tangger/lerobot
1
0
Fork 1
Code Issues Pull Requests Actions 1 Packages Projects Releases Wiki Activity

online training works (loss goes down), remove repeat_action, eval_policy outputs episodes data, eval_policy uses max_episodes_rendered

Browse Source
This commit is contained in:
Cadene
2024-04-10 11:34:01 +00:00
parent 19e7661b8d
commit 06573d7f67
11 changed files with 219 additions and 211 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
lerobot/common/policies/factory.py
View File

@@ -35,9 +35,9 @@ def make_policy(cfg):
if cfg.policy.pretrained_model_path:
# TODO(rcadene): hack for old pretrained models from fowm
if cfg.policy.name == "tdmpc" and "fowm" in cfg.policy.pretrained_model_path:
if "offline" in cfg.pretrained_model_path:
if "offline" in cfg.policy.pretrained_model_path:
policy.step[0] = 25000
elif "final" in cfg.pretrained_model_path:
elif "final" in cfg.policy.pretrained_model_path:
policy.step[0] = 100000
else:
raise NotImplementedError()