williamratcliff commited on
Commit
ca3e97f
·
verified ·
1 Parent(s): ca1263a

Add files using upload-large-folder tool

Browse files
.gitattributes CHANGED
@@ -58,3 +58,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
58
  # Video files - compressed
59
  *.mp4 filter=lfs diff=lfs merge=lfs -text
60
  *.webm filter=lfs diff=lfs merge=lfs -text
 
 
 
58
  # Video files - compressed
59
  *.mp4 filter=lfs diff=lfs merge=lfs -text
60
  *.webm filter=lfs diff=lfs merge=lfs -text
61
+ benchmark_payloads/RRUFF_option1_473_with_buckets_maxnorm.hdf5 filter=lfs diff=lfs merge=lfs -text
62
+ benchmark_payloads/RRUFF_usable_plus_recoverable_325_with_labels_maxnorm.hdf5 filter=lfs diff=lfs merge=lfs -text
.gitignore ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ # Local scratch files
2
+ .DS_Store
3
+ *.tmp
4
+ *.log
README.md CHANGED
@@ -1,3 +1,65 @@
1
- ---
2
- license: mit
3
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ language: en
3
+ license: cc-by-4.0
4
+ task_categories:
5
+ - classification
6
+ tags:
7
+ - materials-science
8
+ - x-ray-diffraction
9
+ - crystallography
10
+ - benchmark
11
+ - neurips
12
+ ---
13
+
14
+ # InformedXRD Benchmarks
15
+
16
+ This staging directory is the draft Hugging Face dataset card for the NeurIPS Evaluations & Datasets artifact release associated with *InformedXRD*.
17
+
18
+ ## Contents
19
+
20
+ The intended dataset repo contains:
21
+
22
+ - the frozen `RRUFF-473` benchmark payload
23
+ - the frozen `RRUFF-325` benchmark payload
24
+ - a small reviewer subset with compact precomputed inference outputs
25
+ - manifests and metadata needed to reproduce the benchmark definition
26
+
27
+ It does **not** contain:
28
+
29
+ - training code
30
+ - the broader PRX archival release
31
+ - model checkpoints
32
+
33
+ ## Why these benchmarks exist
34
+
35
+ The paper's main benchmark contribution is not raw scale. It is a reproducible, auditable evaluation surface for real PXRD symmetry classification:
36
+
37
+ - `RRUFF-473` retains hard real-data cases and supports broad robustness analysis
38
+ - `RRUFF-325` is the cleaner downstream subset used for controlled calibration and error-structure analysis
39
+
40
+ Both benchmarks are tied to explicit curation logic and frozen artifact manifests.
41
+
42
+ ## Intended use
43
+
44
+ These datasets are intended for:
45
+
46
+ - extinction-group evaluation on real powder diffraction patterns
47
+ - hierarchy-aware rescoring of model predictions
48
+ - benchmarking Top-k, prior-baseline, and stratified evaluation claims
49
+
50
+ They are not intended as general-purpose training corpora.
51
+
52
+ ## Companion assets
53
+
54
+ The associated code, manifests, and reviewer notebooks live in the separate GitHub reproducibility repo:
55
+
56
+ - `scattering/neurips-ai-diffraction`
57
+
58
+ The associated model checkpoints live in the separate Hugging Face model repo.
59
+
60
+ ## Publication plan
61
+
62
+ 1. Create a private Hugging Face dataset repo.
63
+ 2. Upload the benchmark payloads, reviewer subset, and metadata.
64
+ 3. Verify the benchmark manifests against the uploaded filenames.
65
+ 4. Switch visibility when the review/release posture is decided.
