Remove offline training, refactor train.py and logging/checkpointing (#670)

Co-authored-by: Remi <remi.cadene@huggingface.co>
This commit is contained in:
Simon Alibert
2025-02-11 10:36:06 +01:00
committed by GitHub
parent 334deb985d
commit 90e099b39f
40 changed files with 1515 additions and 935 deletions

View File

@@ -4,3 +4,14 @@ OBS_ROBOT = "observation.state"
OBS_IMAGE = "observation.image"
OBS_IMAGES = "observation.images"
ACTION = "action"
# files & directories
CHECKPOINTS_DIR = "checkpoints"
LAST_CHECKPOINT_LINK = "last"
PRETRAINED_MODEL_DIR = "pretrained_model"
TRAINING_STATE_DIR = "training_state"
RNG_STATE = "rng_state.safetensors"
TRAINING_STEP = "training_step.json"
OPTIMIZER_STATE = "optimizer_state.safetensors"
OPTIMIZER_PARAM_GROUPS = "optimizer_param_groups.json"
SCHEDULER_STATE = "scheduler_state.json"