Hugging Face's logo

Datasets:
junidude14
/
PBISC-data

Modalities:
Tabular
Text
Formats:
parquet
Languages:
English
Size:
< 1K
Tags:
single-cell-rna-seq
bulk-rna-seq
diffusion
Libraries:
Datasets
pandas
Polars
Dataset card Data Studio Files
xet
 Community
1
junidude14 commited on
Commit
ec85bfc
·
verified ·
1 Parent(s): 1a8c0ed

Archive 4 checkpoints (1-4)

Browse files
Files changed (5) hide show
  1. .gitattributes +1 -0
  2. paper-releases/pbisc-diffusion/updates/20261013T005008Z-reuse-audit-results-deck/diffusion/dataset/server_1/model8/AGENTS.md +55 -0
  3. paper-releases/pbisc-diffusion/updates/20261013T005008Z-reuse-audit-results-deck/diffusion/dataset/server_1/model8/EXPERIMENT_INDEX.md +202 -0
  4. paper-releases/pbisc-diffusion/updates/20261013T005008Z-reuse-audit-results-deck/diffusion/dataset/server_1/model8/EXPERIMENT_REUSE_COVERAGE.json +0 -0
  5. paper-releases/pbisc-diffusion/updates/20261013T005008Z-reuse-audit-results-deck/diffusion/dataset/server_1/model8/EXPERIMENT_REUSE_INDEX.jsonl +3 -0
.gitattributes CHANGED
@@ -230,3 +230,4 @@ paper-releases/pbisc-diffusion/updates/20261012T233237Z-figure-layout-todo/diffu
230
  paper-releases/pbisc-diffusion/updates/20261012T233237Z-figure-layout-todo/diffusion/dataset/server_1/model8/plans/paper_figures_strictpairs_20261012/layout_preview_20261013/panels/Figure_4_layout.pdf filter=lfs diff=lfs merge=lfs -text
231
  paper-releases/pbisc-diffusion/updates/20261012T233237Z-figure-layout-todo/diffusion/dataset/server_1/model8/plans/paper_figures_strictpairs_20261012/layout_preview_20261013/panels/Figure_5_layout.pdf filter=lfs diff=lfs merge=lfs -text
232
  paper-releases/pbisc-diffusion/updates/20261012T233237Z-figure-layout-todo/diffusion/dataset/server_1/model8/plans/paper_figures_strictpairs_20261012/layout_preview_20261013/panels/Figure_6_layout.pdf filter=lfs diff=lfs merge=lfs -text
 
 
230
  paper-releases/pbisc-diffusion/updates/20261012T233237Z-figure-layout-todo/diffusion/dataset/server_1/model8/plans/paper_figures_strictpairs_20261012/layout_preview_20261013/panels/Figure_4_layout.pdf filter=lfs diff=lfs merge=lfs -text
231
  paper-releases/pbisc-diffusion/updates/20261012T233237Z-figure-layout-todo/diffusion/dataset/server_1/model8/plans/paper_figures_strictpairs_20261012/layout_preview_20261013/panels/Figure_5_layout.pdf filter=lfs diff=lfs merge=lfs -text
232
  paper-releases/pbisc-diffusion/updates/20261012T233237Z-figure-layout-todo/diffusion/dataset/server_1/model8/plans/paper_figures_strictpairs_20261012/layout_preview_20261013/panels/Figure_6_layout.pdf filter=lfs diff=lfs merge=lfs -text
233
+ paper-releases/pbisc-diffusion/updates/20261013T005008Z-reuse-audit-results-deck/diffusion/dataset/server_1/model8/EXPERIMENT_REUSE_INDEX.jsonl filter=lfs diff=lfs merge=lfs -text
paper-releases/pbisc-diffusion/updates/20261013T005008Z-reuse-audit-results-deck/diffusion/dataset/server_1/model8/AGENTS.md ADDED
@@ -0,0 +1,55 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Model 8 PBISC-Diffusion paper maintenance
2
+
3
+ This active Model 8 source and its experiments are part of the
4
+ `pbisc-diffusion` paper. At task start, read
5
+ `../../paper-release-20260928/maintenance/POLICY.md` along with this
6
+ directory's README and applicable parent instructions.
7
+
8
+ At the end of each related task, follow that policy's versioned update
9
+ workflow for the exact source, experiment, checkpoint, or evidence assets that
10
+ changed. From the release root, use
11
+ `maintenance/update.py prepare --request FILE --id ID`, inspect the prepared
12
+ update, then run `maintenance/update.py publish ID --execute` and
13
+ `maintenance/update.py status` under the policy. Record a no-op or pending
14
+ status when no
15
+ asset changed or publication failed; never describe that as an upload. Do not
16
+ regenerate or alter the frozen 2026-09-28 release, its root manifests, or the
17
+ HF `paper-20260928` revision. Inference assets are location plus receipt by
18
+ default unless the task explicitly needs their payload.
19
+
20
+ The private dataset must stay private. Include datalake, VAE, or Model 7 assets
21
+ only with exact evidence that they are dependencies of this diffusion paper.
22
+ The parent host GPU safety rules still apply.
23
+
24
+ ## Reuse existing experiments before starting compute
25
+
26
+ User instruction, 2026-10-13: avoid repeating completed Model 8 experiments.
27
+ Start with `EXPERIMENT_INDEX.md` and query `EXPERIMENT_REUSE_INDEX.jsonl` using
28
+ `scripts/audit_experiment_reuse.py`. A missing output in a new plan directory
29
+ does not establish that an experiment has never been done.
30
+
31
+ - Before training, generation, or deconvolution, compare existing source
32
+ receipts: base checkpoint, exact support/test sample and donor IDs, data and
33
+ condition content, measured genes/masks, validation access, optimization and
34
+ selection schedule, seed, and implementation/configuration provenance.