benchmark_payloads/RRUFF_option1_473_with_buckets_maxnorm.hdf5 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:836d30a81879a56afab635cea149d405fb7e2174f972514804f85b470791a5d8
3
+ size 6794891
benchmark_payloads/RRUFF_usable_plus_recoverable_325_with_labels_maxnorm.hdf5 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b3dfbc266120668b53bed0db91e9b02a518264ef40061beabd389999c8d9d7f3
3
+ size 22193796
metadata/ARTIFACT_PLAN.md ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ## Dataset Artifact Plan
2
+
3
+ Candidate contents:
4
+ - frozen `RRUFF-473` benchmark files
5
+ - frozen `RRUFF-325` benchmark files
6
+ - benchmark ID lists / checksums
7
+ - reviewer subset matching the GitHub reviewer docs
8
+ - dataset card text
9
+
10
+ Candidate exclusions:
11
+ - files already kept only for the PRX archival release
12
+ - large unrelated training corpora
13
+ - private intermediate curation outputs
metadata/MANIFEST.csv ADDED
@@ -0,0 +1,60 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ path,kind,description,status,expected_local_path,checkpoint_role
2
+ src/paper_ai_diffraction/core/model.py,source,Core model definition,copied,,
3
+ src/paper_ai_diffraction/core/dataset.py,source,Dataset loading including mixed dual-source support,copied,,
4
+ src/paper_ai_diffraction/core/train.py,source,Training entrypoint,copied,,
5
+ src/paper_ai_diffraction/core/inference.py,source,Inference helpers,copied,,
6
+ src/paper_ai_diffraction/core/streaming_dataset.py,source,Streaming dataset helper retained because train.py imports it,copied,,
7
+ src/paper_ai_diffraction/utils/extinction_multilabel.py,source,Extinction-group multilabel utilities,copied,,
8
+ src/paper_ai_diffraction/eval/evaluate_calibration_metrics.py,source,Calibration metrics evaluation,copied,,
9
+ src/paper_ai_diffraction/eval/evaluate_split_head_validity.py,source,Split-head validity evaluation,copied,,
10
+ src/paper_ai_diffraction/topology/compare_325_failure_modes.py,source,Topology/failure-mode comparison,copied,,
11
+ src/paper_ai_diffraction/topology/plot_extinction_topology_flow.py,source,Topology-flow figure generation,copied,,
12
+ src/paper_ai_diffraction/topology/analyze_topological_error_distance.py,source,Topology distance summary analysis,copied,,
13
+ src/paper_ai_diffraction/topology/plot_topological_error_distance.py,source,Topology distance plotting,copied,,
14
+ src/paper_ai_diffraction/eval/plot_calibration_sweep.py,source,Calibration sweep plotting,copied,,
15
+ assets/figure_data/stage2c_r325_temp_sweep.json,asset,Compact Stage-2c temperature sweep used for the calibration figure,copied,,
16
+ assets/figure_data/physics_pe_curve_82ept35h.json,asset,Compact checkpoint curve used for the physics-PE supplementary figure,copied,,
17
+ configs/final_mixed_2500k_dualsource.json,config,Final champion dual-source config,copied,,
18
+ configs/mixed_200k_pilot.json,config,Mixed-200k pilot config,copied,,
19
+ configs/po_1epoch.json,config,PO 1-epoch config,copied,,
20
+ configs/po_resume2.json,config,PO continuation config,copied,,
21
+ configs/mixed_2500k_merged_reference.json,config,Merged-HDF5 mixed config retained for reference,copied,,
22
+ scripts/train_final_mixed.sh,script,Canonical portable wrapper for final champion training,authored,,
23
+ scripts/train_mixed_200k.sh,script,Canonical portable wrapper for mixed-200k pilot training,authored,,
24
+ scripts/train_po_1epoch.sh,script,Canonical portable wrapper for PO 1-epoch training,authored,,
25
+ scripts/train_po_resume2.sh,script,Canonical portable wrapper for PO continuation training,authored,,
26
+ scripts/tacc_archive/train_final_mixed_2500k_dualsource_tacc.sh,script,Historical TACC launcher for final champion,archived,,
27
+ scripts/tacc_archive/train_mixed_200k_pilot_tacc.sh,script,Historical TACC launcher for mixed-200k pilot,archived,,
28
+ scripts/tacc_archive/train_po_1epoch_tacc.sh,script,Historical TACC launcher for PO 1-epoch run,archived,,
29
+ scripts/tacc_archive/train_po_resume2_tacc.sh,script,Historical TACC launcher for PO continuation,archived,,
30
+ scripts/tacc_archive/eval_final_mixed_rruff_tacc.sh,script,Historical TACC launcher for final benchmark evaluation,archived,,
31
+ scripts/tacc_archive/eval_mixed200k_topology_tacc.sh,script,Historical TACC launcher for mixed-200k topology eval,archived,,
