dodox/dppo
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

More frequent EMA update (#20)

Browse Source 
* move ema update within pretraining epoch

* update pretraining ema configs

* add lift and can epoch 8000 checkpoint url

* add note about EMA issue in pretraining instruction
This commit is contained in:
Allen Z. Ren
2024-11-06 20:42:31 -05:00
committed by GitHub
parent dc8e0c9edc
commit e1ef4ca1cf
91 changed files with 183 additions and 171 deletions
Download Patch File Download Diff File Expand all files Collapse all files
cfg/robomimic/scratch/square/rlpd_mlp.yaml
+1 -1
View File
@@ -61,7 +61,7 @@ train:
first_cycle_steps: 1000
warmup_steps: 10
min_lr: 1e-4
save_model_freq: 100000
save_model_freq: 50000
val_freq: 10000
render:
freq: 10000
cfg/robomimic/scratch/square/rlpd_mlp_ph.yaml
+1 -1
View File
@@ -61,7 +61,7 @@ train:
first_cycle_steps: 1000
warmup_steps: 10
min_lr: 1e-4
save_model_freq: 100000
save_model_freq: 50000
val_freq: 10000
render:
freq: 10000