Conversation
📝 WalkthroughWalkthroughThe pull request introduces project infrastructure and tooling configurations, including linting/formatting setup via Ruff and mypy, pre-commit hooks, and a project manifest via pyproject.toml. Two new Python scripts are added: analyze_dataset.py for stain analysis workflows and demo.py for running pre-trained stain normalization models. The README is updated with corrected capitalization. Changes
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~25 minutes Poem
🚥 Pre-merge checks | ✅ 1 | ❌ 2❌ Failed checks (1 warning, 1 inconclusive)
✅ Passed checks (1 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches
🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
Summary of ChangesHello, I'm Gemini Code Assist1! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed! This pull request significantly enhances the project's development experience and usability by introducing a user-friendly demonstration for the stain normalization model. It also provides a powerful tool for analyzing and comparing dataset stain characteristics, which is crucial for assessing the model's performance and identifying areas for optimization. Furthermore, the integration of modern code quality tools ensures a consistent and maintainable codebase. Highlights
Changelog
Activity
Using Gemini Code AssistThe full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips. Invoking Gemini You can request assistance from Gemini at any point by creating a comment using either
Customization To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a Limitations & Feedback Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for GitHub and other Google products, sign up here. You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension. Footnotes
|
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
Tento pull request pridáva demo, skript na analýzu a rôzne konfiguračné súbory. Moje pripomienky sa zameriavajú na opravu neplatných verzií v pre-commit-config.yaml, zosúladenie konfigurácie v .ruff.toml s projektom a niekoľko drobných vylepšení v README.md a python skriptoch pre lepšiu robustnosť a výkon.
| - repo: https://github.com/pre-commit/pre-commit-hooks | ||
| rev: v5.0.0 | ||
| hooks: | ||
| - id: check-yaml | ||
| args: [--unsafe] | ||
|
|
||
| - repo: https://github.com/commitizen-tools/commitizen | ||
| rev: v3.30.1 | ||
| hooks: | ||
| - id: commitizen | ||
|
|
||
| - repo: https://github.com/astral-sh/ruff-pre-commit | ||
| rev: v0.7.3 | ||
| hooks: |
There was a problem hiding this comment.
Niekoľko revízií pre pre-commit hooks sa zdá byť nesprávnych alebo odkazujú na neexistujúce/predbežné verzie. To pravdepodobne spôsobí zlyhanie pre-commit alebo jeho neočakávané správanie.
- pre-commit-hooks (riadok 5):
v5.0.0nie je platný tag. Posledná stabilná verzia jev4.6.0. - commitizen (riadok 11):
v3.30.1nie je platný tag. Posledná stabilná verzia jev3.28.0. - ruff-pre-commit (riadok 16):
v0.7.3je predbežná verzia Ruff. Všeobecne sa odporúča používať stabilné verzie (napr.v0.5.0), aby sa predišlo neočakávaným problémom.
Prosím, aktualizujte tieto revízie na platné a stabilné tagy.
| @@ -0,0 +1,51 @@ | |||
| fix = true | |||
| line-length = 88 | |||
| target-version = "py311" | |||
There was a problem hiding this comment.
target-version je nastavená na py311, ale pyproject.toml špecifikuje Python 3.12.5. Aby sa zabezpečilo, že opravy a kontroly Ruff sú kompatibilné s verziou Pythonu vášho projektu, toto by sa malo aktualizovať.
| target-version = "py311" | |
| target-version = "py312" |
| "D", # pydocstyle | ||
| ] | ||
| extend-ignore = [ | ||
| "ERA001", # commented out code |
There was a problem hiding this comment.
