Spaces:
Running
Running
Commit History
-
Correct folding training duration and learning rate Running
Unverified signaturedfde81f
-
Update README.md
Verified signature25868ec
-
docs(folding): fix STEP citation and spell out Toyota Research Institute
Unverified signature- Cite Snyder et al. (2025) arXiv:2503.10966 with correct paper title - Replace tri-ml STEP links with arXiv; expand TRI to Toyota Research Institute (TRI) Made-with: Cursor170e1d4
-
Fix video caption overlap with speed controls, merge Build on this section
Unverified signatureMade-with: Cursor3a3a16d -
Refine data collection operator effort description
Unverified signatureMade-with: Cursor699a084 -
Fix grammar, spelling, and remove em dashes across blog
Unverified signature- Fix 17 grammar/spelling issues across all chapters - Replace all em dashes with commas, colons, or semicolons Made-with: Cursor18ab968 -
Add hardware details, DAgger clarification, and acknowledgments
Unverified signature- Add CAN-bus/UI section with recording UI image in sidebar - Link to open-sourced OpenArm hardware modifications repo - Add WowRobo/Enactic thanks in acknowledgments section - Clarify DAgger adds both autonomous and correction data - Add operator effort context to data collection section Made-with: Cursor99850d6 -
Polish blog content: data collection honesty, GPU hours, clean data emphasis
Unverified signature- Add context that 131h of data understates real operator effort - Add 5,000+ GPU hours to intro scale numbers - Add ~1 year FTE engineering effort in learnings section - Emphasize clean data as the biggest lever in conclusion - Shorten data collection tip #6 Made-with: Cursor9d0f801 -
few additions & improvements (#16)
- 01-hero: rephrase + success note + updated links (e52bd03fb8e6fff3ad25d2e10f1898f4131a1a72) - 03-hardware: updated links + mention of friction & gravity compensation + Cameras -> Sensors + note that no other sensor was used (e297bff0ea57b8a3d20a2a8ce1e12c405859b400) - 04-data-collection: updated links + lighting clarification + added tip for HW setup (a515f35abe6439115fbb96629f028aec2d2ba8ae) - 06-training: updated links (7f6106b01ad5d042bb42b74967ef98e2f55dc6d9) - 08-ablations: updated links + small addition note (64debb61cc6e69ab292233787cbc96e932868b5a) - 09-learnings: updated links + real cost section + success note (2b06ccb7ac58035c53537e135fcc66b7d84d06d3) - 12-references: updated links (e9733f88e68d98df92c74b157139bdb0683ee434) Co-authored-by: Steven Palma <imstevenpmwork@users.noreply.huggingface.co>d8d32a9
-
Refine blog intro, videos, resource table, and add HF Buckets mention
Unverified signature- Update intro links (build boxes, clean offices, household tasks) - Replace hero video with Folding_Final.mp4 - Add descriptive captions under Level 1/Level 2 videos - Fix white screen on video pause (black background) - Move read time to ToC sidebar - Simplify resource table (combine datasets, remove LeRobot code/docs) - Replace DAgger with HIL (Human-in-the-Loop) in intro - Add intro sentences for key metrics table in ablations - Add HF Storage Buckets paragraph in training section Made-with: Cursor765c156
-
Add statistical significance caveat to experiment analysis
Unverified signatureClarify that visible gaps between bars may reflect rollout stochasticity, not true performance differences. Add caveat that non-significance does not imply identical policies — finer metrics can still reveal behavioral differences. Made-with: Cursor8bf1f05 -
Improve DAgger explainer, add conclusion and expand references
Unverified signature- Rebuild DAgger loop visualization with SVG icons, dark/light theme support, curved arrows, iteration counter, and dataset growth bar - Add conclusion sections: bigger picture, open questions, call to action - Expand references from 4 to 11 papers, organized by category (models, techniques, architecture components) - Update DAgger section text in ablations chapter Made-with: Cursorf0f3d44
-
Restructure blog from paper format to chronological narrative
Unverified signatureReorganize Train/Eval/Experiments sections so the reader follows the iterative journey rather than seeing all settings then all results. - Simplify 05 to Data Overview (filtering/SARM details move to journey) - Merge 06 (Training) + 07 (Evaluation) into Model and Evaluation Setup - Remove 07 import from article.mdx (content now in 06) - Rewrite 08 as chronological narrative: baseline → data curation → algorithmic improvements (SARM, RABC, relative actions, DAgger) → breakthrough with high-quality data → what didn't work → full comparison - Slim 09 Learnings (DAgger now covered in journey chapter) Made-with: Cursor6419f04
-
fix(CLI tools): fixing CLI tools examples (#15)
- fix(CLI tools): fixing CLI tools examples (71ecf4473c4dda425a8a7a80b6d31adad32db3d8) - fix(typos) (ab769eefad04124eb0be067b0c1ba4ea055f889a) - fix(cuda): removing unncessary flag (8cdc2a80a3b9588c6ec19828875abb33bf49e633)7f41624 -
