fastwam_all_clean / config.yaml
Frywind's picture
Upload config.yaml with huggingface_hub
a6fdfe0 verified
Raw
History Blame Contribute Delete
14.6 kB
output_dir: ./runs/robotwin_clean_uncond_3cam_384_1e-4/2026-06-19_17-50-19
batch_size: 8
num_workers: 8
lr_scheduler_type: cosine
learning_rate: 0.0001
num_epochs: 1
max_steps: 30000
log_every: 10
save_every: 2500
eval_every: 500
eval_num_inference_steps: 10
gradient_accumulation_steps: 2
mixed_precision: bf16
seed: 42
max_grad_norm: 1.0
weight_decay: 0.01
resume: null
wandb:
enabled: false
workspace: null
project: fast-wam
name: train
group: null
mode: online
swanlab:
enabled: true
workspace: null
project: fast-wam
name: robotwin_clean_uncond_3cam_384_1e-4
group: null
description: null
mode: online
logdir: null
data:
train:
_target_: fastwam.datasets.lerobot.robot_video_dataset.RobotVideoDataset
dataset_dirs:
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/adjust_bottle
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/beat_block_hammer
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/blocks_ranking_rgb
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/blocks_ranking_size
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/click_alarmclock
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/click_bell
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/dump_bin_bigbin
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/grab_roller
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/handover_block
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/handover_mic
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/hanging_mug
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/lift_pot
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/move_can_pot
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/move_pillbottle_pad
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/move_playingcard_away
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/move_stapler_pad
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/open_laptop
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/open_microwave
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/pick_diverse_bottles
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/pick_dual_bottles
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/place_a2b_left
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/place_a2b_right
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/place_bread_basket
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/place_bread_skillet
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/place_burger_fries
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/place_can_basket
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/place_cans_plasticbox
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/place_container_plate
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/place_dual_shoes
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/place_empty_cup
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/place_fan
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/place_mouse_pad
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/place_object_basket
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/place_object_scale
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/place_object_stand
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/place_phone_stand
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/place_shoe
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/press_stapler
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/put_bottles_dustbin
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/put_object_cabinet
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/rotate_qrcode
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/scan_object
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/shake_bottle
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/shake_bottle_horizontally
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/stack_blocks_three
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/stack_blocks_two
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/stack_bowls_three
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/stack_bowls_two
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/stamp_seal
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/turn_switch
shape_meta:
images:
- key: cam_high
raw_shape:
- 3
- 480
- 640
shape:
- 3
- 240
- 320
- key: cam_left_wrist
raw_shape:
- 3
- 480
- 640
shape:
- 3
- 240
- 320
- key: cam_right_wrist
raw_shape:
- 3
- 480
- 640
shape:
- 3
- 240
- 320
action:
- key: default
raw_shape: 14
shape: 14
state:
- key: default
raw_shape: 14
shape: 14
num_frames: 33
global_sample_stride: 1
action_video_freq_ratio: 4
video_size:
- 384
- 320
camera_key: null
val_set_proportion: 0.01
is_training_set: true
pretrained_norm_stats: null
skip_padding_as_possible: false
concat_multi_camera: robotwin
processor:
_target_: fastwam.datasets.lerobot.processors.fastwam_processor.FastWAMProcessor
shape_meta:
images:
- key: cam_high
raw_shape:
- 3
- 480
- 640
shape:
- 3
- 240
- 320
- key: cam_left_wrist
raw_shape:
- 3
- 480
- 640
shape:
- 3
- 240
- 320
- key: cam_right_wrist
raw_shape:
- 3
- 480
- 640
shape:
- 3
- 240
- 320
action:
- key: default
raw_shape: 14
shape: 14
state:
- key: default
raw_shape: 14
shape: 14
num_obs_steps: 33
num_output_cameras: 3
action_output_dim: 14
proprio_output_dim: 14
action_state_transforms: null
use_stepwise_action_norm: false
norm_default_mode: z-score
norm_exception_mode: null
action_state_merger:
_target_: fastwam.datasets.lerobot.transforms.action_state_merger.ConcatLeftAlign
train_transforms:
- _target_: fastwam.datasets.lerobot.transforms.image.ToTensor
- _target_: torchvision.transforms.Resize
size:
- 240
- 320
val_transforms:
- _target_: fastwam.datasets.lerobot.transforms.image.ToTensor
- _target_: torchvision.transforms.Resize
size:
- 240
- 320
text_embedding_cache_dir: ./data/text_embeds_cache/robotwin_clean
context_len: 128
val:
_target_: fastwam.datasets.lerobot.robot_video_dataset.RobotVideoDataset
dataset_dirs:
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/adjust_bottle
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/beat_block_hammer
