diff --git a/SHA256SUMS b/SHA256SUMS index f1e6cc9b07c586d077ad8bc27c96f3d848ecca52..40cef93bb87d8f3eb3f6d52c86eb527f2db06896 100644 --- a/SHA256SUMS +++ b/SHA256SUMS @@ -114,14 +114,14 @@ c68cd003370b2dcdb162f848f766e4e62f2653c6c38d205f8cbe53a9aabe2d74 data/vocab/bpe 044030f212648a98fc477352328651c284d185660d9a2a7116f92f73f97e1ae5 data/vocab/confidence_analysis.json ff7bb6a3de636dbee79846ef8083bdab004e9b79cd60715fbe1474d482e119c8 data/vocab/ms_vocabulary.json f315117a51a49e4019812367e72423620c007c9c73bf6e19f357fe4ca395de3a data/vocab/vocabulary.json -736a696448cc1c4dd8f10ae41ba5549c8998e919663fec4590742369d5ca399a provenance/CORPUS_PIPELINE_AND_TOKENIZER_PROVENANCE_2026-04-19.md +820ecfca39cd6311e0499b3c4bef36708ae6cb6e536d86a629e524515ad17d66 provenance/CORPUS_PIPELINE_AND_TOKENIZER_PROVENANCE_2026-04-19.md 670fa0cdd66ca2b21132545a196924723b6ec0943df0bff7e7540731564a7eed provenance/SOURCE_LOG.md b4b19a7d28f543f36bbca2b3ab5444bdafc01e0b173c57faccc79f8bf3e78d2e provenance/compute_provenance/CLUSTER_VERIFIED_MODEL_LINEAGE_20260419.md -6268604d64f62f4e9564f36760fd4c4dc7fa834470cf369b8ee224b153591550 provenance/compute_provenance/COMPUTE_PROVENANCE_LOG.md +0eadc9e9c8e1354e9b5f696f6d80b7257e8e7ed2d4280b389d7bfd1cf46fc750 provenance/compute_provenance/COMPUTE_PROVENANCE_LOG.md 445408075a619e10f1511cbf13d6540f020fe5a238a11f05e83e34adeecea096 provenance/compute_provenance/EXA_SCALABILITY_MAIN_TEXT_FILL_20260419.md -39ab3e327829a873b412baf1df3f5b9713d9923e7ad644d7097677adc2493ec1 provenance/compute_provenance/compute_runtime_summary.csv -5ac925a6e79d2dd3f7c81397ded1b60cab2a251836e74e86e1605f8da1425de7 provenance/compute_provenance/compute_runtime_summary.json -732dc9ee6174342fefcec73bdf7f812e3abac8b71ecb24633672c3e355d131f1 provenance/compute_provenance/launch_script_summary.csv +c762eea25d0e7537363a02e8bb4f5b0d6eea7e558b76574731b536b32121060b provenance/compute_provenance/compute_runtime_summary.csv +838301cb054463d81af4488a9f3e7d8eec13fcf7e860665202f701711458efd2 provenance/compute_provenance/compute_runtime_summary.json +b3882871574b41505c6f46fdc7dbdc716249713e9173715649d591f897dc564d provenance/compute_provenance/launch_script_summary.csv ddc06ca59280bc75410aefa73816e9a6809e7720a6ec3b43198fa99c503d88c1 provenance/compute_provenance/multimodal_config_v5b_cluster_copy.yaml 04d9c188cbd8a2a7bf68349967a0472f33a5f9ee37c7de1f867eec5120979bcd provenance/compute_provenance/run_train_v5b_cluster_verified_copy_20260201.sh 59adc26c35d55c259411d5bd49177bfce0f4269d96184d5c4dbb1f334fc54bf2 provenance/compute_provenance/step3_contrastive_train_cluster_verified_copy_20260130.sh diff --git a/code/benchmarks/__pycache__/calculate_rankings.cpython-312.pyc b/code/benchmarks/__pycache__/calculate_rankings.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..95eaa7c14170f2b820872583dec1ecc8893f9ed9 Binary files /dev/null and b/code/benchmarks/__pycache__/calculate_rankings.cpython-312.pyc differ diff --git a/code/benchmarks/__pycache__/consolidate_results.cpython-312.pyc b/code/benchmarks/__pycache__/consolidate_results.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..cd3562c9271a4319e1bba37e53dbfaa6f1f81aac Binary files /dev/null and b/code/benchmarks/__pycache__/consolidate_results.cpython-312.pyc differ diff --git a/code/benchmarks/__pycache__/create_excluded_dataset.cpython-312.pyc b/code/benchmarks/__pycache__/create_excluded_dataset.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..c23da252eb4ddd78a2176fbf6a242daa0ee7b286 Binary files /dev/null and b/code/benchmarks/__pycache__/create_excluded_dataset.cpython-312.pyc differ diff --git a/code/benchmarks/__pycache__/extract_all_benchmark_results.cpython-312.pyc b/code/benchmarks/__pycache__/extract_all_benchmark_results.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..9d200d18882478a988ff22dc4fc09496fa0cfe4b Binary files /dev/null and b/code/benchmarks/__pycache__/extract_all_benchmark_results.cpython-312.pyc differ diff --git a/code/benchmarks/__pycache__/finetune.cpython-312.pyc b/code/benchmarks/__pycache__/finetune.