V extend-select ste vybrali ERA (kontrola zakomentovaného kódu), ale hneď potom ste ignorovali jeho jediné pravidlo, ERA001. Tým sa kontrola v podstate vypne. Ponechanie zakomentovaného kódu sa vo všeobecnosti neodporúča, pretože zahlcuje kódovú základňu a môže zastarať. Je lepšie spoliehať sa na históriu verzovacieho systému. Zvážte odstránenie tohto riadku, aby ste kontrolu povolili.
| pdm run python demo.py --input ./demo_data/modified | ||
| ``` | ||
|
|
||
| ## Dostupné arguemnty: |
There was a problem hiding this comment.
V slove arguemnty je preklep. Správne má byť argumenty.
| ## Dostupné arguemnty: | |
| ## Dostupné argumenty: |
| ## Dostupné arguemnty: | ||
| - **input**: cesta k obrázku alebo priečinku s obrázkami na normalizáciu (default ./demo_data/modified) | ||
| - **output**: priečinok, kam sa uložia normalizované obrázky (default ./demo_data) | ||
| - **use_cpu**: defaultne nadstavené na použitie GPU ak je dostupná, avšak ak by nastali problémy odporúčam použivať iba CPU No newline at end of file |
There was a problem hiding this comment.
V slove použivať chýba dĺžeň. Správne má byť používať.
| - **use_cpu**: defaultne nadstavené na použitie GPU ak je dostupná, avšak ak by nastali problémy odporúčam použivať iba CPU | |
| - **use_cpu**: defaultne nadstavené na použitie GPU ak je dostupná, avšak ak by nastali problémy odporúčam používať iba CPU |
| comp_iter = iterate_tiles(comp_slides, comp_tiles) | ||
|
|
||
| for (_, orig_tile, image_id), (_, comp_tile, _) in tqdm( | ||
| zip(orig_iter, comp_iter, strict=False), total=len(orig_tiles) |
There was a problem hiding this comment.
Použitie strict=True s zip by tu bolo bezpečnejšie. Komentár na riadku 8 spomína predpoklad, že datasety sú párované. strict=True vyvolá ValueError, ak majú iterovateľné objekty rôzne dĺžky, čo pomáha presadiť tento predpoklad a včas odhaliť potenciálne nezhody v dátach.
| zip(orig_iter, comp_iter, strict=False), total=len(orig_tiles) | |
| zip(orig_iter, comp_iter, strict=True), total=len(orig_tiles) |
| if analyzer is None: | ||
| return |
There was a problem hiding this comment.
Táto kontrola if analyzer is None: sa zdá byť nedosiahnuteľná. Logika vyššie zaisťuje, že sa buď zavolá run_reference_mode alebo run_paired_mode (čo inicializuje analyzer), alebo sa program ukončí cez parser.error(). Tento blok je možné bezpečne odstrániť.
| mean = torch.tensor(self.MEAN).view(3, 1, 1).to(tensor.device) | ||
| std = torch.tensor(self.STD).view(3, 1, 1).to(tensor.device) | ||
| return tensor * std + mean |
There was a problem hiding this comment.
Pre zlepšenie výkonu sa vyhnite vytváraniu nových tenzorov pre mean a std pri každom volaní denormalize. Tieto môžu byť vytvorené raz v metóde __init__, presunuté na správne zariadenie a uložené ako atribúty inštancie (napr. self.mean_tensor, self.std_tensor).
V __init__ pridajte:
self.mean_tensor = torch.tensor(self.MEAN).view(3, 1, 1).to(self.device)
self.std_tensor = torch.tensor(self.STD).view(3, 1, 1).to(self.device)Potom je možné denormalize zjednodušiť.
| mean = torch.tensor(self.MEAN).view(3, 1, 1).to(tensor.device) | |
| std = torch.tensor(self.STD).view(3, 1, 1).to(tensor.device) | |
| return tensor * std + mean | |
| return tensor * self.std_tensor + self.mean_tensor |
| name = "stain-normalization" | ||
| version = "0.1.0" | ||
| authors = [{name = "Adam Lopatka"}] | ||
| requires-python = "==3.12.5" |
There was a problem hiding this comment.