35
+ - Registry matches are candidates, not automatic equivalence. Verify payloads
36
+ and write truthful adoption evidence. Preserve original execution provenance;
37
+ do not relabel an old artifact as newly executed strict work.
38
+ - Changing a sampling seed or receipt format does not justify retraining an
39
+ existing model. Reuse verified checkpoints and run only the genuinely changed
40
+ operation. Missing native receipts call for adoption, not automatic reruns.
41
+ - Record the candidate sources and the exact scientific reason for any new
42
+ operation before launching it. Unknown provenance requires investigation,
43
+ not a conclusion that new compute is necessary. Consult newer live receipts
44
+ when the inventory predates an ongoing campaign.
45
+ - Distinguish existing results, CPU-only statistics/plotting, a new protocol
46
+ update, and missing data in plans and decks. Never label all four as an empty
47
+ TODO. Archive/backup copies and tensor-identical models are not independent
48
+ experimental evidence.
49
+
50
+ The 2026-10-13 audit and confirmed duplicate-training proof are in
51
+ `plans/paper_figures_strictpairs_20261012/audit_reuse_20261013/`.
52
+ Sepsis deconvolution uses verified historical draw-specific outputs through
53
+ `tools/adopt_deconv_legacy.py` in that plan; its methods must not be rerun as a
54
+ fallback when adoption validation fails. Current results review:
55
+ `presentations/2026-09-30--paper-summary--r00/PBISC-Diffusion_results_review_20261013.pdf`.
paper-releases/pbisc-diffusion/updates/20261013T005008Z-reuse-audit-results-deck/diffusion/dataset/server_1/model8/EXPERIMENT_INDEX.md ADDED
@@ -0,0 +1,202 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Model 8 experiment and reuse index
2
+
3
+ Observed on `spark-fa6b` at 2026-10-13T00:45:19.744801+00:00.
4
+
5
+ Before any training, inference, annotation, scoring or figure rebuild, search this index and the JSONL registry. An absent file under a new plan is not evidence that the result does not exist elsewhere.
6
+
7
+ ## Required reuse lookup
8
+
9
+ 1. Find the experiment family below; read its existing README/protocol, run records, generation receipts, score tables and deck source map.
10
+ 2. Check the current cross-host model/inference catalogs. Resolve actual host and path; inspect completed outputs before scheduling compute.
11
+ 3. Compare checkpoint, exact train/validation/test sample IDs, gene order/masks, conditioning/depth, selected/fixed steps, seed, generated cell budget, sampling/readout, scSemi HVG/assignment, and scoring reference.
12
+ 4. Classify the request: existing figure; CPU redraw/reanalysis; existing generated output awaiting scoring; or scientifically different new GPU work. Record the exact mismatch that requires new work.
13
+ 5. Unknown fields block an exact-match claim, not historical display with an accurate protocol label. Candidate matches never authorize deleting or overwriting outputs.
14
+ 6. Follow host memory/GPU policies for any later launch. This audit does not launch anything or verify live process state.
15
+
16
+ ```bash
17
+ python scripts/audit_experiment_reuse.py --query 'immicro/fewshot_n5' --limit 20
18
+ python scripts/audit_experiment_reuse.py --query 'refcal' --kind array_payload --limit 20
19
+ python scripts/audit_experiment_reuse.py --check-training-request /path/to/training-request.json
20
+ python scripts/audit_experiment_reuse.py --refresh
21
+ ```
22
+
23
+ The training-request JSON uses `checkpoint_sha256`, `train_samples` (ID list), `validation_samples` (ID list, empty when none), `mode`, `lora_rank`, `learning_rate`, `fixed_steps`, `training_seed`, `condition_sha256`, `training_data_sha256`, `gene_order` (ordered gene-ID list), `batch_size`, `warmup`, `ema_decay`, `training_code_sha256`, and `effective_config_sha256`; `cohort` is an optional filter. The check reports `reuse_candidate`, unknown fields, or a different protocol. A complete match has `metadata_match=true`, never an exact-reuse verdict. A generation/sampling seed is ignored for checkpoint-training comparison. Verified adoption still requires all scientifically relevant inputs/configuration/code, current checkpoint/output receipts, and payload identity or equivalence; fields outside this schema also require review.
24
+
25
+ ## Evidence already available
26
+
27
+ - **Pretraining and checkpoint parents:** [runs](runs); [plans/full_recipe_allcells_6k_broad_20260909](plans/full_recipe_allcells_6k_broad_20260909); [plans/floor01_campaign_20260923](plans/floor01_campaign_20260923). Use RUN_STATUS/history/config and checkpoint receipts together; payload presence alone is not completion.
28
+ - **Architecture and gene-panel ablations:** [plans/arch_ablation_6k_20260913](plans/arch_ablation_6k_20260913); [plans/gene_panel_scaling_20260908](plans/gene_panel_scaling_20260908); [analysis/floor01_benchmark_20260923](analysis/floor01_benchmark_20260923); [presentations/2026-09-30--paper-summary--r00/data/sec03_width_bench.csv](presentations/2026-09-30--paper-summary--r00/data/sec03_width_bench.csv); [presentations/2026-09-30--paper-summary--r00/data/sec03_panel_bench.csv](presentations/2026-09-30--paper-summary--r00/data/sec03_panel_bench.csv). Existing width/context/gene-panel results precede the strict-pair campaign.
29
+ - **Canonical paired training, exact-five historical draws and learning curves:** [plans/paired_finetune_intradataset_20260927](plans/paired_finetune_intradataset_20260927). ImMicro fewshot_n5: five samples, no inner validation, fixed3000, five draws. Historical lora_n curves: n+7 validation samples and early stopping. Do not conflate these.