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..cd466721c1ab3fe24a9a9fa09f1e51467e595632 Binary files /dev/null and b/code/benchmarks/__pycache__/finetune.cpython-312.pyc differ diff --git a/code/bertint/__pycache__/bertint_v8.cpython-312.pyc b/code/bertint/__pycache__/bertint_v8.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..b6cf9ef1cfbc26f76fbff75245aeff9893ee24e8 Binary files /dev/null and b/code/bertint/__pycache__/bertint_v8.cpython-312.pyc differ diff --git a/code/bertint/__pycache__/bertint_v8_inference.cpython-312.pyc b/code/bertint/__pycache__/bertint_v8_inference.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..6331216ec6edf710b2fd70b7471f4dd89dcdd8dd Binary files /dev/null and b/code/bertint/__pycache__/bertint_v8_inference.cpython-312.pyc differ diff --git a/code/bertint/__pycache__/build_combined_dataset.cpython-312.pyc b/code/bertint/__pycache__/build_combined_dataset.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..688fd6cb573a73e46a9bfeb9949f02b5d6618037 Binary files /dev/null and b/code/bertint/__pycache__/build_combined_dataset.cpython-312.pyc differ diff --git a/code/bertint/__pycache__/compare_ablation.cpython-312.pyc b/code/bertint/__pycache__/compare_ablation.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..10b94fa8f5a27298441031d7b2a8434d7b6deff5 Binary files /dev/null and b/code/bertint/__pycache__/compare_ablation.cpython-312.pyc differ diff --git a/code/bertint/__pycache__/dataset_v8.cpython-312.pyc b/code/bertint/__pycache__/dataset_v8.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..400f5017c9374ef19ef665ab6624fecb572447a6 Binary files /dev/null and b/code/bertint/__pycache__/dataset_v8.cpython-312.pyc differ diff --git a/code/bertint/__pycache__/generate_glycan_splits.cpython-312.pyc b/code/bertint/__pycache__/generate_glycan_splits.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..8c3da4c694c5642d244ef0b4cd8b169abef30862 Binary files /dev/null and b/code/bertint/__pycache__/generate_glycan_splits.cpython-312.pyc differ diff --git a/code/bertint/__pycache__/training_v8.cpython-312.pyc b/code/bertint/__pycache__/training_v8.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..2a0b6e5b2e64fc22cd03da227488d102c7280b28 Binary files /dev/null and b/code/bertint/__pycache__/training_v8.cpython-312.pyc differ diff --git a/code/contrastive/__pycache__/contrastive_trainer_v51_FINAL.cpython-312.pyc b/code/contrastive/__pycache__/contrastive_trainer_v51_FINAL.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..018cc647dc1e842563cfd7ceada9ee0ca53c91c5 Binary files /dev/null and b/code/contrastive/__pycache__/contrastive_trainer_v51_FINAL.cpython-312.pyc differ diff --git a/code/contrastive/__pycache__/contrastive_trainer_v5_SIMPLE.cpython-312.pyc b/code/contrastive/__pycache__/contrastive_trainer_v5_SIMPLE.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..2c41153d6ff2893954361a15506c9171c47d83ac Binary files /dev/null and b/code/contrastive/__pycache__/contrastive_trainer_v5_SIMPLE.cpython-312.pyc differ diff --git a/code/contrastive/__pycache__/extract_resolved_glycans.cpython-312.pyc b/code/contrastive/__pycache__/extract_resolved_glycans.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..5468ef56740bdbcefba12ea3421a85ba62c00c1a Binary files /dev/null and b/code/contrastive/__pycache__/extract_resolved_glycans.cpython-312.pyc differ diff --git a/code/contrastive/__pycache__/generate_negatives_v4_FINAL.cpython-312.pyc b/code/contrastive/__pycache__/generate_negatives_v4_FINAL.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..7bc687889f4d63db1d82520c0fc0581f49ec398e Binary files /dev/null and b/code/contrastive/__pycache__/generate_negatives_v4_FINAL.cpython-312.pyc differ diff --git a/code/contrastive/__pycache__/mcns_filter_75k_v3_FINAL.cpython-312.pyc b/code/contrastive/__pycache__/mcns_filter_75k_v3_FINAL.