Pevné viazanie verzie Pythonu na konkrétnu patch verziu (==3.12.5) môže byť príliš obmedzujúce a môže spôsobiť problémy prispievateľom používajúcim mierne odlišnú patch verziu. Všeobecne je lepšie špecifikovať kompatibilný rozsah, napríklad requires-python = ">=3.12".
| requires-python = "==3.12.5" | |
| requires-python = ">=3.12" |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 8
🧹 Nitpick comments (1)
demo.py (1)
28-32:torch.load()already defaults toweights_only=Truein PyTorch 2.6+.In PyTorch 2.6+,
torch.load()defaults toweights_only=True, making the explicit parameter redundant for modern versions. However, if the code targets PyTorch 1.13–2.5 compatibility, explicitly settingweights_only=Trueprovides consistent security across versions and helps avoid the arbitrary code execution risk from unpickling untrusted checkpoints.Optional for compatibility
- checkpoint = torch.load(self.CHECKPOINT_PATH, map_location=self.device) + checkpoint = torch.load( + self.CHECKPOINT_PATH, map_location=self.device, weights_only=True + )🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@demo.py` around lines 28 - 32, The torch.load call that reads self.CHECKPOINT_PATH should explicitly pass weights_only=True to ensure consistent behavior and avoid unpickling arbitrary objects on older PyTorch versions; update the torch.load(self.CHECKPOINT_PATH, map_location=self.device) call to include weights_only=True (keeping map_location=self.device) and leave the subsequent checkpoint handling logic (the "state_dict" branch and self.model.load_state_dict calls) unchanged.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In @.gitignore:
- Around line 181-183: The .gitignore entry '/demo_*' incorrectly excludes the
demo_data directory referenced in README.md; update the .gitignore by removing
or narrowing the '/demo_*' pattern and add an explicit allow rule for demo_data
(e.g., add '!/demo_data/' and '!/demo_data/**') so the './demo_data/...' demo
assets can be committed; target the '/demo_*' line in .gitignore and the
README.md mention of './demo_data' to verify consistency.
In @.pre-commit-config.yaml:
- Around line 19-23: The pre-commit hooks replaced Ruff's default entrypoints
with `pdm lint` and `pdm format`, which breaks hooks because `pdm` is not
available in pre-commit's isolated environment; change the `id: ruff` and `id:
ruff-format` hook entries back to Ruff's default executables (`ruff` and
`ruff-format`) and move `--force-exclude` into the hook `args` array so the
flags are passed to Ruff without depending on `pdm` in the hook environment.
In `@analyze_dataset.py`:
- Around line 16-28: Reorder the imports in analyze_dataset.py into
standard-library, third-party, and local/project groups (each group
alphabetized) and ensure single blank lines between groups so Ruff/I001 passes;
specifically place argparse, collections.abc.Generator, pathlib.Path, and
typing.Any first, then third-party imports numpy, pandas, PIL.Image (from PIL
import Image), tqdm, and
rationai.mlkit.data.datasets.MetaTiledSlides/OpenSlideTilesDataset, and finally
local package imports stain_normalization.analysis.StainAnalyzer and
stain_normalization.analysis.report.REPORT_METRICS; run isort/ruff to verify
formatting and remove any unused imports if flagged.
- Around line 90-102: The code silently pairs tiles by iteration order which can
misalign data; instead ensure explicit matching by a unique key and validate
equality before comparing: reindex comp_tiles to orig_tiles using comp_tiles =
comp_tiles.reindex(orig_tiles.index) (or perform a merge/join on the tile
identifier column) and check that no NaNs were introduced and that
len(comp_tiles) == len(orig_tiles) (raise an error or exit if there are
missing/mismatched rows); then create iterators via iterate_tiles(orig_slides,
orig_tiles) and iterate_tiles(comp_slides, comp_tiles) and remove zip(...,
strict=False) so you iterate only aligned rows when calling
analyzer.compare(comp_tile, image_id=image_id, reference=orig_tile).
