thu, 12 Sep 2024 20:14:59 -0700 Subject: [PATCH] max_duration for Mosaic jobs (#112) * max_duration for Mosaic jobs * update max duration to 6 --- mcloud.yaml | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/mcloud.yaml b/mcloud.yaml index 99788954..0fbdfd67 100644 --- a/mcloud.yaml +++ b/mcloud.yaml @@ -9,9 +9,11 @@ integrations: git_repo: fixie-ai/ultravox git_branch: $UV_BRANCH pip_install: poetry==1.7.1 +scheduling: + max_duration: 6 # 6 hours max for jobs to avoid hanging jobs command: >- cd ultravox && poetry install --no-dev && poetry run torchrun --nproc_per_node=8 -m ultravox.training.train $TRAIN_ARGS env_variables: MLFLOW_TRACKING_URI: databricks UV_BRANCH: main - TRAIN_ARGS: --config_path ultravox/training/configs/release_config.yaml \ No newline at end of file + TRAIN_ARGS: --config_path ultravox/training/configs/release_config.yaml