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..fa311f9ac9fcb908fd17c3e33bb37526c8860a45 Binary files /dev/null and b/code/contrastive/__pycache__/mcns_filter_75k_v3_FINAL.cpython-312.pyc differ diff --git a/code/contrastive/__pycache__/score_negative_difficulty.cpython-312.pyc b/code/contrastive/__pycache__/score_negative_difficulty.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..a0d8b568fc2c3e004d143bccc4768b3c29a03895 Binary files /dev/null and b/code/contrastive/__pycache__/score_negative_difficulty.cpython-312.pyc differ diff --git a/code/data_processing/__pycache__/augment_pretraining_data.cpython-312.pyc b/code/data_processing/__pycache__/augment_pretraining_data.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..f13b37032a5fb7ed41683e9fc4262248b1986b93 Binary files /dev/null and b/code/data_processing/__pycache__/augment_pretraining_data.cpython-312.pyc differ diff --git a/code/data_processing/__pycache__/convert_iupac_glycosmos.cpython-312.pyc b/code/data_processing/__pycache__/convert_iupac_glycosmos.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..ffaf6794406cb31e7bc655cb00cd502bca7ed7bf Binary files /dev/null and b/code/data_processing/__pycache__/convert_iupac_glycosmos.cpython-312.pyc differ diff --git a/code/downstream_tasks/__pycache__/__init__.cpython-312.pyc b/code/downstream_tasks/__pycache__/__init__.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..82bfea973a43c1dab401d6de2e40bce7657bc992 Binary files /dev/null and b/code/downstream_tasks/__pycache__/__init__.cpython-312.pyc differ diff --git a/code/downstream_tasks/utils/__pycache__/__init__.cpython-312.pyc b/code/downstream_tasks/utils/__pycache__/__init__.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..32ec4fe5de46faf33a863e7cace09c29927b7721 Binary files /dev/null and b/code/downstream_tasks/utils/__pycache__/__init__.cpython-312.pyc differ diff --git a/code/downstream_tasks/utils/__pycache__/baseline_results.cpython-312.pyc b/code/downstream_tasks/utils/__pycache__/baseline_results.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..e4eb53d4e10da481629ee825744ef3d54f14f461 Binary files /dev/null and b/code/downstream_tasks/utils/__pycache__/baseline_results.cpython-312.pyc differ diff --git a/code/downstream_tasks/utils/__pycache__/dataset.cpython-312.pyc b/code/downstream_tasks/utils/__pycache__/dataset.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..f95774f108709ee2bb20bcdf43ffde62acbbd8a0 Binary files /dev/null and b/code/downstream_tasks/utils/__pycache__/dataset.cpython-312.pyc differ diff --git a/code/downstream_tasks/utils/__pycache__/tokenizer.cpython-312.pyc b/code/downstream_tasks/utils/__pycache__/tokenizer.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..3ebf66c8a785ccb5d367431957226870eec9a62b Binary files /dev/null and b/code/downstream_tasks/utils/__pycache__/tokenizer.cpython-312.pyc differ diff --git a/code/downstream_tasks/utils/__pycache__/wurcs_bpe_tokenizer.cpython-312.pyc b/code/downstream_tasks/utils/__pycache__/wurcs_bpe_tokenizer.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..7e7bb3cad2a079210c91bff36288e881b17b3b26 Binary files /dev/null and b/code/downstream_tasks/utils/__pycache__/wurcs_bpe_tokenizer.cpython-312.pyc differ diff --git a/code/model/__pycache__/__init__.cpython-312.pyc b/code/model/__pycache__/__init__.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..07e20b96db65ac7d6dbd39af4cabfb39dc842014 Binary files /dev/null and b/code/model/__pycache__/__init__.cpython-312.pyc differ diff --git a/code/model/__pycache__/dataset.cpython-312.pyc b/code/model/__pycache__/dataset.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..c2b0e14e43b7ac885c36c7d1cd366ed8023cba96 Binary files /dev/null and b/code/model/__pycache__/dataset.cpython-312.pyc differ diff --git a/code/model/__pycache__/enhanced_classifier.cpython-312.pyc b/code/model/__pycache__/enhanced_classifier.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..8d4d9647046106c3c56cdab608526163884530b5 Binary files /dev/null and b/code/model/__pycache__/enhanced_classifier.cpython-312.pyc differ diff --git a/code/model/__pycache__/glycan_bert.cpython-312.pyc b/code/model/__pycache__/glycan_bert.