32
+ scripts/tacc_archive/eval_wiseft_r325_tacc.sh,script,Historical TACC launcher for WiSE-FT R325 eval,archived,,
33
+ scripts/eval_rruff_325_473.sh,script,Paper-facing benchmark evaluation wrapper,authored,,
34
+ scripts/make_topology_flow_figure.sh,script,Paper-facing topology-flow wrapper,authored,,
35
+ scripts/make_topology_distance_figure.sh,script,Paper-facing topology-distance wrapper,authored,,
36
+ scripts/make_calibration_figure.sh,script,Paper-facing calibration sweep wrapper,authored,,
37
+ scripts/make_curriculum_real_holdout.py,script,Paper-facing Figure 2 generator from bundled benchmark values,adapted,,
38
+ scripts/make_stage_decoder_tradeoffs_rruff473.py,script,Paper-facing Figure 3 generator from bundled benchmark values,adapted,,
39
+ scripts/make_physics_pe_q2_ruler.py,script,Paper-facing S8 generator from bundled checkpoint-curve JSON,authored,,
40
+ scripts/reconstruct_rruff_473.py,script,Public algorithm for reconstructing the frozen RRUFF-473 benchmark from upstream manifest and XY data,adapted,,
41
+ scripts/build_rruff_325_from_473.py,script,Public deterministic builder for RRUFF-325 from frozen RRUFF-473,adapted,,
42
+ scripts/make_main_tables.py,script,Paper-facing table-generation script,authored,,
43
+ results/,directory,Compact JSON outputs used in paper; generated figures are intentionally excluded from git,copied,,
44
+ docs/TRAINING.md,doc,Paper-facing training instructions,authored,,
45
+ docs/EVALUATION.md,doc,Paper-facing evaluation instructions,authored,,
46
+ docs/FIGURES.md,doc,Paper-facing figure instructions,authored,,
47
+ docs/BENCHMARKS.md,doc,Benchmark-construction and non-redistribution notes for RRUFF-473 and RRUFF-325,authored,,
48
+ assets/topology/extinction_group_adjacency.json,asset,Condensed extinction-group topology graph,copied,,
49
+ reproducibility/checkpoint_manifest.csv,metadata,Checkpoint linkage manifest,generated,,
50
+ reproducibility/dataset_manifest.csv,metadata,Dataset and external-asset manifest,generated,,
51
+ notebooks/provenance/wandb_confusion_matrix_from_table.ipynb,notebook,Provenance notebook for the supplemental ViT confusion matrix from a W&B table artifact,adapted,,
52
+ notebooks/provenance/supplemental_vit_inference.ipynb,notebook,Supplemental ViT batch inference — runs any of the 11 supplemental checkpoints against a RRUFF benchmark HDF5 and reports top-1/3/5 accuracy,authored,,
53
+ notebooks/provenance/csv_to_db_with_bravais_vit_model.ipynb,notebook,Supplemental Fig S5 — VIT attention overlay with HKL annotations using checkpoint pi7r8pah,adapted,,
54
+ assets/figure_data/1k_structures.csv,asset,1000 crystal structures (structure_id/space group/unit cell/atoms); source for runtime-generated intrep_crystal_xrd.db,copied,,
55
+ assets/figure_data/interp_metadata_clean.csv,asset,Maps HDF5 index to structure_id for the interpretability subset; used by csv_to_db notebook,copied,,
56
+ assets/figure_data/conf_mat_top5_copper-sweep-1_table_103_ff53214644fd32c50e63.table.json,asset,99x99 W&B confusion matrix table artifact (run 76zuzb4h); full artifact filename; primary backing Fig S3; used by wandb_confusion_matrix_from_table.ipynb,copied,,
57
+ src/paper_ai_diffraction/core/rt_model.py,source,Regular Transformer (RT) model definition for supplemental checkpoints,adapted,,
58
+ notebooks/provenance/supplemental_rt_inference.ipynb,notebook,Supplemental RT batch inference — runs an RT checkpoint against a RRUFF benchmark HDF5 and reports top-1/3/5 accuracy (tab:real_ablation),adapted,,
59
+ notebooks/provenance/supplemental_rt_cross_test.ipynb,notebook,Supplemental RT 4x4 training-distribution cross-test (tab:rt_distribution),adapted,,
60
+ notebooks/provenance/supplemental_rt_attention.ipynb,notebook,Supplemental RT attention overlay with HKL annotations (fig:rt_attention),adapted,,
metadata/STAGED_INVENTORY.json ADDED
@@ -0,0 +1,32 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "staged_model_files": [
3
+ "pxrd_cnn_control_checkpoint.pt",
4
+ "xrd_model_82ept35h_best.pth",
5
+ "xrd_model_9rwv1qly_best.pth",
6
+ "xrd_model_cscjfdwk_best.pth",
7
+ "xrd_model_wwuvp1kj_best.pth"
8
+ ],
9
+ "staged_dataset_benchmark_payloads": [
10
+ "RRUFF_option1_473_with_buckets_maxnorm.hdf5",