30
+ - **scSemiProfiler and matched-budget comparisons:** [plans/paired_finetune_intradataset_20260927/scsemi](plans/paired_finetune_intradataset_20260927/scsemi); [plans/immicro_scsemi_comparison_20260927](plans/immicro_scsemi_comparison_20260927); [analysis/scsemi_benchmark_diffusion_20260913](analysis/scsemi_benchmark_diffusion_20260913). Check representative IDs, gene axis, hvset, seed index, assignment and exports before reusing training.
31
+ - **Cross-cohort transfer and from-scratch controls:** [plans/paired_crosscohort_20261002](plans/paired_crosscohort_20261002). LOCO and same-cohort adaptation are different estimands; retain support-cohort lineage.
32
+ - **Real-reference UMAP, external calibration and composition:** [plans/utility_studies_final6k_20260925](plans/utility_studies_final6k_20260925). Existing ImMicro real-reference UMAP, AIFI reference-calibration output and composition tables are reusable inputs; LoRA/scSemi panels require method-specific projection.
33
+ - **Deconvolution:** [plans/bayesprism_paired_20261006](plans/bayesprism_paired_20261006); [plans/utility_studies_final6k_20260925/deconv](plans/utility_studies_final6k_20260925/deconv). Match support pairs and supported endpoints; deconvolution is not cell generation.
34
+ - **Rare-state screen and Behcet diagnostic:** [plans/rare_states_paired_20261008](plans/rare_states_paired_20261008); [plans/rare_states_screen_20261004](plans/rare_states_screen_20261004); [plans/behcet_c1q_lora3_20261004](plans/behcet_c1q_lora3_20261004). Retain zero draws, donor dependence, marker rules and the descriptive status of Behcet.
35
+ - **Bulk-only generation and research:** [plans/bulkonly_lora5_20261006](plans/bulkonly_lora5_20261006); [plans/bulkonly_lora5_k4000_20261006](plans/bulkonly_lora5_k4000_20261006). Existing generated arrays and analyses are candidates for CPU reanalysis, not grounds to repeat inference.
36
+ - **76-slide evidence deck:** [presentations/2026-09-30--paper-summary--r00](presentations/2026-09-30--paper-summary--r00). slides_spec.json maps each slide to figures, source CSV and full provenance notes. Slides23–34 contain healthy metrics/curves/runtime.
37
+ - **Strict-pair protocol and new manuscript layouts:** [plans/paper_figures_strictpairs_20261012](plans/paper_figures_strictpairs_20261012). New fold support sets differ from historical draws/curves. Check generated outputs and receipts before considering missing scoring a reason to regenerate.
38
+ - **Current cross-host archive catalogs:** [../../paper-release-20260928/CURRENT_MODEL_INDEX.csv](../../paper-release-20260928/CURRENT_MODEL_INDEX.csv); [../../paper-release-20260928/CURRENT_INFERENCE_LOCATIONS.csv](../../paper-release-20260928/CURRENT_INFERENCE_LOCATIONS.csv). Resolve canonical host and location from these catalogs; identical paths can exist on both hosts. Catalog observations are not a live process check.
39
+ - **Separately verified duplicate-training evidence:** [plans/paper_figures_strictpairs_20261012/audit_reuse_20261013/CONFIRMED_DUPLICATE_TRAINING.json](plans/paper_figures_strictpairs_20261012/audit_reuse_20261013/CONFIRMED_DUPLICATE_TRAINING.json). Read the coordinator's exact scoped payload comparison and provenance. This audit links that evidence; generic metadata signatures do not inherit its confirmed status.
40
+
41
+ ## Coverage and limits
42
+
43
+ - Files with filesystem metadata: **66,604** across **138** families.
44
+ - JSON documents parsed under 16,777,216 bytes: **17,919**.
45
+ - JSON over the cap: **1**; invalid/unreadable JSON: **1**.
46
+ - JSON with unrecognized schema or no extracted scientific fields: **4,050**. These remain indexed with top-level keys.
47
+ - Excluded subtrees: **110**; directory symlinks not followed: **323**; traversal errors: **0**.
48
+ - Candidate signature groups: **120**; identical small scientific-metadata groups: **197**. Neither is a deletion list.
49
+ - Primary working-tree file observations: **63,947**; inherited archive/backup/snapshot files: **2,657**. File or metadata counts are not independent experiment counts.
50
+ - Existing cross-host archive location records indexed: **2,109**; these preserve declared source host/path and receipt pointers without claiming live verification.
51
+ - Separately confirmed duplicate-training aliases: **5**; proof status `validated_report_structure`. Their RUN/checkpoint records map to the original model, retain distinct serialization hashes, and are not independent model evidence. Generated outputs are not classified as duplicates.
52
+ - Large array/checkpoint contents were not inspected, loaded or hashed. Apparent bytes can double-count hardlinks and copies.
53
+ - No remote host or live process query is performed. `recorded_running_or_pending` is historical evidence, not current GPU status.
54
+ - CSV headers and JSONL/log/text prefixes are bounded samples. JSON records over the cap and excluded subtree descendants are explicitly uninspected.
55
+ - Files may change during this point-in-time inventory. Refresh before a new execution decision.
56
+
57
+ Detailed machine-readable records: [EXPERIMENT_REUSE_INDEX.jsonl](EXPERIMENT_REUSE_INDEX.jsonl). Counts, exclusions, family evidence, and candidate groups: [EXPERIMENT_REUSE_COVERAGE.json](EXPERIMENT_REUSE_COVERAGE.json).
58
+
59
+ ## Every inventoried family
60
+
61
+ State columns count records, not completed experiments; multiple states within one family are expected.