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..1eeb316ac99c041c2ccaea92f1149b434cb7d25a Binary files /dev/null and b/code/model/__pycache__/glycan_bert.cpython-312.pyc differ diff --git a/code/model/__pycache__/hierarchical_glycan.cpython-312.pyc b/code/model/__pycache__/hierarchical_glycan.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..e57c50d1e91a377e9f82136e49ade30b302e45d5 Binary files /dev/null and b/code/model/__pycache__/hierarchical_glycan.cpython-312.pyc differ diff --git a/code/model/__pycache__/multimodal_glycan_bert_v3.cpython-312.pyc b/code/model/__pycache__/multimodal_glycan_bert_v3.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..01d7478ebadf3b7a355a9c0f8fc6641cb274e234 Binary files /dev/null and b/code/model/__pycache__/multimodal_glycan_bert_v3.cpython-312.pyc differ diff --git a/code/model/__pycache__/tokenizer.cpython-312.pyc b/code/model/__pycache__/tokenizer.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..1fc739e1ae3b2fd33b73ae3cbf9c17c163515206 Binary files /dev/null and b/code/model/__pycache__/tokenizer.cpython-312.pyc differ diff --git a/code/model/__pycache__/wurcs_bpe_tokenizer.cpython-312.pyc b/code/model/__pycache__/wurcs_bpe_tokenizer.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..dcac97783c0f2818a1d6fc0b2e1029eb32d3cc6f Binary files /dev/null and b/code/model/__pycache__/wurcs_bpe_tokenizer.cpython-312.pyc differ diff --git a/code/probes/__pycache__/analyze_core_biology.cpython-312.pyc b/code/probes/__pycache__/analyze_core_biology.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..64ee5c50c60b57685080b290f95993a275b14b56 Binary files /dev/null and b/code/probes/__pycache__/analyze_core_biology.cpython-312.pyc differ diff --git a/code/probes/__pycache__/analyze_embeddings.cpython-312.pyc b/code/probes/__pycache__/analyze_embeddings.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..9a24864bd3656684a84ee38e151118f247c3f041 Binary files /dev/null and b/code/probes/__pycache__/analyze_embeddings.cpython-312.pyc differ diff --git a/code/probes/__pycache__/embed_benchmark_tasks.cpython-312.pyc b/code/probes/__pycache__/embed_benchmark_tasks.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..aea055282274cfcd85bd36630edfc191cb8d96a9 Binary files /dev/null and b/code/probes/__pycache__/embed_benchmark_tasks.cpython-312.pyc differ diff --git a/code/probes/__pycache__/extract_embeddings.cpython-312.pyc b/code/probes/__pycache__/extract_embeddings.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..b61e8b6fc28e4d767161241d451b959fb3007133 Binary files /dev/null and b/code/probes/__pycache__/extract_embeddings.cpython-312.pyc differ diff --git a/code/probes/__pycache__/novel_probes_v2.cpython-312.pyc b/code/probes/__pycache__/novel_probes_v2.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..4552e2d85dc43b2c392c936ba3c683e505ce7850 Binary files /dev/null and b/code/probes/__pycache__/novel_probes_v2.cpython-312.pyc differ diff --git a/code/probes/__pycache__/probe_01_tsne_clustering.cpython-312.pyc b/code/probes/__pycache__/probe_01_tsne_clustering.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..842f86d3afb0b7b9597f0549f4c127bd25aedc0f Binary files /dev/null and b/code/probes/__pycache__/probe_01_tsne_clustering.cpython-312.pyc differ diff --git a/code/probes/__pycache__/probe_10_lectin_motifs.cpython-312.pyc b/code/probes/__pycache__/probe_10_lectin_motifs.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..c28614e56d67ef537201396b7ce8bdb738f58523 Binary files /dev/null and b/code/probes/__pycache__/probe_10_lectin_motifs.cpython-312.pyc differ diff --git a/code/probes/__pycache__/probe_10_tissue_presence.cpython-312.pyc b/code/probes/__pycache__/probe_10_tissue_presence.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..06307aae77f67f282e374c672e5c17f104b1cd69 Binary files /dev/null and b/code/probes/__pycache__/probe_10_tissue_presence.cpython-312.pyc differ diff --git a/code/probes/__pycache__/probe_14_retrieval.cpython-312.pyc b/code/probes/__pycache__/probe_14_retrieval.