11
+ "RRUFF_usable_plus_recoverable_325_with_labels_maxnorm.hdf5"
12
+ ],
13
+ "staged_dataset_reviewer_subset": [
14
+ "correct_case_015_Arsenopyrite__R050071-1__6130.csv",
15
+ "correct_case_015_Arsenopyrite__R050071-1__6130.json",
16
+ "ext_group_priors.csv",
17
+ "ext_group_priors.json",
18
+ "failure_case_000_Actinolite__R050336-1__5330.csv",
19
+ "failure_case_000_Actinolite__R050336-1__5330.json",
20
+ "manifest.json",
21
+ "reviewer_case_metadata.csv",
22
+ "rruff325_precomputed_inference.json"
23
+ ],
24
+ "staged_dataset_metadata": [
25
+ "ARTIFACT_PLAN.md",
26
+ "MANIFEST.csv",
27
+ "STAGED_INVENTORY.json",
28
+ "checkpoint_manifest.csv",
29
+ "dataset_manifest.csv"
30
+ ],
31
+ "missing_expected_candidates": []
32
+ }
metadata/checkpoint_manifest.csv ADDED
@@ -0,0 +1,33 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ checkpoint_role,archive_status,archive_filename,expected_local_path,notes
2
+ cnn_control,external_model_repo,pxrd_cnn_control_checkpoint.pt,external/checkpoints/pxrd_cnn_control_checkpoint.pt,1D residual CNN control from the validated in-house RRUFF bs128 lineage; selected public-facing artifact name for the NeurIPS evaluation release
3
+ stage1_base,external_archive,xrd_model_ic6gfmvm_best.pth,external/checkpoints/xrd_model_ic6gfmvm_best.pth,base initialization for later realism runs
4
+ stage2a_decoder_tradeoff,external_archive,xrd_model_h7dqlpfp_best.pth,external/checkpoints/xrd_model_h7dqlpfp_best.pth,Stage-2a checkpoint named in the manuscript decoder tradeoff analysis
5
+ stage2b_decoder_tradeoff,external_archive,xrd_model_guvsasrs_best.pth,external/checkpoints/xrd_model_guvsasrs_best.pth,Stage-2b checkpoint named in the manuscript decoder tradeoff analysis
6
+ legacy_replay_baseline,external_archive,xrd_model_wwuvp1kj_best.pth,external/checkpoints/xrd_model_wwuvp1kj_best.pth,Legacy replay-era baseline named in the supplementary checkpoint map
7
+ baseline_350k,external_archive,xrd_model_7pv3pv3y_best.pth,external/checkpoints/xrd_model_7pv3pv3y_best.pth,350k baseline named in the supplementary checkpoint map
8
+ stage2c_baseline,external_archive,xrd_model_9rwv1qly_best.pth,external/checkpoints/xrd_model_9rwv1qly_best.pth,non-PO baseline used in paper
9
+ stage2c_aux025_ablation,external_archive,xrd_model_0879pvmm_best.pth,external/checkpoints/xrd_model_0879pvmm_best.pth,Stage-2c auxiliary-loss ablation named in the supplement
10
+ po_1epoch,external_archive,xrd_model_cscjfdwk_best.pth,external/checkpoints/xrd_model_cscjfdwk_best.pth,best Top-1 PO ablation
11
+ po_resume2,external_archive,xrd_model_dsi7ehiv_best.pth,external/checkpoints/xrd_model_dsi7ehiv_best.pth,PO continuation comparison
12
+ mixed_200k_pilot,external_archive,xrd_model_eeru8svx_best.pth,external/checkpoints/xrd_model_eeru8svx_best.pth,small mixed pilot
13
+ final_mixed_champion,external_archive,xrd_model_82ept35h_best.pth,external/checkpoints/xrd_model_82ept35h_best.pth,final large mixed champion
14
+ late_uniform_precursor,external_archive,xrd_model_xgy3h6uw_best.pth,external/checkpoints/xrd_model_xgy3h6uw_best.pth,exploratory larger-uniform precursor named in the supplement
15
+ late_stage2_continuation,external_archive,xrd_model_2vegisf1_best.pth,external/checkpoints/xrd_model_2vegisf1_best.pth,exploratory larger-uniform continuation named in the supplement
16
+ wiseft_a03_optional,external_archive,xrd_model_wiseft_9rwv1qly_cscjfdwk_a03_best.pth,external/checkpoints/xrd_model_wiseft_9rwv1qly_cscjfdwk_a03_best.pth,optional WiSE-FT ablation described as exploratory rather than part of the main manuscript lineage
17
+ wiseft_a05_optional,external_archive,xrd_model_wiseft_9rwv1qly_cscjfdwk_a05_best.pth,external/checkpoints/xrd_model_wiseft_9rwv1qly_cscjfdwk_a05_best.pth,optional WiSE-FT ablation described as exploratory rather than part of the main manuscript lineage
18
+ supplemental_vt_balanced,external_archive,xrd_model_tite66yo.pth,external/checkpoints/xrd_model_tite66yo.pth,Table S7 VT-Balanced; wandb_run_id=tite66yo; trained on 10M PyXtal balanced label distribution (5–90°)
19
+ supplemental_vt_icsd,external_archive,xrd_model_k0it6fz2.pth,external/checkpoints/xrd_model_k0it6fz2.pth,Table S7 VT-ICSD; wandb_run_id=k0it6fz2; trained on 10M PyXtal ICSD label distribution (5–90°)
20
