docs: fix the tutorial's checksum, install line and add a scale note - #6
Merged
Conversation
- The MD5 check read the whole ~1 GB slide into memory with `fh.read()`, spiking RSS by more than the model uses. Hash in 8 MB chunks instead. - The install line asked for `deepspotm pyvips matplotlib`, which drifts from the `wsi` extra the README documents. Use `deepspotm[wsi]` so there is one documented way to install the whole-slide dependencies. - Say up front what "a whole slide" means in practice, using the figures from the committed run: 99,960 x 65,991 px, a 446 x 294 grid, 45,751 candidate tiles, 36,392 scored after background filtering. Readers currently have no idea of the scale until they are several cells in. Cell outputs, execution counts and the source line-array formatting are unchanged, so the notebook still renders and diffs cleanly. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Three small fixes to
examples/predict_tcga_skcm.ipynb.Checksum read the whole slide into memory.
hashlib.md5(fh.read())on a ~1 GB.svsspikes RSS by more than the model itself uses, on a machine that is about to load a foundation model onto the GPU. Now hashed in 8 MB chunks. Verified the chunked digest matches a whole-file digest.Install line drifted from the README. The notebook asked for
pip install deepspotm pyvips matplotlib; the README documents awsiextra. Nowpip install "deepspotm[wsi]" matplotlib, so there is one documented way to get the whole-slide dependencies.No sense of scale up front. Readers had no idea what "a whole slide" costs until several cells in. Added the figures from the committed run: 99,960 x 65,991 px, a 446 x 294 tile grid, 45,751 candidate tiles from the thumbnail pass, 36,392 scored after the exact background filter. No timings, since the committed outputs do not record them.
Verification
nbformat.validatepasses.sourcestays a line array in every cell, so the notebook still diffs and renders normally.Docs only; no code, packaging or CI changes.
🤖 Generated with Claude Code