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..28f9d9ccbc067fc98afcf4722fc379c4f29d9ec5 Binary files /dev/null and b/code/probes/__pycache__/probe_14_retrieval.cpython-312.pyc differ diff --git a/code/probes/__pycache__/probe_15_seal_milk_ood.cpython-312.pyc b/code/probes/__pycache__/probe_15_seal_milk_ood.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..fed859fa159df618b56693b1f5e643250853c1f8 Binary files /dev/null and b/code/probes/__pycache__/probe_15_seal_milk_ood.cpython-312.pyc differ diff --git a/code/probes/__pycache__/probe_15h_phylogenetic.cpython-312.pyc b/code/probes/__pycache__/probe_15h_phylogenetic.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..995150e71f5afa08de4ed8bfd377b13d2f424d26 Binary files /dev/null and b/code/probes/__pycache__/probe_15h_phylogenetic.cpython-312.pyc differ diff --git a/code/probes/__pycache__/probe_2_layerwise_cls.cpython-312.pyc b/code/probes/__pycache__/probe_2_layerwise_cls.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..d5f2c890c5301c28e4e996dc8415c2bbde74ad4a Binary files /dev/null and b/code/probes/__pycache__/probe_2_layerwise_cls.cpython-312.pyc differ diff --git a/code/probes/__pycache__/probe_3_4_attention_structure.cpython-312.pyc b/code/probes/__pycache__/probe_3_4_attention_structure.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..4134af74661e3b5a6d9b43a82ab3fee1072f5d8f Binary files /dev/null and b/code/probes/__pycache__/probe_3_4_attention_structure.cpython-312.pyc differ diff --git a/code/probes/__pycache__/probe_7_contrastive_ablation.cpython-312.pyc b/code/probes/__pycache__/probe_7_contrastive_ablation.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..5a84807e93b70563e44a5d0386ec83556a0b3596 Binary files /dev/null and b/code/probes/__pycache__/probe_7_contrastive_ablation.cpython-312.pyc differ diff --git a/code/probes/__pycache__/probe_8_disease_tissue.cpython-312.pyc b/code/probes/__pycache__/probe_8_disease_tissue.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..43647e34cd04e7f822a5e22561c1ea0c66a02318 Binary files /dev/null and b/code/probes/__pycache__/probe_8_disease_tissue.cpython-312.pyc differ diff --git a/code/probes/__pycache__/probe_9_biosynthesis_order.cpython-312.pyc b/code/probes/__pycache__/probe_9_biosynthesis_order.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..5c24fbdc4457735b737df7c9dcde9af420f50b54 Binary files /dev/null and b/code/probes/__pycache__/probe_9_biosynthesis_order.cpython-312.pyc differ diff --git a/code/probes/__pycache__/probe_exploratory_biology.cpython-312.pyc b/code/probes/__pycache__/probe_exploratory_biology.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..db635c8cf7713e3817935587f7524b852b9c0ae0 Binary files /dev/null and b/code/probes/__pycache__/probe_exploratory_biology.cpython-312.pyc differ diff --git a/code/tokenizer/__pycache__/generate_bpe_ambiguity.cpython-312.pyc b/code/tokenizer/__pycache__/generate_bpe_ambiguity.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..356e96ccbb122f6b9e99e7f52198b15bff90a96f Binary files /dev/null and b/code/tokenizer/__pycache__/generate_bpe_ambiguity.cpython-312.pyc differ diff --git a/code/training/__pycache__/__init__.cpython-312.pyc b/code/training/__pycache__/__init__.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..18d6bc51bb21b1b9d8478990530216be4567754b Binary files /dev/null and b/code/training/__pycache__/__init__.cpython-312.pyc differ diff --git a/code/training/__pycache__/analyze_confidence.cpython-312.pyc b/code/training/__pycache__/analyze_confidence.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..464c1c20e582b17a1c08c3b6a73fb4f59792f3f7 Binary files /dev/null and b/code/training/__pycache__/analyze_confidence.cpython-312.pyc differ diff --git a/code/training/__pycache__/ipa_bpe_distillation.cpython-312.pyc b/code/training/__pycache__/ipa_bpe_distillation.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..b14a2b2d001947032ed3e2750caf82df4755a15b Binary files /dev/null and b/code/training/__pycache__/ipa_bpe_distillation.cpython-312.pyc differ diff --git a/code/training/__pycache__/masking.cpython-312.pyc b/code/training/__pycache__/masking.