62
+
63
+ | Family | Files | Inherited copies | Figures | Tables | Arrays/checkpoints | Recorded completed/budget reached | Recorded running/pending | Recorded failed/incomplete |
64
+ |---|---:|---:|---:|---:|---:|---:|---:|---:|
65
+ | [analysis](analysis) | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
66
+ | [analysis/additive_2k_e1_e2_20260910](analysis/additive_2k_e1_e2_20260910) | 15 | 0 | 4 | 1 | 0 | 2 | 0 | 0 |
67
+ | [analysis/archive](analysis/archive) | 5 | 5 | 2 | 1 | 1 | 0 | 0 | 0 |
68
+ | [analysis/c1q_ifn_validation_20260926T2314](analysis/c1q_ifn_validation_20260926T2314) | 50 | 0 | 8 | 20 | 1 | 0 | 0 | 0 |
69
+ | [analysis/diffusion_paper_after_2k_e3_20260910_analysis_unknown_v1](analysis/diffusion_paper_after_2k_e3_20260910_analysis_unknown_v1) | 546 | 0 | 8 | 465 | 21 | 1 | 0 | 0 |
70
+ | [analysis/diffusion_paper_after_2k_e3_20260910_generation_v1](analysis/diffusion_paper_after_2k_e3_20260910_generation_v1) | 235 | 0 | 0 | 0 | 225 | 5 | 0 | 0 |
71
+ | [analysis/diffusion_trajectory_20260917](analysis/diffusion_trajectory_20260917) | 49 | 0 | 11 | 2 | 1 | 2 | 0 | 0 |
72
+ | [analysis/diffusion_trajectory_biology_20260917](analysis/diffusion_trajectory_biology_20260917) | 258 | 0 | 104 | 7 | 6 | 16 | 0 | 0 |
73
+ | [analysis/floor01_benchmark_20260923](analysis/floor01_benchmark_20260923) | 115 | 0 | 0 | 36 | 38 | 0 | 0 | 0 |
74
+ | [analysis/gse135779_s0_2000hvg_deg_dualprotocol_seed20260907](analysis/gse135779_s0_2000hvg_deg_dualprotocol_seed20260907) | 113 | 0 | 10 | 36 | 47 | 2 | 0 | 0 |
75
+ | [analysis/gse135779_s0_2000hvg_deg_dualprotocol_smoke_seed20260907](analysis/gse135779_s0_2000hvg_deg_dualprotocol_smoke_seed20260907) | 5 | 0 | 0 | 0 | 2 | 1 | 0 | 0 |
76
+ | [analysis/proportion_k1000_20260910_analysis_v1](analysis/proportion_k1000_20260910_analysis_v1) | 16 | 0 | 4 | 7 | 0 | 1 | 0 | 0 |
77
+ | [analysis/proportion_k1000_20260910_generation_v1](analysis/proportion_k1000_20260910_generation_v1) | 52 | 0 | 0 | 0 | 44 | 4 | 0 | 0 |
78
+ | [analysis/s0_1000hvg_val48x64_ddim50_collection](analysis/s0_1000hvg_val48x64_ddim50_collection) | 2 | 0 | 0 | 0 | 1 | 1 | 0 | 0 |
79
+ | [analysis/s0_1000hvg_val48x64_ddim50_context](analysis/s0_1000hvg_val48x64_ddim50_context) | 4 | 0 | 0 | 2 | 1 | 1 | 0 | 0 |
80
+ | [analysis/s0_1000hvg_val48x64_ddim50_reference_umap](analysis/s0_1000hvg_val48x64_ddim50_reference_umap) | 9 | 0 | 2 | 5 | 1 | 1 | 0 | 0 |
81
+ | [analysis/s0_2000hvg_model_explainer_site](analysis/s0_2000hvg_model_explainer_site) | 165 | 0 | 2 | 0 | 0 | 0 | 0 | 0 |
82
+ | [analysis/s0_2000hvg_val48x64_ddim50_analysis_bundle](analysis/s0_2000hvg_val48x64_ddim50_analysis_bundle) | 18 | 0 | 10 | 3 | 0 | 0 | 1 | 0 |
83
+ | [analysis/s0_2000hvg_val48x64_ddim50_analysis_bundle_globalprime_v2](analysis/s0_2000hvg_val48x64_ddim50_analysis_bundle_globalprime_v2) | 20 | 0 | 10 | 3 | 0 | 2 | 0 | 0 |
84
+ | [analysis/s0_2000hvg_val48x64_ddim50_collection](analysis/s0_2000hvg_val48x64_ddim50_collection) | 3 | 0 | 0 | 0 | 1 | 1 | 0 | 0 |
85
+ | [analysis/s0_2000hvg_val48x64_ddim50_context](analysis/s0_2000hvg_val48x64_ddim50_context) | 4 | 0 | 0 | 2 | 1 | 1 | 0 | 0 |
86
+ | [analysis/s0_2000hvg_val48x64_ddim50_context_globalprime_v2](analysis/s0_2000hvg_val48x64_ddim50_context_globalprime_v2) | 4 | 0 | 0 | 2 | 1 | 1 | 0 | 0 |
87
+ | [analysis/s0_2000hvg_val48x64_ddim50_reference_umap](analysis/s0_2000hvg_val48x64_ddim50_reference_umap) | 9 | 0 | 2 | 5 | 1 | 1 | 0 | 0 |
88
+ | [analysis/s0_2000hvg_val48x64_ddim50_reference_umap_globalprime_v2](analysis/s0_2000hvg_val48x64_ddim50_reference_umap_globalprime_v2) | 9 | 0 | 2 | 5 | 1 | 1 | 0 | 0 |
89
+ | [analysis/s0_2k_6k_nearest_cells_20260910](analysis/s0_2k_6k_nearest_cells_20260910) | 39 | 3 | 4 | 22 | 1 | 1 | 0 | 0 |
90