+ supplemental_vt_rruff,external_archive,xrd_model_l0yy9wv9.pth,external/checkpoints/xrd_model_l0yy9wv9.pth,Table S7 VT-RRUFF; wandb_run_id=l0yy9wv9; trained on 10M PyXtal RRUFF label distribution (5–90°)
21
+ supplemental_vt_augmented,external_archive,xrd_model_cpv8k8ha.pth,external/checkpoints/xrd_model_cpv8k8ha.pth,Table S7 VT-Augmented; wandb_run_id=cpv8k8ha; trained on 10M PyXtal augmented RRUFF-type distribution (5–90°)
22
+ supplemental_vt_reduced_reflection,external_archive,xrd_model_smqmqi14.pth,external/checkpoints/xrd_model_smqmqi14.pth,Tables S4/S5 VT-Reduced Reflection and Fig S3 confusion matrix source; wandb_run_id=smqmqi14; spec_length=701 (10–80°); crashed run
23
+ supplemental_vt_full_reflection,external_archive,xrd_model_mth7zg2w.pth,external/checkpoints/xrd_model_mth7zg2w.pth,Tables S3/S5 VT-Full Reflection; wandb_run_id=mth7zg2w; spec_length=1000; crashed run
24
+ supplemental_vt_reduced_pyxtal,external_archive,xrd_model_bj1m83oz.pth,external/checkpoints/xrd_model_bj1m83oz.pth,Tables S4/S5 VT-Reduced PyXtal; wandb_run_id=bj1m83oz; spec_length=700 (10–80°)
25
+ supplemental_vt_full_pyxtal,external_archive,xrd_model_3zmiyil8.pth,external/checkpoints/xrd_model_3zmiyil8.pth,Tables S3/S5 VT-Full PyXtal; wandb_run_id=3zmiyil8; spec_length=1000; use_rope=True
26
+ supplemental_vt_biased_reflection,external_archive,xrd_model_f3sdux88.pth,external/checkpoints/xrd_model_f3sdux88.pth,Table S3 VT-Biased Reflection; wandb_run_id=f3sdux88; ICSD-biased label distribution; spec_length=1000; crashed run
27
+ supplemental_vt_biased_pyxtal,external_archive,xrd_model_kd1znx23.pth,external/checkpoints/xrd_model_kd1znx23.pth,Table S3 VT-Biased PyXtal; wandb_run_id=kd1znx23; ICSD-biased label distribution; spec_length=1000
28
+ supplemental_vt_fig_s5_attention,external_archive,xrd_model_pi7r8pah.pth,external/checkpoints/xrd_model_pi7r8pah.pth,Fig S5 VIT attention overlay with HKL annotations; wandb_run_id=pi7r8pah; trained on augmented RRUFF-type distribution; spec_length=8501
29
+ rt_distribution_balanced,external_archive,xrd_model_yv1m76u6.pth,external/checkpoints/xrd_model_yv1m76u6.pth,RT-Balanced checkpoint for tab:rt_distribution; depth=8 num_heads=4 spec_length=8501
30
+ rt_distribution_icsd,external_archive,xrd_model_4hv17ttu.pth,external/checkpoints/xrd_model_4hv17ttu.pth,RT-ICSD checkpoint for tab:rt_distribution and real-RRUFF ablation notebook; depth=6 num_heads=4 spec_length=8501
31
+ rt_distribution_rruff,external_archive,xrd_model_hwixtnv7.pth,external/checkpoints/xrd_model_hwixtnv7.pth,RT-RRUFF checkpoint for tab:rt_distribution and fig:rt_attention; depth=6 num_heads=4 spec_length=8501
32
+ rt_distribution_augmented,external_archive,xrd_model_mq1l94p7.pth,external/checkpoints/xrd_model_mq1l94p7.pth,RT-Augmented checkpoint for tab:rt_distribution and real-RRUFF ablation notebook; depth=6 num_heads=4 spec_length=8501
33
+ rt_real_ablation_16heads,external_archive,xrd_model_7brb1pir.pth,external/checkpoints/xrd_model_7brb1pir.pth,RT 16-head ablation checkpoint for the supplemental real-RRUFF ablation notebook; depth=6 num_heads=16 spec_length=8501
metadata/dataset_manifest.csv ADDED
@@ -0,0 +1,42 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ logical_name,asset_class,redistributed_in_repo,expected_env_var,example_filename,expected_local_path_or_source,required_for,notes
2
+ cnn_control_checkpoint,external_checkpoint,no,,pxrd_cnn_control_checkpoint.pt,external/checkpoints/pxrd_cnn_control_checkpoint.pt,baseline comparison,Public-facing CNN control checkpoint for the NeurIPS evaluation release; selected from the validated in-house RRUFF bs128 lineage.
3
+ stage1_base_checkpoint,external_checkpoint,no,CHECKPOINT_PATH,xrd_model_ic6gfmvm_best.pth,external/checkpoints/xrd_model_ic6gfmvm_best.pth,training and ablation lineage,Download from external archive. This repo does not bundle checkpoints.
4
+ stage2a_checkpoint,external_checkpoint,no,,xrd_model_h7dqlpfp_best.pth,external/checkpoints/xrd_model_h7dqlpfp_best.pth,decoder tradeoff reproduction,Named manuscript checkpoint for the Stage-2a decoder comparison.
5
+ stage2b_checkpoint,external_checkpoint,no,,xrd_model_guvsasrs_best.pth,external/checkpoints/xrd_model_guvsasrs_best.pth,decoder tradeoff reproduction,Named manuscript checkpoint for the Stage-2b decoder comparison.
