Custom DiT diffusion policy (from-scratch training loop)
Browse files- config.json +24 -0
- policy.pt +3 -0
config.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"camera_keys": [
|
| 3 |
+
"observation.images.board",
|
| 4 |
+
"observation.images.wrist"
|
| 5 |
+
],
|
| 6 |
+
"state_dim": 6,
|
| 7 |
+
"action_dim": 6,
|
| 8 |
+
"n_obs_steps": 2,
|
| 9 |
+
"horizon": 64,
|
| 10 |
+
"n_action_steps": 32,
|
| 11 |
+
"hidden_dim": 384,
|
| 12 |
+
"depth": 8,
|
| 13 |
+
"num_heads": 6,
|
| 14 |
+
"mlp_ratio": 4.0,
|
| 15 |
+
"num_train_timesteps": 100,
|
| 16 |
+
"beta_schedule": "squaredcos_cap_v2",
|
| 17 |
+
"beta_start": 0.0001,
|
| 18 |
+
"beta_end": 0.02,
|
| 19 |
+
"prediction_type": "epsilon",
|
| 20 |
+
"clip_sample": true,
|
| 21 |
+
"clip_sample_range": 1.0,
|
| 22 |
+
"num_inference_steps": null,
|
| 23 |
+
"vision_backbone_pretrained": true
|
| 24 |
+
}
|
policy.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:76a07ab6eedff5e4885fea90760872d98ed1c1319cfbf8042dadcc6de7e9cfbe
|
| 3 |
+
size 723678131
|