+ | [analysis/s0_2k_6k_raw_counts_20260910](analysis/s0_2k_6k_raw_counts_20260910) | 70 | 0 | 4 | 55 | 0 | 1 | 0 | 0 |
91
+ | [analysis/s0_2k_vs_6k_epoch1_umap_20260909](analysis/s0_2k_vs_6k_epoch1_umap_20260909) | 10 | 0 | 6 | 1 | 1 | 1 | 0 | 0 |
92
+ | [analysis/s0_6k_epoch1_val48x64_ddim50_collection_20260909](analysis/s0_6k_epoch1_val48x64_ddim50_collection_20260909) | 4 | 0 | 0 | 0 | 2 | 1 | 0 | 0 |
93
+ | [analysis/scsemi_benchmark_diffusion_20260913](analysis/scsemi_benchmark_diffusion_20260913) | 5844 | 0 | 0 | 3028 | 912 | 3 | 0 | 0 |
94
+ | [analysis/sdd_gene_width_feasibility_seed20260627_v3](analysis/sdd_gene_width_feasibility_seed20260627_v3) | 1 | 0 | 0 | 1 | 0 | 0 | 0 | 0 |
95
+ | [analysis/umap_smoke_v4_test8x64_s20](analysis/umap_smoke_v4_test8x64_s20) | 5 | 0 | 2 | 1 | 1 | 0 | 0 | 0 |
96
+ | [artifacts/diffusion_paper_native_reference_20260910](artifacts/diffusion_paper_native_reference_20260910) | 2 | 0 | 0 | 0 | 1 | 1 | 0 | 0 |
97
+ | [artifacts/gse135779_s0_2000hvg_pediatric44_seed20260907](artifacts/gse135779_s0_2000hvg_pediatric44_seed20260907) | 18 | 0 | 0 | 4 | 10 | 1 | 0 | 0 |
98
+ | [artifacts/gse135779_scsemi_split_target42_20260913](artifacts/gse135779_scsemi_split_target42_20260913) | 6 | 0 | 0 | 2 | 1 | 1 | 0 | 0 |
99
+ | [artifacts/hvg_1000_train64x64_seed20260627](artifacts/hvg_1000_train64x64_seed20260627) | 3 | 0 | 0 | 1 | 1 | 0 | 0 | 0 |
100
+ | [artifacts/hvg_2000_train256x128_seed20260627](artifacts/hvg_2000_train256x128_seed20260627) | 3 | 0 | 0 | 1 | 1 | 0 | 0 | 0 |
101
+ | [artifacts/hvg_3000_train256x128_seed20260627](artifacts/hvg_3000_train256x128_seed20260627) | 3 | 0 | 0 | 1 | 1 | 0 | 0 | 0 |
102
+ | [artifacts/panel_broad_mustkeep_seed20260627_v1](artifacts/panel_broad_mustkeep_seed20260627_v1) | 4 | 0 | 0 | 1 | 1 | 1 | 0 | 0 |
103
+ | [artifacts/panel_full_20097_20260924](artifacts/panel_full_20097_20260924) | 3 | 0 | 0 | 1 | 1 | 0 | 0 | 0 |
104
+ | [artifacts/panel_ladder_1m_20260922](artifacts/panel_ladder_1m_20260922) | 46 | 0 | 0 | 14 | 18 | 0 | 0 | 0 |
105
+ | [artifacts/panel_ladder_floor01_1m_20260923](artifacts/panel_ladder_floor01_1m_20260923) | 23 | 0 | 0 | 7 | 9 | 0 | 0 | 0 |
106
+ | [artifacts/panel_ladder_floor01_1m_20260924_r2](artifacts/panel_ladder_floor01_1m_20260924_r2) | 26 | 0 | 0 | 8 | 10 | 0 | 0 | 0 |
107
+ | [artifacts/panel_ladder_train256x128_seed20260627_v2](artifacts/panel_ladder_train256x128_seed20260627_v2) | 18 | 0 | 0 | 7 | 5 | 0 | 0 | 0 |
108
+ | [artifacts/panel_ladder_train256x128_seed20260627_v3](artifacts/panel_ladder_train256x128_seed20260627_v3) | 24 | 0 | 0 | 7 | 9 | 0 | 0 | 0 |
109
+ | [artifacts/proportion_classifier_k1000_20260910_v1](artifacts/proportion_classifier_k1000_20260910_v1) | 7 | 0 | 0 | 4 | 1 | 2 | 0 | 0 |
110
+ | [artifacts/proportion_reference_k1000_20260910_v1](artifacts/proportion_reference_k1000_20260910_v1) | 13 | 0 | 0 | 5 | 5 | 1 | 0 | 0 |
111
+ | [configs](configs) | 13 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
112
+ | [logs](logs) | 23 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
113
+ | [plans/additive_2k_allcells_20260909](plans/additive_2k_allcells_20260909) | 70 | 28 | 0 | 1 | 0 | 3 | 3 | 2 |
114
+ | [plans/aifi_scanvi_rebuild_20260916](plans/aifi_scanvi_rebuild_20260916) | 648 | 0 | 0 | 47 | 185 | 6 | 0 | 0 |
115
+ | [plans/arch_ablation_6k_20260913](plans/arch_ablation_6k_20260913) | 154 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
116
+ | [plans/bayesprism_paired_20261006](plans/bayesprism_paired_20261006) | 3217 | 0 | 178 | 473 | 12 | 0 | 0 | 0 |
117
+ | [plans/behcet_c1q_lora3_20261004](plans/behcet_c1q_lora3_20261004) | 266 | 0 | 0 | 132 | 34 | 2 | 0 | 0 |
118
+ | [plans/behcet_c1q_repro_20260921](plans/behcet_c1q_repro_20260921) | 86 | 0 | 0 | 35 | 9 | 0 | 0 | 0 |
119
+ | [plans/behcet_pair_recovery_20261004](plans/behcet_pair_recovery_20261004) | 39 | 0 | 0 | 22 | 0 | 0 | 0 | 0 |