s
Unverified signature1628767 -
Add CLD takeaway discussion and clarify STEP sequential benefit
Unverified signatureMade-with: Cursorc231589 -
feat(video): adding new teaser video (#14)
- feat(video): adding new video (aaf9f25ef8d1bf0ba75bda87d33a5d6cfd6324b9)23d30c8 -
Delete app/src/content/assets/image/Folding_V2.mp4
Verified signature45fb2d2 -
Create Folding_V2.mp4
Verified signature69b5d60 -
fix(build): fixing imports (#13)
- fix(build): fixing imports (e917726c7eb4982c7a2ea745e543ae6ebad05584)b5df06a -
feat(layout): improving overall layout (#12)
- feat(layout): improving overall layout by introducing key concepts early on and concenrtrating code snippets at the end. (500c38713f1cc4a1282af55f8665efcf5b9d96c4)e5a2eaf -
chore(typos): fixing typos, syntax and grammar issues (#10)
- chore(typos, syntax and stuff): fixing typos, syntax and grammatical errors (bb483005abe6db32c7db079097d13e9857cd1a76) - chore(format): making the levels stand out more ! (13f03ee46d8d5c75b412de899dfaa18da1975378) - chore(link): adding link for Evaluation and adapting link for SARM (ab280313ec792497a0eff57b23ab90019e3da8c8) - chore(diverse): fixing divers links, typos and syntax/format issues (074807b572cf438cda49af9309b0e9bbe095f78b) - fix(OpenArm/WowRobo): fix OpenArm/WowRobo naming (f7943426f656060412e61cb8d856d6658323dda4)13cc431 -
Switch to Clopper-Pearson CIs and STEP sequential testing
Unverified signature- Replace Wilson CIs with exact Clopper-Pearson intervals in success rate chart - Switch from Barnard's exact test to STEP (TRI) for pairwise statistical comparisons with Bonferroni correction (n_max=50) - Recompute CLD assignments using STEP - Remove separate "Sort by" controls; auto-sort by selected metric - Add Leandro von Werra to authors and acknowledgments - Link STEP project page and TRI blog throughout Made-with: Cursor32be670 -
Update blog title to be more inviting
Unverified signatureMade-with: Cursor255daaf -
Fix naming: OpenArm, Enactic, WowRobo and clarify manufacturer framing
Unverified signature- "Open Arms" → "OpenArm" throughout - "Enetic" → "Enactic" - "WowRobot" → "WowRobo" - Clarify WowRobo is one of several manufacturers of the Enactic-designed open-source hardware Made-with: Cursorcdf35c1
-
Remove Models & Datasets section from references
Unverified signatureMade-with: Cursordd04a36
-
Remove Trajectory Diversity Grid heading
Unverified signatureMade-with: Cursorf09ee63 -
Rewrite intro, add code snippets for HIL/RTC/relative actions, link repos
Unverified signature- Rewrite hero intro to be more direct and less formulaic - Add code snippets for relative actions training, HIL/DAgger, and eval with RTC - Link Open Arms Mini repo and action representations docs Made-with: Cursorf833fb3
-
Add custom grippers section to hardware chapter
Unverified signatureMade-with: Cursorcdd9e6f -
Add data interpretability call-to-action in learnings section
Unverified signatureMade-with: Cursor2116145 -
Rephrase data collection tip for clarity
Unverified signatureMade-with: Cursor73aff12 -
Clean up action representation diagram
Unverified signatureClearer arrow separation, cleaner legend swatches, chunk brackets. Made-with: Cursore2080a3 -
Add action representation diagram explaining relative vs delta vs absolute
Unverified signatureAdapted from UMI paper (Chi et al., 2024). Shows why relative trajectory avoids error accumulation and calibration requirements. Made-with: Cursor516e6e9 -
Rename delta actions to relative actions throughout
Unverified signatureOpenPI's DeltaActions class implements relative trajectory (UMI terminology), not true delta actions. Each action in the chunk is relative to the current state at prediction time, avoiding error accumulation. Made-with: Cursor9a36854 -
Update app/src/content/chapters/folding/03-hardware.mdx
Verified signature2bde1de -
Fix text colors for dark/light mode in failure analysis and statistical analysis
Unverified signatureMade-with: Cursora72b4f6 -
Fix insight box styling for light mode
Unverified signatureUse neutral grey background with dark text colors for readability on white backgrounds. Made-with: Cursor27447c2 -
Fix colorTo in HF space metadata
Unverified signatureorange is not a valid HF colorTo value, use yellow instead. Made-with: Cursora55899f -
Improve chart clarity and fix visualization issues
Unverified signature- Replace dual-axis L1 Time/Quality chart with bubble scatter plot - Add counts/percentage toggle to failure analysis - Fix label overlaps in heatmap, scatter, and failure charts - Remove misleading series divider lines from sorted charts - Fix total-score description for 50% reference line - Increase margins and use smart label placement throughout Made-with: Cursor5f268a7
-
Robot folding blog post — initial deployment
Unverified signatureMade-with: Cursorfb7e94b
-
Clean repository - remove missing LFS files
6afedde