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..d2018a50bbb3e9da09b38b6b736e1be4d9b21730 Binary files /dev/null and b/code/training/__pycache__/masking.cpython-312.pyc differ diff --git a/code/training/__pycache__/multimodal_dataset.cpython-312.pyc b/code/training/__pycache__/multimodal_dataset.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..2a36592c846f05f7bb1f67237a72b2935a678652 Binary files /dev/null and b/code/training/__pycache__/multimodal_dataset.cpython-312.pyc differ diff --git a/code/training/__pycache__/multimodal_masking.cpython-312.pyc b/code/training/__pycache__/multimodal_masking.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..3a0c469f0f10b1063b72820b7e03c49d68e6d285 Binary files /dev/null and b/code/training/__pycache__/multimodal_masking.cpython-312.pyc differ diff --git a/code/training/__pycache__/train_multimodal.cpython-312.pyc b/code/training/__pycache__/train_multimodal.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..1c6a785be3c9e49e5229943b0bedeb971e4734cb Binary files /dev/null and b/code/training/__pycache__/train_multimodal.cpython-312.pyc differ diff --git a/code/training/__pycache__/train_wurcs_bpe.cpython-312.pyc b/code/training/__pycache__/train_wurcs_bpe.cpython-312.pyc new file mode 100644 index 0000000000000000000000000000000000000000..369e5c512d5a6fe493f92928a6ee5c641f963ffd Binary files /dev/null and b/code/training/__pycache__/train_wurcs_bpe.cpython-312.pyc differ diff --git a/provenance/CORPUS_PIPELINE_AND_TOKENIZER_PROVENANCE_2026-04-19.md b/provenance/CORPUS_PIPELINE_AND_TOKENIZER_PROVENANCE_2026-04-19.md index 2576e86e1a0b2f3b07211e1e818d97f3115dd1bf..0ee08fba672cf8f45ec376b3373526ebb9ac6c8b 100644 --- a/provenance/CORPUS_PIPELINE_AND_TOKENIZER_PROVENANCE_2026-04-19.md +++ b/provenance/CORPUS_PIPELINE_AND_TOKENIZER_PROVENANCE_2026-04-19.md @@ -45,7 +45,7 @@ No checked current or older local corpus artifact reproduced a BPE median of `26 ### Stage 1 | Raw source accounting for the current manuscript -From [table_s1_corpus_source_breakdown.csv](/Users/supantha/Documents/esm-g/plan/bert_training_v3/v3.1_cluster_training/writing/submission/nature_methods_submission_bundle_20260415/supplementary/table_s1_corpus_source_breakdown.csv): +From [table_s1_corpus_source_breakdown.csv](/writing/submission/nature_methods_submission_bundle_20260415/supplementary/table_s1_corpus_source_breakdown.csv): - GlyTouCan entries: `248,390` - CSDB entries: `39,112` @@ -67,7 +67,7 @@ Arithmetic check: ### Stage 3 | Original pre-augmentation corpus -From [probe_results.md](/Users/supantha/Documents/esm-g/plan/bert_training_v3/v3.1_cluster_training/writing/source_docs/probe_results.md:692) and direct local index checks: +From [probe_results.md](/writing/source_docs/probe_results.md:692) and direct local index checks: - `bert_training_v4/data/multimodal_index.csv` has `254,732` rows - all `254,732` rows are unique WURCS entries @@ -97,7 +97,7 @@ Direct local verification on `2026-04-19`: ### Stage 5 | Leakage-proof retraining corpus -From [table_s1_corpus_source_breakdown.csv](/Users/supantha/Documents/esm-g/plan/bert_training_v3/v3.1_cluster_training/writing/submission/nature_methods_submission_bundle_20260415/supplementary/table_s1_corpus_source_breakdown.csv): +From [table_s1_corpus_source_breakdown.csv](/writing/submission/nature_methods_submission_bundle_20260415/supplementary/table_s1_corpus_source_breakdown.csv): - leakage-proof corpus: `259,405` - benchmark-overlap glycans excluded: `1,298` @@ -177,10 +177,10 @@ Practical rule: ### Files checked -- `/Users/supantha/Documents/esm-g/plan/bert_training_v3/v3.1_cluster_training/data/sequences.pkl` -- `/Users/supantha/Documents/esm-g/plan/bert_training_v3/v3.1_cluster_training/data/sequences_bpe.pkl` -- `/Users/supantha/Documents/esm-g/plan/bert_training_v3/v3.1_cluster_training/bert_training_v4/data/sequences.pkl` -- tokenizer-generation