120
+ | [plans/broad_shape_smoke](plans/broad_shape_smoke) | 4 | 0 | 0 | 0 | 0 | 0 | 1 | 0 |
121
+ | [plans/bulkonly_lora5_20261006](plans/bulkonly_lora5_20261006) | 5384 | 4 | 57 | 2018 | 392 | 5 | 0 | 0 |
122
+ | [plans/bulkonly_lora5_k4000_20261006](plans/bulkonly_lora5_k4000_20261006) | 936 | 0 | 1 | 58 | 322 | 10 | 0 | 0 |
123
+ | [plans/c1q_crossdisease_20261004](plans/c1q_crossdisease_20261004) | 71 | 0 | 0 | 6 | 24 | 0 | 0 | 0 |
124
+ | [plans/conditioning_use_20260921](plans/conditioning_use_20260921) | 14 | 0 | 0 | 0 | 3 | 0 | 0 | 0 |
125
+ | [plans/context_rank_three_models_20260908](plans/context_rank_three_models_20260908) | 2 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
126
+ | [plans/floor01_campaign_20260923](plans/floor01_campaign_20260923) | 69 | 0 | 0 | 0 | 8 | 3 | 0 | 0 |
127
+ | [plans/full20k_followup_20260912](plans/full20k_followup_20260912) | 3 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
128
+ | [plans/full_recipe_allcells_6k_broad_20260909](plans/full_recipe_allcells_6k_broad_20260909) | 395 | 111 | 0 | 24 | 7 | 6 | 23 | 4 |
129
+ | [plans/gene_panel_scaling_20260908](plans/gene_panel_scaling_20260908) | 70 | 15 | 8 | 15 | 1 | 3 | 0 | 0 |
130
+ | [plans/immicro_scsemi_comparison_20260927](plans/immicro_scsemi_comparison_20260927) | 2234 | 0 | 21 | 21 | 1218 | 228 | 3 | 0 |
131
+ | [plans/paired_crosscohort_20261002](plans/paired_crosscohort_20261002) | 1694 | 4 | 0 | 307 | 353 | 0 | 0 | 0 |
132
+ | [plans/paired_finetune_intradataset_20260927](plans/paired_finetune_intradataset_20260927) | 27553 | 186 | 15 | 1469 | 5823 | 7899 | 0 | 0 |
133
+ | [plans/panel_floor01_20260923](plans/panel_floor01_20260923) | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
134
+ | [plans/panel_ladder_1m_20260922](plans/panel_ladder_1m_20260922) | 9 | 0 | 0 | 5 | 0 | 0 | 0 | 0 |
135
+ | [plans/panel_ladder_train256x128_seed20260627_v3](plans/panel_ladder_train256x128_seed20260627_v3) | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
136
+ | [plans/panel_scaling_e16_20260922](plans/panel_scaling_e16_20260922) | 11 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
137
+ | [plans/paper_eval_after_2k_e3_20260910](plans/paper_eval_after_2k_e3_20260910) | 221 | 115 | 0 | 4 | 1 | 10 | 11 | 2 |
138
+ | [plans/paper_figures_strictpairs_20261012](plans/paper_figures_strictpairs_20261012) | 8225 | 1841 | 419 | 557 | 1576 | 1915 | 12 | 3 |
139
+ | [plans/prop_condition_6k_20260916](plans/prop_condition_6k_20260916) | 975 | 0 | 0 | 3 | 887 | 0 | 0 | 0 |
140
+ | [plans/proportion_preservation_20260910](plans/proportion_preservation_20260910) | 95 | 9 | 0 | 1 | 1 | 4 | 3 | 2 |
141
+ | [plans/rare_states_paired_20261008](plans/rare_states_paired_20261008) | 106 | 0 | 0 | 3 | 96 | 0 | 0 | 0 |
142
+ | [plans/rare_states_screen_20261004](plans/rare_states_screen_20261004) | 270 | 0 | 0 | 107 | 38 | 0 | 0 | 0 |
143
+ | [plans/realbulk_generation_20260922](plans/realbulk_generation_20260922) | 45 | 0 | 0 | 0 | 18 | 0 | 0 | 0 |
144
+ | [plans/roundrobin_until_saturation_20260910](plans/roundrobin_until_saturation_20260910) | 238 | 5 | 0 | 1 | 0 | 20 | 15 | 0 |
145
+ | [plans/s0_1000hvg_allmeasured_seed20260627](plans/s0_1000hvg_allmeasured_seed20260627) | 4 | 0 | 0 | 0 | 0 | 0 | 1 | 0 |
146
+ | [plans/s0_2000hvg_fullpool16_allmeasured_seed20260627](plans/s0_2000hvg_fullpool16_allmeasured_seed20260627) | 14 | 0 | 0 | 0 | 0 | 0 | 3 | 0 |
147
+ | [plans/s0_2000hvg_gse135779_deg_dualprotocol_seed20260907](plans/s0_2000hvg_gse135779_deg_dualprotocol_seed20260907) | 6 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
148
+ | [plans/s0_condition_rank_20260908](plans/s0_condition_rank_20260908) | 4 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
149
