Remove data/ research scripts#9
Merged
Merged
Conversation
The data/ dir was 49 model-training/labeling/prototype one-off scripts (how the Orange models were built) — not used by the package or tests. Their narrative lives in work.log. - block_chunker.py duplicated pulpie.chunker; the 3 eval scripts now import from pulpie.chunker directly (DATA_DIR kept for runtime dataset/ckpt paths) - chunk_page (block_chunker's only extra fn) was unused by eval - gitignore: data/ is now a fully-ignored runtime dir (GPU-box datasets/ckpts) No package code touched; 65 parity tests still pass.
There was a problem hiding this comment.
Pull request overview
This PR removes the repository’s data/ research / training / benchmarking scripts and artifacts (one-off model-building pipeline), while keeping the historical narrative in work.log. It also updates a small set of evaluation scripts to use the canonical chunking utilities from pulpie.chunker, and adjusts .gitignore so data/ becomes a fully runtime-only directory.
Changes:
- Delete the
data/directory contents (training, labeling, prototypes, benchmarks, and metadata files) that are not used by the package/tests. - Update eval scripts to import chunking helpers from
pulpie.chunkerinstead ofdata/block_chunker.py. - Simplify
.gitignoreto ignoredata/entirely as an untracked runtime directory.
Reviewed changes
Copilot reviewed 52 out of 53 changed files in this pull request and generated no comments.
Show a summary per file
| File | Description |
|---|---|
| eval/eval_latte_large_vs_dripper.py | Switch chunking import to pulpie.chunker (removes reliance on data/ on sys.path). |
| eval/eval_all_methods_200.py | Switch chunking import to pulpie.chunker (removes reliance on data/ on sys.path). |
| eval/bench_all_methods.py | Switch chunking import to pulpie.chunker inside run_pulpie(). |
| .gitignore | Ignore data/ as a runtime-only directory (no longer tracks any data/* files). |
| data/tune_gbm.py | Remove unused one-off GBM tuning script. |
| data/train_test_split.py | Remove unused split-eval training script. |
| data/train_potion_classifier.py | Remove unused potion classifier training script. |
| data/train_gbm.py | Remove unused GBM training/feature-selection script. |
| data/train_gbm_v3_cc.py | Remove unused fast CC model training script. |
| data/train_gbm_v2.py | Remove unused GBM v2 training/tuning script. |
| data/train_final.py | Remove unused final-model training script. |
| data/train_combined.py | Remove unused combined-dataset training script. |
| data/test_llm_classifier.py | Remove unused LLM labeling experiment script. |
| data/test_llm_classifier_v2.py | Remove unused LLM labeling experiment (v2) script. |
| data/test_ensemble.py | Remove unused GBM+LLM ensemble experiment script. |
| data/selected_features.json | Remove model-training metadata artifact (no longer tracked). |
| data/sample_cc.py | Remove unused Common Crawl sampling script. |
| data/run_label.sh | Remove unused labeling runner shell script. |
| data/relabel_truncated.py | Remove unused relabeling utility script. |
| data/prototype_text_classifier.py | Remove unused prototype text-model comparison script. |
| data/prototype_crf.py | Remove unused CRF prototype script. |
| data/prototype_crf_v2.py | Remove unused CRF prototype (v2) script. |
| data/make_wmb_eval_sample.py | Remove unused eval-sample generation script. |
| data/label_with_dripper.py | Remove unused Dripper labeling script. |
| data/label_cc.py | Remove unused Bedrock-based CC labeling script. |
| data/generate_training_data.py | Remove unused WMB training-data generation script. |
| data/generate_training_data_dom.py | Remove unused DOM-based training-data generation script. |
| data/generate_cc_training_data.py | Remove unused CC training-data generation script. |
| data/filter_and_audit_cc.py | Remove unused CC filtering/audit preparation script. |
| data/estimate_label_noise.py | Remove unused label-noise estimation script. |
| data/download_warcs.py | Remove unused WARC download helper script. |
| data/cc_relabel_urls.json | Remove unused relabel URL list artifact. |
| data/cascade_analysis.py | Remove unused cascade analysis script. |
| data/block_chunker.py | Remove duplicated chunker implementation (superseded by pulpie.chunker). |
| data/bench_throughput_real.py | Remove unused throughput benchmark script. |
| data/bench_tensorrt.py | Remove unused TensorRT benchmark script. |
| data/bench_onnx_cpu.py | Remove unused ONNX CPU benchmark script. |
| data/bench_modernbert.py | Remove unused ModernBERT vs EuroBERT benchmark script. |
| data/bench_gpu.py | Remove unused GPU microbenchmark script. |
| data/bench_gpu_local.py | Remove unused local-checkpoint GPU benchmark script. |
| data/bench_gpu_compile.py | Remove unused torch.compile GPU benchmark script. |
| data/audit_labels_llm.py | Remove unused label-auditing script. |
| data/analyze_errors.py | Remove unused error analysis script. |
| data/add_embeddings.py | Remove unused embeddings augmentation script. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
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
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.
Prunes the
data/directory — 49 model-training / labeling / prototype / benchmark one-off scripts that built the Orange models. Not used by the package or tests; the narrative is preserved inwork.log.block_chunker.pyduplicatedpulpie.chunker; the 3 eval scripts that imported it now usefrom pulpie.chunker import ...(theirDATA_DIRis kept — it points at the gitignored runtime dataset/checkpoint location on GPU boxes)..gitignore:data/is now a fully-ignored runtime dir.~12.4K lines removed.