In `@demo.py`:
- Around line 1-11: The imports in demo.py are unsorted (ruff I001); reorder
them into the configured import groups and alphabetic order (standard library,
third-party, local) so ruff lint passes — i.e., ensure argparse and pathlib.Path
come first, then third-party imports like albumentations (A),
albumentations.pytorch.ToTensorV2, numpy (np), PIL.Image, torch, and finally the
local StainNormalizationModel import; you can run `ruff check --fix demo.py` or
`ruff format demo.py` to apply the correct ordering automatically.
In `@pyproject.toml`:
- Around line 30-33: The package entrypoints 'train', 'validate', 'test', and
'predict' currently run "python -m stain_normalization" but this repo doesn't
contain a stain_normalization module; either add that package/module or update
those scripts in pyproject.toml to point to the real CLI module that exists in
this checkout (replace "stain_normalization" with the correct module name),
ensuring the module exposes the expected mode argument handlers used by the
project.
- Line 11: The pyproject.toml dependency line for "torchmetrics" uses an invalid
constraint "torchmetrics>=1.4.14"; update the constraint to a valid PyPI version
(e.g., "torchmetrics>=1.4.3" or a newer official release) so dependency
resolution succeeds—edit the "torchmetrics" entry in pyproject.toml to replace
">=1.4.14" with a valid version specifier.
In `@README.md`:
- Around line 31-34: Fix the typos in the CLI flags section header and
description: change the header "Dostupné arguemnty:" to "Dostupné argumenty:"
and correct "nadstavené" to "nastavené" in the line describing "use_cpu"; keep
the existing flag names (input, output, use_cpu) and their default values intact
so only spelling is updated.
---
Nitpick comments:
In `@demo.py`:
- Around line 28-32: The torch.load call that reads self.CHECKPOINT_PATH should
explicitly pass weights_only=True to ensure consistent behavior and avoid
unpickling arbitrary objects on older PyTorch versions; update the
torch.load(self.CHECKPOINT_PATH, map_location=self.device) call to include
weights_only=True (keeping map_location=self.device) and leave the subsequent
checkpoint handling logic (the "state_dict" branch and
self.model.load_state_dict calls) unchanged.
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro
Run ID: 4eca22a1-4ca7-4ee5-a739-3f7bfb33de67
⛔ Files ignored due to path filters (2)
pdm.lockis excluded by!**/*.lockuv.lockis excluded by!**/*.lock
📒 Files selected for processing (9)
.flake8.gitignore.mypy.ini.pre-commit-config.yaml.ruff.tomlREADME.mdanalyze_dataset.pydemo.pypyproject.toml
| # Project | ||
| /data | ||
| /demo_* No newline at end of file |
There was a problem hiding this comment.
/demo_* conflicts with the documented demo assets.
README.md tells users to run the demo from ./demo_data/..., but this pattern ignores that directory. That prevents the sample inputs/checkpoint from being added or updated in git, which undercuts the demo flow this PR introduces.
Suggested fix
# Project
/data
/demo_*
+!/demo_data/
+!/demo_data/**🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In @.gitignore around lines 181 - 183, The .gitignore entry '/demo_*'
incorrectly excludes the demo_data directory referenced in README.md; update the
.gitignore by removing or narrowing the '/demo_*' pattern and add an explicit
allow rule for demo_data (e.g., add '!/demo_data/' and '!/demo_data/**') so the
'./demo_data/...' demo assets can be committed; target the '/demo_*' line in
.gitignore and the README.md mention of './demo_data' to verify consistency.
| - id: ruff | ||
| entry: pdm lint --force-exclude | ||
| # Run the formatter. | ||
| - id: ruff-format | ||
| entry: pdm format --force-exclude |
There was a problem hiding this comment.