6
+ legacy_replay_checkpoint,external_checkpoint,no,,xrd_model_wwuvp1kj_best.pth,external/checkpoints/xrd_model_wwuvp1kj_best.pth,supplementary baselines,Named supplementary replay-era baseline checkpoint.
7
+ baseline_350k_checkpoint,external_checkpoint,no,,xrd_model_7pv3pv3y_best.pth,external/checkpoints/xrd_model_7pv3pv3y_best.pth,supplementary baselines,Named supplementary 350k baseline checkpoint.
8
+ stage2c_baseline_checkpoint,external_checkpoint,no,,xrd_model_9rwv1qly_best.pth,external/checkpoints/xrd_model_9rwv1qly_best.pth,baseline comparison,Download from external archive if reproducing the Stage-2c baseline comparison.
9
+ stage2c_aux025_checkpoint,external_checkpoint,no,,xrd_model_0879pvmm_best.pth,external/checkpoints/xrd_model_0879pvmm_best.pth,supplementary ablation,Named supplementary Stage-2c-aux025 ablation checkpoint.
10
+ po_1epoch_checkpoint,external_checkpoint,no,,xrd_model_cscjfdwk_best.pth,external/checkpoints/xrd_model_cscjfdwk_best.pth,PO ablation,Download from external archive if reproducing the PO-only ablation.
11
+ po_resume2_checkpoint,external_checkpoint,no,,xrd_model_dsi7ehiv_best.pth,external/checkpoints/xrd_model_dsi7ehiv_best.pth,PO continuation,Download from external archive if reproducing the continuation comparison.
12
+ mixed200k_checkpoint,external_checkpoint,no,,xrd_model_eeru8svx_best.pth,external/checkpoints/xrd_model_eeru8svx_best.pth,mixed pilot,Download from external archive if rerunning mixed-200k checkpoint evaluation.
13
+ final_mixed2500k_checkpoint,external_checkpoint,no,CHECKPOINT_PATH,xrd_model_82ept35h_best.pth,external/checkpoints/xrd_model_82ept35h_best.pth,main paper evaluation and figures,Default checkpoint path expected by the paper-facing evaluation wrapper.
14
+ late_uniform_precursor_checkpoint,external_checkpoint,no,,xrd_model_xgy3h6uw_best.pth,external/checkpoints/xrd_model_xgy3h6uw_best.pth,exploratory supplement continuation family,Named supplementary larger-uniform precursor checkpoint.
15
+ late_stage2_continuation_checkpoint,external_checkpoint,no,,xrd_model_2vegisf1_best.pth,external/checkpoints/xrd_model_2vegisf1_best.pth,exploratory supplement continuation family,Named supplementary continuation checkpoint.
16
+ standard_trainready_h5,external_dataset,no,STANDARD_TRAINREADY_H5,rruff_conditioned_2346k_v1_trainready.hdf5,/scratch/.../rruff_conditioned_2346k_v1_trainready.hdf5,training only,External trainready dataset not redistributed in this repo.
17
+ po_trainready_h5,external_dataset,no,PO_TRAINREADY_H5,rruff_conditioned_500k_po_v1_trainready.hdf5,/scratch/.../rruff_conditioned_500k_po_v1_trainready.hdf5,training only,External trainready dataset not redistributed in this repo.
18
+ rruff325_h5,external_dataset,no,RRUFF_325_H5,RRUFF_usable_plus_recoverable_325_with_labels_maxnorm.hdf5,/work2/.../RRUFF_usable_plus_recoverable_325_with_labels_maxnorm.hdf5,evaluation,Primary harsh real-data benchmark. Required by `scripts/eval_rruff_325_473.sh`.
19
+ rruff473_h5,external_dataset,no,RRUFF_473_H5,RRUFF_option1_473_with_buckets_maxnorm.hdf5,/work2/.../RRUFF_option1_473_with_buckets_maxnorm.hdf5,evaluation,Broader curated real-data benchmark. Required by `scripts/eval_rruff_325_473.sh`.
20
+ prior_h5,external_dataset,no,PRIOR_H5,rruff_conditioned_2346k_v1_trainready.hdf5,/scratch/.../rruff_conditioned_2346k_v1_trainready.hdf5,evaluation,External prior/train HDF5 used for prior-derived calibration in the evaluation wrapper.
21
+ canonical_extinction_csv,external_dataset,no,CANONICAL_CSV,canonical_extinction_to_space_group.csv,/scratch/.../ai-diffraction/Code/Post_Processing/canonical_extinction_to_space_group.csv,topology figures,External canonical extinction-group table required by `scripts/make_topology_flow_figure.sh`. The validated Stampede example lived in an `ai-diffraction` checkout on scratch.
22
+ topology_graph_json,bundled_asset,yes,GRAPH_JSON,extinction_group_adjacency.json,assets/topology/extinction_group_adjacency.json,topology figures,Bundled in this repo.