code in [train_wurcs_bpe.py](/Users/supantha/Documents/esm-g/plan/bert_training_v3/v3.1_cluster_training/writing/cluster/nature_submission/code/training/train_wurcs_bpe.py) +- `/data/sequences.pkl` +- `/data/sequences_bpe.pkl` +- `/bert_training_v4/data/sequences.pkl` +- tokenizer-generation code in [train_wurcs_bpe.py](/writing/cluster/nature_submission/code/training/train_wurcs_bpe.py) ### What was verified @@ -199,7 +199,7 @@ So the BPE median is not coming from: ### Current live tokenizer stats -From [tokenizer_length_summary.csv](/Users/supantha/Documents/esm-g/plan/bert_training_v3/v3.1_cluster_training/writing/submission/nature_methods_submission_bundle_20260415/supplementary/tokenizer_support_evidence/tokenizer_length_summary.csv): +From [tokenizer_length_summary.csv](/writing/submission/nature_methods_submission_bundle_20260415/supplementary/tokenizer_support_evidence/tokenizer_length_summary.csv): - literal WURCS character length: - median `121` diff --git a/provenance/compute_provenance/COMPUTE_PROVENANCE_LOG.md b/provenance/compute_provenance/COMPUTE_PROVENANCE_LOG.md index 60bb4ab49f6f9dcf9a9ee37767a33d99446e19d0..ae9454958208d163aa91f85de01b431c6eab0d76 100644 --- a/provenance/compute_provenance/COMPUTE_PROVENANCE_LOG.md +++ b/provenance/compute_provenance/COMPUTE_PROVENANCE_LOG.md @@ -71,7 +71,7 @@ training-runtime wording. ### `good_log` -- Log path: `/Users/supantha/Documents/esm-g/plan/bert_training_v3/v3.1_cluster_training/logs/good_log.out` +- Log path: `/logs/good_log.out` - Checkpoint destination: `/work/ratul1/supantha/glycan-SD-VS/bert_training_v3/v3.1_cluster_training/checkpoints_v4_bpe` - Started line present: `not present in recovered log` - Completed training: `True` @@ -88,7 +88,7 @@ training-runtime wording. ### `current_log` -- Log path: `/Users/supantha/Documents/esm-g/plan/bert_training_v3/v3.1_cluster_training/logs/current_log.out` +- Log path: `/logs/current_log.out` - Checkpoint destination: `/work/ratul1/supantha/glycan-SD-VS/bert_training_v3/v3.1_cluster_training/checkpoints_v4_bpe_topology` - Started line present: `Tue Dec 23 10:04:07 PM CST 2025` - Completed training: `False` diff --git a/provenance/compute_provenance/compute_runtime_summary.csv b/provenance/compute_provenance/compute_runtime_summary.csv index 77c65ff7db1667a708d4e69888f45272bd7d74e0..d4111994cbd4afd3bdd7f3c666f69aae42a902a2 100644 --- a/provenance/compute_provenance/compute_runtime_summary.csv +++ b/provenance/compute_provenance/compute_runtime_summary.csv @@ -1,3 +1,3 @@ label,log_path,started,checkpoint_dir,completed_training,early_stopped,total_params,trainable_params,train_samples,val_samples,batch_size,epochs_requested,observed_total_epochs_field,observed_epochs_in_log,steps_per_epoch,avg_epoch_seconds,avg_epoch_minutes,median_epoch_minutes,min_epoch_seconds,max_epoch_seconds,approx_batches_per_second,approx_samples_per_second,total_epochs_reported,total_steps_reported,approx_total_hours_if_using_logged_epoch_average,note -good_log,/Users/supantha/Documents/esm-g/plan/bert_training_v3/v3.1_cluster_training/logs/good_log.out,,/work/ratul1/supantha/glycan-SD-VS/bert_training_v3/v3.1_cluster_training/checkpoints_v4_bpe,True,True,36997516,36997516,208562,52141,128,100,100,85,1630,620.2,10.34,9.97,594,690,2.6282,336.28,85,138550,14.64,Recovered completed pretraining log for the smaller checkpoint-writing run that saves into checkpoints_v4_bpe. -current_log,/Users/supantha/Documents/esm-g/plan/bert_training_v3/v3.1_cluster_training/logs/current_log.out,Tue Dec 23 10:04:07 PM CST 2025,/work/ratul1/supantha/glycan-SD-VS/bert_training_v3/v3.1_cluster_training/checkpoints_v4_bpe_topology,False,False,136470307,136470307,208562,52141,64,100,100,17,3259,3158.82,52.65,50.75,3040,3529,1.0317,66.03,,,,Recovered partial pretraining log for the heavier topology-enabled run that saves into checkpoints_v4_bpe_topology. +good_log,/logs/good_log.out,,/work/ratul1/supantha/glycan-SD-VS/bert_training_v3/v3.1_cluster_training/checkpoints_v4_bpe,True,True,36997516,36997516,208562,52141,128,100,100,85,1630,620.2,10.34,9.97,594,690,2.6282,336.28,85,138550,14.64,Recovered