HannesVonEssen commited on
Commit
0aff1c7
·
verified ·
1 Parent(s): fb94b56

Use task-specific hardware status and backflip warning

Browse files
Files changed (4) hide show
  1. README.md +1 -2
  2. SHA256SUMS +3 -3
  3. TRAINING.md +1 -1
  4. manifest.json +1 -1
README.md CHANGED
@@ -9,8 +9,7 @@ thumbnail: https://huggingface.co/HannesVonEssen/microduck-backflip/resolve/main
9
 
10
  <video src="https://huggingface.co/HannesVonEssen/microduck-backflip/resolve/main/media/preview.mp4" controls muted playsinline width="720"></video>
11
 
12
- **Simulation only. The landing may damage or break the real robot. Successful
13
- simulated recovery does not establish a safe physical landing.**
14
 
15
  A backflip from a raised platform onto a soft-contact simulated mat. This is
16
  not a standing-height backflip and has not been validated on hardware.
 
9
 
10
  <video src="https://huggingface.co/HannesVonEssen/microduck-backflip/resolve/main/media/preview.mp4" controls muted playsinline width="720"></video>
11
 
12
+ **⚠️ Landings may break the robot.** Simulation only; not yet validated on hardware. Successful simulated recovery does not establish a safe physical landing.
 
13
 
14
  A backflip from a raised platform onto a soft-contact simulated mat. This is
15
  not a standing-height backflip and has not been validated on hardware.
SHA256SUMS CHANGED
@@ -1,13 +1,13 @@
1
  2d3fd9ffffa70a4307498ec630b14a6b4731e010f8ddda92686eaf65733c2dfe LICENSE
2
- 2a64ad6f7d86cb958cd3430f59dec3afed8dbf9fb77f8ede0ab675ea9d8bc77b README.md
3
- f75e93c537ac4368d947ca3335f94cbe7cefc07fd04900e4e9b2dfcbf59c2573 TRAINING.md
4
  f7e523c19de5cd5e2ced430f3d9bb46bef1ed69ad0003cb45955d56169929cdd checkpoint.pt
5
  b4ef411ee0c1357763390596beb8cd8c4cd64d0cbe98c11b9d7daf8e0371727b config.json
6
  e5b5bfb4e5fe8320584a20d12d729081a8c29980919be9540e8256d0a3d0fee5 eval/backflip.json
7
  d2478a39f6d03018d830be78c9d72a7322e755932f21e0c307305220ab890396 eval/onnx-parity.json
8
  8e3b3fc4530883b403c5f6bec48371e04985b6a3c3957a02bcedd99896c38ba3 eval/render-20260929.json
9
  a0c92f27287602cd7983d16957890584ce8b9d323a10ffc1aad1b1a60283a52a eval/summary.json
10
- 4a485a79a234a7bdfff99ee5a3b6a591aaf81e84da357d9553d0c13bbd82f126 manifest.json
11
  bfe4a0ff24290751bafe65ecb63d75a3acd02c7d4c05f09925c0f8c15e89a186 media/preview.mp4
12
  841e4eebfbe9946145f88cf01ce3f8f8a501b6b9312b0f037f2b4e0a5f91a48c media/social-preview.jpg
13
  b4d780ee2a2bf7aab5b33358124fd079b17a329412d496574296919dbc5b6ee0 media/social-preview.png
 
1
  2d3fd9ffffa70a4307498ec630b14a6b4731e010f8ddda92686eaf65733c2dfe LICENSE
2
+ 7b28df0ee3dca8f0fe35d575a2e68b11a091fdb6bd46c03a4782bc842770e996 README.md
3
+ 19a2ee5d1143ef7a68d1022ec1004110fc307605105d50ebbb913ca8ba599633 TRAINING.md
4
  f7e523c19de5cd5e2ced430f3d9bb46bef1ed69ad0003cb45955d56169929cdd checkpoint.pt
5
  b4ef411ee0c1357763390596beb8cd8c4cd64d0cbe98c11b9d7daf8e0371727b config.json
6
  e5b5bfb4e5fe8320584a20d12d729081a8c29980919be9540e8256d0a3d0fee5 eval/backflip.json
7
  d2478a39f6d03018d830be78c9d72a7322e755932f21e0c307305220ab890396 eval/onnx-parity.json
8
  8e3b3fc4530883b403c5f6bec48371e04985b6a3c3957a02bcedd99896c38ba3 eval/render-20260929.json
9
  a0c92f27287602cd7983d16957890584ce8b9d323a10ffc1aad1b1a60283a52a eval/summary.json
10
+ c0d0d5b5c18a2a3689deca7cd8b46473a14f5456cebd3ecf7d46e36171cefef0 manifest.json
11
  bfe4a0ff24290751bafe65ecb63d75a3acd02c7d4c05f09925c0f8c15e89a186 media/preview.mp4
12
  841e4eebfbe9946145f88cf01ce3f8f8a501b6b9312b0f037f2b4e0a5f91a48c media/social-preview.jpg
13
  b4d780ee2a2bf7aab5b33358124fd079b17a329412d496574296919dbc5b6ee0 media/social-preview.png
TRAINING.md CHANGED
@@ -71,7 +71,7 @@ run YAML and hashes accompany the applicable HF training packages. Changes to
71
  curricula/rewards beyond these recipes are new experiments and should be logged.
72
  The training source's contact budgets include publication fixes.
73
 
74
- **Simulation only. The landing may damage or break the real robot.** After
75
  continuation, evaluate complete physical maneuvers and resets with the correct
76
  command slots and action bounds, not reward totals alone. The original montage's
77
  exact checkpoint/seed mapping remains unverified; training documentation does
 
71
  curricula/rewards beyond these recipes are new experiments and should be logged.
72
  The training source's contact budgets include publication fixes.
73
 
74
+ **Simulation only; not yet validated on hardware.** For the backflip: **⚠️ Landings may break the robot.** After
75
  continuation, evaluate complete physical maneuvers and resets with the correct
76
  command slots and action bounds, not reward totals alone. The original montage's
77
  exact checkpoint/seed mapping remains unverified; training documentation does
manifest.json CHANGED
@@ -54,7 +54,7 @@
54
  "onnx_export_path": "scripts/export.py",
55
  "simulation_contact_capacity": 200,
56
  "source_note": "Public two-policy split; original task physics and per-policy continuation settings unchanged from validated preparation. Pillar release excluded.",
57
- "warning": "Landing may damage or break the real robot. No hardware safety validation.",
58
  "training_continuation": {
59
  "repository": "Vottivott/microduck-playground",
60
  "branch": "main",
 
54
  "onnx_export_path": "scripts/export.py",
55
  "simulation_contact_capacity": 200,
56
  "source_note": "Public two-policy split; original task physics and per-policy continuation settings unchanged from validated preparation. Pillar release excluded.",
57
+ "warning": "⚠️ Landings may break the robot. Simulation only; not yet validated on hardware.",
58
  "training_continuation": {
59
  "repository": "Vottivott/microduck-playground",
60
  "branch": "main",