23
+ mixed200k_failure_json,bundled_result_json,yes,MIXED200K_FAILURE_JSON,mixed200k_compare_325_failure_modes_eeru8svx.json,results/mixed200k_compare_325_failure_modes_eeru8svx.json,topology figures,Bundled compact paper result artifact.
24
+ mixed2500k_failure_json,bundled_result_json,yes,FAILURE_JSON,mixed2500k_compare_325_failure_modes_655279.json,results/mixed2500k_compare_325_failure_modes_655279.json,topology figures,Bundled compact paper result artifact.
25
+ calibration_sweep_json,bundled_result_json,yes,CAL_SWEEP_JSON,stage2c_r325_temp_sweep.json,assets/figure_data/stage2c_r325_temp_sweep.json,calibration figure,Bundled compact Stage-2c sweep JSON used by `scripts/make_calibration_figure.sh` by default.
26
+ physics_pe_curve_json,bundled_result_json,yes,PHYSICS_PE_CURVE_JSON,physics_pe_curve_82ept35h.json,assets/figure_data/physics_pe_curve_82ept35h.json,S8 physics PE figure,Bundled compact checkpoint-curve JSON used by `scripts/make_physics_pe_q2_ruler.py`.
27
+ supplemental_train_pyxtal_balanced,external_dataset,no,,10Million_pyxtal_RRUFFF-type_repacked.hdf5,/scratch/<USER>/Data/10Million_pyxtal_RRUFFF-type_repacked.hdf5,supplemental Table S7 (tite66yo),10M PyXtal balanced label distribution; 5–90°; spec_length=8501
28
+ supplemental_train_pyxtal_icsd,external_dataset,no,,10mil_pyxtal_rruff-type_icsd_distrib_repacked.hdf5,/scratch/<USER>/Data/10mil_pyxtal_rruff-type_icsd_distrib_repacked.hdf5,supplemental Table S7 (k0it6fz2),10M PyXtal ICSD label distribution; 5–90°; spec_length=8501
29
+ supplemental_train_pyxtal_rruff,external_dataset,no,,10mil_pyxtal_rruff-type_rruff_distrib_repacked.hdf5,/scratch/<USER>/Data/10mil_pyxtal_rruff-type_rruff_distrib_repacked.hdf5,supplemental Table S7 (l0yy9wv9),10M PyXtal RRUFF label distribution; 5–90°; spec_length=8501
30
+ supplemental_train_pyxtal_aug,external_dataset,no,,rruff_aug_10M_trainready_repacked.hdf5,/scratch/<USER>/Data/rruff_aug_10M_trainready_repacked.hdf5,supplemental Table S7 (cpv8k8ha) and Fig S5 (pi7r8pah),10M PyXtal augmented RRUFF-type distribution; 5–90°; spec_length=8501
31
+ supplemental_train_reflection_highres,external_dataset,no,,10Million_ref_ext_highres.h5,/scratch/<USER>/Data/10Million_ref_ext_highres.h5,supplemental Tables S3/S4/S5 (smqmqi14 spec_length=701; mth7zg2w spec_length=1000),10M reflection high-res
32
+ supplemental_train_pyxtal_cleaned,external_dataset,no,,10million_pyxtal_repacked_cleaned.hdf5,/scratch/<USER>/Data/10million_pyxtal_repacked_cleaned.hdf5,supplemental Tables S4/S5 (bj1m83oz spec_length=700; 3zmiyil8 spec_length=1000),10M PyXtal cleaned
33
+ supplemental_train_reflection_icsd_bias,external_dataset,no,,10M_W_ICSD_Bias_reflection_repack.hdf5,/scratch/<USER>/Data/10M_W_ICSD_Bias_reflection_repack.hdf5,supplemental Table S3 (f3sdux88),10M reflection ICSD-biased label distribution
34
+ supplemental_train_pyxtal_icsd_bias,external_dataset,no,,10M_W_ICSD_Bias_pyxtal_repack.hdf5,/scratch/<USER>/Data/10M_W_ICSD_Bias_pyxtal_repack.hdf5,supplemental Table S3 (kd1znx23),10M PyXtal ICSD-biased label distribution
35
+ supplemental_test_rruff_reduced_700,external_dataset,no,,RRUFF_norm_reduced_range_10_80_deg_700_intensity.hdf5,/scratch/<USER>/Data/RRUFF_norm_reduced_range_10_80_deg_700_intensity.hdf5,supplemental Table S5 inference run e0sq2tus (bj1m83oz),Reduced range RRUFF; 700 pts; 10–80°
36
+ supplemental_test_rruff_reduced_701,external_dataset,no,,RRUFF_norm_reduced_range_10_80_deg_701_intensity.hdf5,/scratch/<USER>/Data/RRUFF_norm_reduced_range_10_80_deg_701_intensity.hdf5,supplemental Table S5 inference runs njur2oso/x9ye2mrx (smqmqi14),Reduced range RRUFF; 701 pts; 10–80°