completed pretraining log for the smaller checkpoint-writing run that saves into checkpoints_v4_bpe. +current_log,/logs/current_log.out,Tue Dec 23 10:04:07 PM CST 2025,/work/ratul1/supantha/glycan-SD-VS/bert_training_v3/v3.1_cluster_training/checkpoints_v4_bpe_topology,False,False,136470307,136470307,208562,52141,64,100,100,17,3259,3158.82,52.65,50.75,3040,3529,1.0317,66.03,,,,Recovered partial pretraining log for the heavier topology-enabled run that saves into checkpoints_v4_bpe_topology. diff --git a/provenance/compute_provenance/compute_runtime_summary.json b/provenance/compute_provenance/compute_runtime_summary.json index 300ac97dfbd36ec0548c8a012828f56096fabed7..91602506be9711aa38c8fd858bd34ece3bb80a5f 100644 --- a/provenance/compute_provenance/compute_runtime_summary.json +++ b/provenance/compute_provenance/compute_runtime_summary.json @@ -1,18 +1,18 @@ { "generated_at": "2026-04-19T11:18:49", - "project_root": "/Users/supantha/Documents/esm-g/plan/bert_training_v3/v3.1_cluster_training", + "project_root": "", "support_scope": "These files support single-GPU runtime characterization from recovered local training logs and representative launch scripts. They do not, by themselves, support multi-GPU scaling efficiency, strong/weak scaling, GPU utilization, or deployment-latency claims.", "representative_launch_scripts": [ { "label": "run_training", - "script_path": "/Users/supantha/Documents/esm-g/plan/bert_training_v3/v3.1_cluster_training/run_training.sh", + "script_path": "/run_training.sh", "note": "Representative single-GPU v3.1 training launcher.", "requested_gpus": 1, "requests_a100": true }, { "label": "run_v4_bpe_pretraining", - "script_path": "/Users/supantha/Documents/esm-g/plan/bert_training_v3/v3.1_cluster_training/run_v4_bpe_pretraining.sh", + "script_path": "/run_v4_bpe_pretraining.sh", "note": "Representative single-GPU v4 BPE/topology pretraining launcher.", "requested_gpus": 1, "requests_a100": true @@ -21,7 +21,7 @@ "recovered_training_logs": [ { "label": "good_log", - "log_path": "/Users/supantha/Documents/esm-g/plan/bert_training_v3/v3.1_cluster_training/logs/good_log.out", + "log_path": "/logs/good_log.out", "note": "Recovered completed pretraining log for the smaller checkpoint-writing run that saves into checkpoints_v4_bpe.", "started": null, "checkpoint_dir": "/work/ratul1/supantha/glycan-SD-VS/bert_training_v3/v3.1_cluster_training/checkpoints_v4_bpe", @@ -49,7 +49,7 @@ }, { "label": "current_log", - "log_path": "/Users/supantha/Documents/esm-g/plan/bert_training_v3/v3.1_cluster_training/logs/current_log.out", + "log_path": "/logs/current_log.out", "note": "Recovered partial pretraining log for the heavier topology-enabled run that saves into checkpoints_v4_bpe_topology.", "started": "Tue Dec 23 10:04:07 PM CST 2025", "checkpoint_dir": "/work/ratul1/supantha/glycan-SD-VS/bert_training_v3/v3.1_cluster_training/checkpoints_v4_bpe_topology", diff --git a/provenance/compute_provenance/launch_script_summary.csv b/provenance/compute_provenance/launch_script_summary.csv index 7f6491f487bf20a7ca818cd261385f04921262da..8de77b9581082f1d84bc55413038e54c4c0621e6 100644 --- a/provenance/compute_provenance/launch_script_summary.csv +++ b/provenance/compute_provenance/launch_script_summary.csv @@ -1,3 +1,3 @@ label,script_path,requested_gpus,requests_a100,note -run_training,/Users/supantha/Documents/esm-g/plan/bert_training_v3/v3.1_cluster_training/run_training.sh,1,True,Representative single-GPU v3.1 training launcher. -run_v4_bpe_pretraining,/Users/supantha/Documents/esm-g/plan/bert_training_v3/v3.1_cluster_training/run_v4_bpe_pretraining.sh,1,True,Representative single-GPU v4 BPE/topology pretraining launcher. +run_training,/run_training.sh,1,True,Representative single-GPU v3.1 training launcher. +run_v4_bpe_pretraining,/run_v4_bpe_pretraining.sh,1,True,Representative single-GPU v4 BPE/topology pretraining launcher.