+ | [plans/sdd_3000hvg_allmeasured_16ep_seed20260627](plans/sdd_3000hvg_allmeasured_16ep_seed20260627) | 1 | 0 | 0 | 0 | 0 | 0 | 1 | 0 |
150
+ | [plans/sdd_gene_width_feasibility_seed20260627_v3](plans/sdd_gene_width_feasibility_seed20260627_v3) | 3 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
151
+ | [plans/shape_smoke_ladder](plans/shape_smoke_ladder) | 5 | 0 | 0 | 0 | 0 | 0 | 1 | 0 |
152
+ | [plans/sparsity_threshold_20260921](plans/sparsity_threshold_20260921) | 5 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
153
+ | [plans/step_sweep_prop_e8_20260918](plans/step_sweep_prop_e8_20260918) | 22 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
154
+ | [plans/three_panel_resume_after_evaluations_20260910](plans/three_panel_resume_after_evaluations_20260910) | 19 | 7 | 0 | 0 | 0 | 1 | 2 | 0 |
155
+ | [plans/threshold_rescore_20260922](plans/threshold_rescore_20260922) | 37 | 0 | 0 | 1 | 0 | 0 | 0 | 0 |
156
+ | [plans/umap_2k_6k_epoch1_20260909](plans/umap_2k_6k_epoch1_20260909) | 36 | 13 | 0 | 0 | 0 | 1 | 3 | 0 |
157
+ | [plans/utility_studies_final6k_20260925](plans/utility_studies_final6k_20260925) | 3538 | 0 | 115 | 1194 | 443 | 1 | 0 | 0 |
158
+ | [plans/writeup_20260922](plans/writeup_20260922) | 3 | 0 | 2 | 0 | 0 | 0 | 0 | 0 |
159
+ | [presentations/2026-09-30--paper-summary--r00](presentations/2026-09-30--paper-summary--r00) | 675 | 274 | 322 | 154 | 0 | 8 | 0 | 2 |
160
+ | [root](.) | 7 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
161
+ | [runs/archive](runs/archive) | 37 | 37 | 0 | 3 | 17 | 4 | 0 | 0 |
162
+ | [runs/f01_2k_1x_seed20260627](runs/f01_2k_1x_seed20260627) | 77 | 0 | 0 | 0 | 1 | 2 | 0 | 0 |
163
+ | [runs/f01_3k_1x_seed20260627](runs/f01_3k_1x_seed20260627) | 2 | 0 | 0 | 0 | 0 | 1 | 0 | 0 |
164
+ | [runs/f01_6k_1x_seed20260627](runs/f01_6k_1x_seed20260627) | 78 | 0 | 0 | 0 | 2 | 2 | 0 | 0 |
165
+ | [runs/f01_p15k_1x_seed20260627](runs/f01_p15k_1x_seed20260627) | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
166
+ | [runs/hvg_smoke_16](runs/hvg_smoke_16) | 3 | 0 | 0 | 1 | 1 | 0 | 0 | 0 |
167
+ | [runs/panel_scaling_20260908_pilot_v1_2k_128steps_seed20260627](runs/panel_scaling_20260908_pilot_v1_2k_128steps_seed20260627) | 12 | 0 | 0 | 0 | 3 | 3 | 0 | 0 |
168
+ | [runs/panel_scaling_20260908_pilot_v1_6k_128steps_seed20260627](runs/panel_scaling_20260908_pilot_v1_6k_128steps_seed20260627) | 12 | 0 | 0 | 0 | 3 | 3 | 0 | 0 |
169
+ | [runs/panel_scaling_20260908_pilot_v1_broad_128steps_seed20260627](runs/panel_scaling_20260908_pilot_v1_broad_128steps_seed20260627) | 12 | 0 | 0 | 0 | 3 | 3 | 0 | 0 |
170
+ | [runs/panel_scaling_20260908_pilot_v1_full_128steps_seed20260627](runs/panel_scaling_20260908_pilot_v1_full_128steps_seed20260627) | 12 | 0 | 0 | 0 | 3 | 3 | 0 | 0 |
171
+ | [runs/s0_1000hvg_allmeasured_seed20260627](runs/s0_1000hvg_allmeasured_seed20260627) | 4 | 0 | 0 | 0 | 0 | 0 | 0 | 1 |
172
+ | [runs/s0_1000hvg_allmeasured_seed20260627_attempt2](runs/s0_1000hvg_allmeasured_seed20260627_attempt2) | 7 | 0 | 0 | 0 | 1 | 1 | 0 | 0 |
173
+ | [runs/s0_2000hvg_fullpool16_allmeasured_seed20260627](runs/s0_2000hvg_fullpool16_allmeasured_seed20260627) | 7 | 0 | 0 | 0 | 1 | 1 | 0 | 0 |
174
+ | [runs/s0_2k_allK_topup16_seed20260627_additive_v1](runs/s0_2k_allK_topup16_seed20260627_additive_v1) | 84 | 0 | 0 | 0 | 0 | 6 | 6 | 0 |
175
+ | [runs/s0_2k_allK_topup16_seed20260627_additive_v1_profile](runs/s0_2k_allK_topup16_seed20260627_additive_v1_profile) | 10 | 0 | 0 | 0 | 1 | 2 | 1 | 0 |
176
+ | [runs/s0_6k_allK_topup16_seed20260627_arch16x_v1](runs/s0_6k_allK_topup16_seed20260627_arch16x_v1) | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
177
+ | [runs/s0_6k_allK_topup16_seed20260627_arch4x_ctx512_v1](runs/s0_6k_allK_topup16_seed20260627_arch4x_ctx512_v1) | 85 | 0 | 0 | 0 | 1 | 2 | 4 | 0 |
178