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash
set -euo pipefail
curl -fsSL https://raw.githubusercontent.com/astral-sh/ruff-pre-commit/v0.7.3/.pre-commit-hooks.yaml \
| sed -n '/^- id: ruff/,+12p;/^- id: ruff-format/,+8p'Repository: RationAI/stain-normalization
Length of output: 653
Keep the Ruff hooks on Ruff's default entrypoint.
Replacing the hook entry with pdm lint and pdm format makes the hook depend on pdm being available inside pre-commit's isolated hook environment, where it typically isn't available. This causes hook execution to fail before Ruff even runs. Use the default Ruff entrypoints and pass tool-specific flags via args instead.
Suggested fix
- repo: https://github.com/astral-sh/ruff-pre-commit
rev: v0.7.3
hooks:
# Run the linter.
- id: ruff
- entry: pdm lint --force-exclude
+ args: [--fix, --force-exclude]
# Run the formatter.
- id: ruff-format
- entry: pdm format --force-exclude
+ args: [--force-exclude]📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
| - id: ruff | |
| entry: pdm lint --force-exclude | |
| # Run the formatter. | |
| - id: ruff-format | |
| entry: pdm format --force-exclude | |
| - id: ruff | |
| args: [--fix, --force-exclude] | |
| # Run the formatter. | |
| - id: ruff-format | |
| args: [--force-exclude] |
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In @.pre-commit-config.yaml around lines 19 - 23, The pre-commit hooks replaced
Ruff's default entrypoints with `pdm lint` and `pdm format`, which breaks hooks
because `pdm` is not available in pre-commit's isolated environment; change the
`id: ruff` and `id: ruff-format` hook entries back to Ruff's default executables
(`ruff` and `ruff-format`) and move `--force-exclude` into the hook `args` array
so the flags are passed to Ruff without depending on `pdm` in the hook
environment.
| import argparse | ||
| from collections.abc import Generator | ||
| from pathlib import Path | ||
| from typing import Any | ||
|
|
||
| import numpy as np | ||
| import pandas as pd | ||
| from PIL import Image | ||
| from rationai.mlkit.data.datasets import MetaTiledSlides, OpenSlideTilesDataset | ||
| from tqdm import tqdm | ||
|
|
||
| from stain_normalization.analysis import StainAnalyzer | ||
| from stain_normalization.analysis.report import REPORT_METRICS |
There was a problem hiding this comment.
Sort this import block so lint passes.
The pipeline is already failing with Ruff I001 on this block, so this script cannot merge cleanly until the imports are re-ordered/formatted.
🧰 Tools
🪛 GitHub Actions: Python Lint (RationAI Standard)
[error] 16-16: I001 Import block is unsorted or un-formatted. Organize imports (ruff).
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@analyze_dataset.py` around lines 16 - 28, Reorder the imports in
analyze_dataset.py into standard-library, third-party, and local/project groups
(each group alphabetized) and ensure single blank lines between groups so
Ruff/I001 passes; specifically place argparse, collections.abc.Generator,
pathlib.Path, and typing.Any first, then third-party imports numpy, pandas,
PIL.Image (from PIL import Image), tqdm, and
rationai.mlkit.data.datasets.MetaTiledSlides/OpenSlideTilesDataset, and finally
local package imports stain_normalization.analysis.StainAnalyzer and
stain_normalization.analysis.report.REPORT_METRICS; run isort/ruff to verify
formatting and remove any unused imports if flagged.
| if args.max_tiles and len(orig_tiles) > args.max_tiles: | ||
| orig_tiles = orig_tiles.sample(n=args.max_tiles, random_state=42) | ||
| comp_tiles = comp_tiles.loc[orig_tiles.index] | ||
| print(f"Subsampled to {args.max_tiles} tile pairs") | ||
|
|
||
| analyzer = StainAnalyzer() | ||
| orig_iter = iterate_tiles(orig_slides, orig_tiles) | ||
| comp_iter = iterate_tiles(comp_slides, comp_tiles) | ||
|
|
||
| for (_, orig_tile, image_id), (_, comp_tile, _) in tqdm( | ||
| zip(orig_iter, comp_iter, strict=False), total=len(orig_tiles) | ||
| ): | ||
| analyzer.compare(comp_tile, image_id=image_id, reference=orig_tile) |
There was a problem hiding this comment.