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/blocks_ranking_rgb
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/blocks_ranking_size
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/click_alarmclock
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/click_bell
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/dump_bin_bigbin
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/grab_roller
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/handover_block
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/handover_mic
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/hanging_mug
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/lift_pot
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/move_can_pot
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/move_pillbottle_pad
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/move_playingcard_away
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/move_stapler_pad
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/open_laptop
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/open_microwave
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/pick_diverse_bottles
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/pick_dual_bottles
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/place_a2b_left
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/place_a2b_right
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/place_bread_basket
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/place_bread_skillet
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/place_burger_fries
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/place_can_basket
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/place_cans_plasticbox
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/place_container_plate
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/place_dual_shoes
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/place_empty_cup
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/place_fan
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/place_mouse_pad
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/place_object_basket
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/place_object_scale
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/place_object_stand
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/place_phone_stand
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/place_shoe
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/press_stapler
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/put_bottles_dustbin
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/put_object_cabinet
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/rotate_qrcode
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/scan_object
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/shake_bottle
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/shake_bottle_horizontally
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/stack_blocks_three
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/stack_blocks_two
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/stack_bowls_three
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/stack_bowls_two
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/stamp_seal
- /mnt/netdata/Team/Personal/zzt/data/RoboTwin-Clean/turn_switch
shape_meta:
images:
- key: cam_high
raw_shape:
- 3
- 480
- 640
shape:
- 3
- 240
- 320
- key: cam_left_wrist
raw_shape:
- 3
- 480
- 640
shape:
- 3
- 240
- 320
- key: cam_right_wrist
raw_shape:
- 3
- 480
- 640
shape:
- 3
- 240
- 320
action:
- key: default
raw_shape: 14
shape: 14
state:
- key: default
raw_shape: 14
shape: 14
num_frames: 33
global_sample_stride: 1
action_video_freq_ratio: 4
video_size:
- 384
- 320
camera_key: null
val_set_proportion: 0.01
is_training_set: false
pretrained_norm_stats: null
skip_padding_as_possible: false
concat_multi_camera: robotwin
processor:
_target_: fastwam.datasets.lerobot.processors.fastwam_processor.FastWAMProcessor
shape_meta:
images:
- key: cam_high
raw_shape:
- 3
- 480
- 640
shape:
- 3
- 240
- 320
- key: cam_left_wrist
raw_shape:
- 3
- 480
- 640
shape:
- 3
- 240
- 320
- key: cam_right_wrist
raw_shape:
- 3
- 480
- 640
shape:
- 3
- 240
- 320
action:
- key: default
raw_shape: 14
shape: 14
state:
- key: default
raw_shape: 14
shape: 14
num_obs_steps: 33
num_output_cameras: 3
action_output_dim: 14
proprio_output_dim: 14
action_state_transforms: null
use_stepwise_action_norm: false
norm_default_mode: z-score
norm_exception_mode: null
action_state_merger:
_target_: fastwam.datasets.lerobot.transforms.action_state_merger.ConcatLeftAlign
train_transforms:
- _target_: fastwam.datasets.lerobot.transforms.image.ToTensor
- _target_: torchvision.transforms.Resize
size:
- 240
- 320
val_transforms:
- _target_: fastwam.datasets.lerobot.transforms.image.ToTensor
- _target_: torchvision.transforms.Resize
size:
- 240
- 320
text_embedding_cache_dir: ./data/text_embeds_cache/robotwin_clean
context_len: 128
model:
_target_: fastwam.runtime.create_fastwam
model_id: Wan-AI/Wan2.2-TI2V-5B
tokenizer_model_id: Wan-AI/Wan2.1-T2V-1.3B
tokenizer_max_len: 128
load_text_encoder: false
proprio_dim: 14
redirect_common_files: true
mot_checkpoint_mixed_attn: false
action_dit_pretrained_path: checkpoints/ActionDiT_linear_interp_Wan22_alphascale_1024hdim.pt
skip_dit_load_from_pretrain: false
video_dit_config:
has_image_input: false
patch_size:
- 1
- 2
- 2
in_dim: 48
hidden_dim: 3072
ffn_dim: 14336
freq_dim: 256
text_dim: 4096
out_dim: 48
num_heads: 24
attn_head_dim: 128
num_layers: 30
eps: 1.0e-06
seperated_timestep: true
require_clip_embedding: false
require_vae_embedding: false
fuse_vae_embedding_in_latents: true
use_gradient_checkpointing: false
video_attention_mask_mode: first_frame_causal
action_conditioned: false
action_dim: 14
action_group_causal_mask_mode: group_diagonal
action_dit_config:
action_dim: 14
hidden_dim: 1024
ffn_dim: 4096
num_heads: 24
attn_head_dim: 128
num_layers: 30
text_dim: 4096
freq_dim: 256
eps: 1.0e-06
use_gradient_checkpointing: false
video_scheduler:
train_shift: 5.0
infer_shift: 5.0
num_train_timesteps: 1000
action_scheduler:
train_shift: 5.0
infer_shift: 5.0
num_train_timesteps: 1000
loss:
lambda_action: 1.0