Fixes: Turn off prevalidation by default, update README with new files - #54
Conversation
Ubiquinone-dot
changed the title
There was a problem hiding this comment.
Pull request overview
This pull request makes two key changes: turns off prevalidation by default in the RFdiffusion3 inference engine configuration, and updates documentation to reflect new files and improve clarity. The documentation restructuring improves the user experience by providing a clearer quick start section and better organization of input specification details.
- Changed default
prevalidate_inputsfromTruetoFalsein the inference engine config - Restructured
input.mddocumentation with improved quick start, better organization, and additional examples - Updated
README.mdwith clarified command examples and new conditioning options diagram
Reviewed changes
Copilot reviewed 3 out of 7 changed files in this pull request and generated 9 comments.
| File | Description |
|---|---|
| models/rfd3/docs/input.md | Major documentation reorganization: moved quick start earlier, added partial diffusion examples, debugging recommendations, and improved FAQ structure |
| models/rfd3/configs/inference_engine/rfdiffusion3.yaml | Changed prevalidate_inputs default from True to False to disable input validation by default |
| models/rfd3/README.md | Updated example command to include prevalidate_inputs=True flag and added conditioning options diagram |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
| // ... args for the second (independent) configuration for design. | ||
| } | ||
| } | ||
| ``` |
There was a problem hiding this comment.
Typo: "unecessary" should be spelled "unnecessary"
| <summary><b>What do the logged bfactors mean?</b></summary> | ||
|
|
||
| You gave input but no contig, unindex, or partial_t. | ||
| The sequence head from RFD3 logs it's confidence for each token in the output structure, you can run `spectrum b` in `pymol` to see it. It usually doesn't mean anything but can give you some idea if the model has gone vastly distribution if the entropy is high (uncertain assignment of sequence). |
There was a problem hiding this comment.
Unclear phrasing: "has gone vastly distribution" should be "has gone vastly off distribution" or "is vastly off distribution"
| The sequence head from RFD3 logs it's confidence for each token in the output structure, you can run `spectrum b` in `pymol` to see it. It usually doesn't mean anything but can give you some idea if the model has gone vastly distribution if the entropy is high (uncertain assignment of sequence). | |
| The sequence head from RFD3 logs it's confidence for each token in the output structure, you can run `spectrum b` in `pymol` to see it. It usually doesn't mean anything but can give you some idea if the model is vastly off distribution if the entropy is high (uncertain assignment of sequence). |
| <details> | ||
| <summary><b>Can I guide on secondary structure?</b></summary> | ||
| Currently no - in future models we may do so, however, you can use `is_non_loopy: true` to make fewer loops. We find this produces a lot more helices and fewer loops (and less sheets). | ||
| </details> |
There was a problem hiding this comment.
Incorrect HTML structure: There's a nested <details> tag inside the parent <details> tag without a proper summary. The nested details tag at line 139 should be removed or the structure should be reorganized so each details section is at the same level.
|
|
||
| --- | ||
| ## InputSpecification fields | ||
|
|
There was a problem hiding this comment.
Typo: "infernece" should be spelled "inference"
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Fixes: Turn off prevalidation by default, update README with new files
No description provided.