Don't silently pair tiles by iteration order.
comp_tiles = comp_tiles.loc[orig_tiles.index] and zip(..., strict=False) assume both datasets have identical indexes and ordering. Any reordering or missing row silently truncates or misaligns pairs, which corrupts the reported metrics.
Minimal guard
- for (_, orig_tile, image_id), (_, comp_tile, _) in tqdm(
- zip(orig_iter, comp_iter, strict=False), total=len(orig_tiles)
- ):
+ for (orig_slide, orig_tile, image_id), (
+ comp_slide,
+ comp_tile,
+ comp_image_id,
+ ) in tqdm(zip(orig_iter, comp_iter, strict=True), total=len(orig_tiles)):
+ if (orig_slide, image_id) != (comp_slide, comp_image_id):
+ raise ValueError("Original and compared datasets are not aligned")
analyzer.compare(comp_tile, image_id=image_id, reference=orig_tile)🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@analyze_dataset.py` around lines 90 - 102, The code silently pairs tiles by
iteration order which can misalign data; instead ensure explicit matching by a
unique key and validate equality before comparing: reindex comp_tiles to
orig_tiles using comp_tiles = comp_tiles.reindex(orig_tiles.index) (or perform a
merge/join on the tile identifier column) and check that no NaNs were introduced
and that len(comp_tiles) == len(orig_tiles) (raise an error or exit if there are
missing/mismatched rows); then create iterators via iterate_tiles(orig_slides,
orig_tiles) and iterate_tiles(comp_slides, comp_tiles) and remove zip(...,
strict=False) so you iterate only aligned rows when calling
analyzer.compare(comp_tile, image_id=image_id, reference=orig_tile).
| import argparse | ||
| from pathlib import Path | ||
|
|
||
| import albumentations as A | ||
| import numpy as np | ||
| import torch | ||
| from albumentations.pytorch import ToTensorV2 | ||
| from PIL import Image | ||
|
|
||
| from stain_normalization.stain_normalization_model import StainNormalizationModel | ||
|
|
There was a problem hiding this comment.
Fix import sorting to pass linting.
The pipeline is failing with ruff error I001 indicating imports are unsorted. Run ruff check --fix or ruff format to auto-fix.
Suggested fix
import argparse
from pathlib import Path
import albumentations as A
-import numpy as np
-import torch
from albumentations.pytorch import ToTensorV2
+import numpy as np
from PIL import Image
+import torch
from stain_normalization.stain_normalization_model import StainNormalizationModelNote: The exact ordering depends on your ruff configuration. Running ruff check --fix demo.py will apply the correct order automatically.
🧰 Tools
🪛 GitHub Actions: Python Lint (RationAI Standard)
[error] 1-1: I001 Import block is unsorted or un-formatted. Organize imports (ruff).
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@demo.py` around lines 1 - 11, The imports in demo.py are unsorted (ruff
I001); reorder them into the configured import groups and alphabetic order
(standard library, third-party, local) so ruff lint passes — i.e., ensure
argparse and pathlib.Path come first, then third-party imports like
albumentations (A), albumentations.pytorch.ToTensorV2, numpy (np), PIL.Image,
torch, and finally the local StainNormalizationModel import; you can run `ruff
check --fix demo.py` or `ruff format demo.py` to apply the correct ordering
automatically.
| dependencies = [ | ||
| "lightning>=2.0.0", | ||
| "albumentations>=1.4.14", | ||
| "torchmetrics>=1.4.14", |
There was a problem hiding this comment.
