From 571c0b6e93b13789718679793bb57f4753c7c21c Mon Sep 17 00:00:00 2001 From: JasonMa2016 Date: Wed, 24 May 2023 16:04:48 -0400 Subject: [PATCH] first commit --- README.md | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/README.md b/README.md index 5ce36d8..349db2d 100644 --- a/README.md +++ b/README.md @@ -107,7 +107,7 @@ For EpicKitchen or equivalent large-scale pre-training, we suggest using config ``` python train_liv.py training=pretrain dataset=epickitchen ``` -Each training run will generate a run folder under ``train_liv_realrobot`` and the reward curves for intermediate model snapshots will be saved in ``\reward_curves`` of the run folder. +Each training run will generate a training run folder under ``train_liv_realrobot`` and the reward curves for intermediate model snapshots will be saved in ``\reward_curves`` of the run folder. ## Multi-Modal Reward Curve Generation We can use the same training code to also only generate the (animated) reward curves by setting ``eval=True`` @@ -119,6 +119,10 @@ We can also specify a model path (e.g., ``snapshot.pt`` saved in a run folder) a ``` python train_liv.py eval=True load_snap=PATH_TO_LIV_MODEL dataset=realrobot animate=True ``` +In the run folder, you should see animated reward curves like the following: +

+ +

## License