+ | [runs/s0_6k_allK_topup16_seed20260627_arch4x_ctx512_v1_profile](runs/s0_6k_allK_topup16_seed20260627_arch4x_ctx512_v1_profile) | 14 | 0 | 0 | 0 | 1 | 2 | 1 | 0 |
179
+ | [runs/s0_6k_allK_topup16_seed20260627_arch4x_v1](runs/s0_6k_allK_topup16_seed20260627_arch4x_v1) | 56 | 0 | 0 | 0 | 0 | 0 | 6 | 0 |
180
+ | [runs/s0_6k_allK_topup16_seed20260627_arch4x_v1_profile](runs/s0_6k_allK_topup16_seed20260627_arch4x_v1_profile) | 10 | 0 | 0 | 0 | 1 | 2 | 1 | 0 |
181
+ | [runs/s0_6k_allK_topup16_seed20260627_archctx2048_v1](runs/s0_6k_allK_topup16_seed20260627_archctx2048_v1) | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
182
+ | [runs/s0_6k_allK_topup16_seed20260627_bundle_v1](runs/s0_6k_allK_topup16_seed20260627_bundle_v1) | 10 | 0 | 0 | 0 | 0 | 0 | 2 | 0 |
183
+ | [runs/s0_6k_allK_topup16_seed20260627_bundle_v1_profile](runs/s0_6k_allK_topup16_seed20260627_bundle_v1_profile) | 10 | 0 | 0 | 0 | 1 | 2 | 1 | 0 |
184
+ | [runs/s0_6k_allK_topup16_seed20260627_bundle_v2](runs/s0_6k_allK_topup16_seed20260627_bundle_v2) | 112 | 0 | 0 | 0 | 1 | 11 | 5 | 0 |
185
+ | [runs/s0_6k_allK_topup16_seed20260627_bundle_v2_profile](runs/s0_6k_allK_topup16_seed20260627_bundle_v2_profile) | 10 | 0 | 0 | 0 | 1 | 2 | 1 | 0 |
186
+ | [runs/s0_6k_allK_topup16_seed20260627_lossactive_v1](runs/s0_6k_allK_topup16_seed20260627_lossactive_v1) | 77 | 0 | 0 | 0 | 1 | 2 | 0 | 0 |
187
+ | [runs/s0_6k_allK_topup16_seed20260627_prop_v1](runs/s0_6k_allK_topup16_seed20260627_prop_v1) | 95 | 0 | 0 | 0 | 1 | 2 | 6 | 0 |
188
+ | [runs/s0_6k_allK_topup16_seed20260627_prop_v1_profile](runs/s0_6k_allK_topup16_seed20260627_prop_v1_profile) | 14 | 0 | 0 | 0 | 1 | 2 | 1 | 0 |
189
+ | [runs/s0_broad_allK_topup16_seed20260627_bundle_v1_profile](runs/s0_broad_allK_topup16_seed20260627_bundle_v1_profile) | 10 | 0 | 0 | 0 | 1 | 2 | 1 | 0 |
190
+ | [runs/s0_broad_allK_topup16_seed20260627_bundle_v2](runs/s0_broad_allK_topup16_seed20260627_bundle_v2) | 47 | 0 | 0 | 0 | 0 | 6 | 4 | 0 |
191
+ | [runs/s0_broad_allK_topup16_seed20260627_bundle_v2_profile](runs/s0_broad_allK_topup16_seed20260627_bundle_v2_profile) | 10 | 0 | 0 | 0 | 1 | 2 | 1 | 0 |
192
+ | [runs/sdd_ladder_seed20260627_v3_3k_shape_smoke](runs/sdd_ladder_seed20260627_v3_3k_shape_smoke) | 8 | 0 | 0 | 0 | 1 | 1 | 0 | 0 |
193
+ | [runs/sdd_ladder_seed20260627_v3_6k_shape_smoke](runs/sdd_ladder_seed20260627_v3_6k_shape_smoke) | 8 | 0 | 0 | 0 | 1 | 1 | 0 | 0 |
194
+ | [runs/sdd_ladder_seed20260627_v3_broad11589_shape_smoke](runs/sdd_ladder_seed20260627_v3_broad11589_shape_smoke) | 8 | 0 | 0 | 0 | 1 | 1 | 0 | 0 |
195
+ | [runs/smoke_native_e2](runs/smoke_native_e2) | 12 | 0 | 0 | 0 | 6 | 1 | 0 | 0 |
196
+ | [runs/smoke_s0_1000hvg_gpu_seed20260627](runs/smoke_s0_1000hvg_gpu_seed20260627) | 7 | 0 | 0 | 0 | 1 | 1 | 0 | 0 |
197
+ | [runs/smoke_s0_2000hvg_fullbatch_seed20260627](runs/smoke_s0_2000hvg_fullbatch_seed20260627) | 7 | 0 | 0 | 0 | 1 | 1 | 0 | 0 |
198
+ | [runs/smoke_v5_ema_heartbeat_20260904](runs/smoke_v5_ema_heartbeat_20260904) | 7 | 0 | 0 | 0 | 2 | 1 | 0 | 0 |
199
+ | [runs/smoke_v5_txn_order_20260904](runs/smoke_v5_txn_order_20260904) | 6 | 0 | 0 | 0 | 0 | 0 | 1 | 0 |
200
+ | [scripts](scripts) | 121 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
201
+ | [src](src) | 11 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
202
+ | [tests](tests) | 89 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
paper-releases/pbisc-diffusion/updates/20261013T005008Z-reuse-audit-results-deck/diffusion/dataset/server_1/model8/EXPERIMENT_REUSE_COVERAGE.json ADDED
The diff for this file is too large to render. See raw diff
 
paper-releases/pbisc-diffusion/updates/20261013T005008Z-reuse-audit-results-deck/diffusion/dataset/server_1/model8/EXPERIMENT_REUSE_INDEX.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:96e59c20a9718afcfcaa729bf30b645fb979eeacd8c9a81922b3cc68f9121980
3
+ size 57450669