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash
set -euo pipefail
python - <<'PY'
import json
import urllib.request
with urllib.request.urlopen("https://pypi.org/pypi/torchmetrics/json") as resp:
data = json.load(resp)
releases = sorted(data["releases"])
print("1.4.14 exists:", "1.4.14" in data["releases"])
print("Recent 1.4.x releases:", [v for v in releases if v.startswith("1.4")][-10:])
PYRepository: RationAI/stain-normalization
Length of output: 2653
🌐 Web query:
torchmetrics 1.4.14 PyPI release version
💡 Result:
torchmetrics==1.4.14 is not a version published on PyPI. The 1.4.x releases on PyPI are 1.4.0, 1.4.0.post0, 1.4.1, 1.4.2, and 1.4.3. The current latest release shown on PyPI is 1.9.0 (Mar 9, 2026). [1]
Source(s):
- [1] PyPI “torchmetrics” release history (shows available versions and dates)
Fix torchmetrics version constraint—1.4.14 does not exist on PyPI.
The constraint torchmetrics>=1.4.14 is unsatisfiable. PyPI only has torchmetrics 1.4.0, 1.4.0.post0, 1.4.1, 1.4.2, and 1.4.3 in the 1.4.x series. This will cause pdm install to fail during dependency resolution. Change this to a valid version constraint, such as torchmetrics>=1.4.3 or update to a more recent release.
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@pyproject.toml` at line 11, The pyproject.toml dependency line for
"torchmetrics" uses an invalid constraint "torchmetrics>=1.4.14"; update the
constraint to a valid PyPI version (e.g., "torchmetrics>=1.4.3" or a newer
official release) so dependency resolution succeeds—edit the "torchmetrics"
entry in pyproject.toml to replace ">=1.4.14" with a valid version specifier.
| train = "python -m stain_normalization mode=fit" | ||
| validate = "python -m stain_normalization mode=validate" | ||
| test = "python -m stain_normalization mode=test" | ||
| predict = "python -m stain_normalization mode=predict" |
There was a problem hiding this comment.
These scripts point at a module the checkout doesn't contain.
train / validate / test / predict all call python -m stain_normalization, but this PR doesn't add that package. In this state those entrypoints fail before any app code runs.
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@pyproject.toml` around lines 30 - 33, The package entrypoints 'train',
'validate', 'test', and 'predict' currently run "python -m stain_normalization"
but this repo doesn't contain a stain_normalization module; either add that
package/module or update those scripts in pyproject.toml to point to the real
CLI module that exists in this checkout (replace "stain_normalization" with the
correct module name), ensuring the module exposes the expected mode argument
handlers used by the project.
| ## Dostupné arguemnty: | ||
| - **input**: cesta k obrázku alebo priečinku s obrázkami na normalizáciu (default ./demo_data/modified) | ||
| - **output**: priečinok, kam sa uložia normalizované obrázky (default ./demo_data) | ||
| - **use_cpu**: defaultne nadstavené na použitie GPU ak je dostupná, avšak ak by nastali problémy odporúčam použivať iba CPU No newline at end of file |
There was a problem hiding this comment.
Fix the CLI section typos.
arguemnty and nadstavené are user-facing typos in the only section that explains the flags, so the README reads unfinished right where users are expected to follow it.
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@README.md` around lines 31 - 34, Fix the typos in the CLI flags section
header and description: change the header "Dostupné arguemnty:" to "Dostupné
argumenty:" and correct "nadstavené" to "nastavené" in the line describing
"use_cpu"; keep the existing flag names (input, output, use_cpu) and their
default values intact so only spelling is updated.
1 participant
The rest includes a demo+README for BC and analyze_dataset for easier datasets analysis, to find out whether the optimization is actually needed and how much the model helps.
Summary by CodeRabbit
New Features
Chores
Documentation