37
+ supplemental_test_rruff_full_1000,external_dataset,no,,RRUFF_norm_uniform_sampling_5_90_deg_1000_intensity.hdf5,/scratch/<USER>/Data/RRUFF_norm_uniform_sampling_5_90_deg_1000_intensity.hdf5,supplemental Table S5 inference runs hrhk538h (mth7zg2w) and ollqf5gm (3zmiyil8),Full range RRUFF; 1000 pts; 5–90°
38
+ supplemental_test_rruff_low_bkg_8501,external_dataset,no,,RRUFF_low_bkg_full_intensity.hdf5,/scratch/<USER>/Data/RRUFF_low_bkg_full_intensity.hdf5,supplemental Table S7 inference runs,Full RRUFF low background; 8501 pts; 5–90°
39
+ supplemental_test_rruff_interp,external_dataset,no,,RRUFF_interp_data.hdf5,/scratch/<USER>/Data/RRUFF_interp_data.hdf5,supplemental Fig S5 attention overlay inference,RRUFF interpretability set with structure_id dataset
40
+ supplemental_1k_structures_csv,bundled_asset,yes,,1k_structures.csv,assets/figure_data/1k_structures.csv,supplemental Fig S5 (csv_to_db_with_bravais_vit_model.ipynb),1000 crystal structures with structure_id/space group/unit cell/atoms; source for intrep_crystal_xrd.db
41
+ supplemental_interp_metadata_csv,bundled_asset,yes,,interp_metadata_clean.csv,assets/figure_data/interp_metadata_clean.csv,supplemental Fig S5 (csv_to_db_with_bravais_vit_model.ipynb),Maps HDF5 index to structure_id for interpretability subset
42
+ supplemental_confusion_matrix_json,bundled_asset,yes,,conf_mat_top5_copper-sweep-1_table_103_ff53214644fd32c50e63.table.json,assets/figure_data/conf_mat_top5_copper-sweep-1_table_103_ff53214644fd32c50e63.table.json,supplemental Fig S3 (wandb_confusion_matrix_from_table.ipynb),99x99 W&B confusion matrix table artifact from run 76zuzb4h; columns: Actual/Predicted/nPredictions
reviewer_subset/correct_case_015_Arsenopyrite__R050071-1__6130.csv ADDED
The diff for this file is too large to render. See raw diff
 
reviewer_subset/correct_case_015_Arsenopyrite__R050071-1__6130.json ADDED
@@ -0,0 +1,27 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "tag": "correct_case",
3
+ "benchmark_index": 15,
4
+ "case_id": "Arsenopyrite__R050071-1__6130",
5
+ "mineral": "Arsenopyrite",
6
+ "severity": "recoverable",
7
+ "fit_bucket": "fail",
8
+ "rwp": 0.12289926409721375,
9
+ "true_eg": 99,
10
+ "true_space_group": 2,
11
+ "true_space_group_symbol": "P-1",
12
+ "true_space_group_source": "rruff_fit_dif",
13
+ "paper_model_summary": {
14
+ "index": 15,
15
+ "true_eg": 99,
16
+ "pred_eg": 99,
17
+ "top5_eg": [
18
+ 99,
19
+ 4,
20
+ 6,
21
+ 5,
22
+ 17
23
+ ],
24
+ "correct": true
25
+ },
26
+ "csv_path": "assets/reviewer_examples/correct_case_015_Arsenopyrite__R050071-1__6130.csv"
27
+ }
reviewer_subset/ext_group_priors.csv ADDED
@@ -0,0 +1,100 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ext_group,count_smoothed,probability,log_prior
2
+ 1,5950.0,0.0031674730857948682,-5.7548211427917035
3
+ 2,82459.0,0.04389691818177462,-3.1259111622359455
4
+ 3,28565.0,0.01520653255390427,-4.186030170172816
5
+ 4,230649.0,0.12278563021268916,-2.0973152879281907
6
+ 5,168237.0,0.08956070076216323,-2.412838662782828
7
+ 6,148099.0,0.0788402683248965,-2.5403313932927483
8
+ 7,1.0,5.323484177806501e-07,-14.445967641331379
9
+ 8,1.0,5.323484177806501e-07,-14.445967641331379
10
+ 9,1505.0,0.0008011843687598784,-7.129419464148403
11
+ 10,33151.0,0.017647882397846332,-4.0371394802598335
12
+ 11,5892.0,0.0031365968775635905,-5.764616863748858
13
+ 12,19579.0,0.010422849671727349,-4.563754799074106
14
+ 13,1.0,5.323484177806501e-07,-14.445967641331379
15
+ 14,10481.0,0.005579543766758993,-5.188648268161202
16
+ 15,1460.0,0.0007772286899597491,-7.159775926628997
17
+ 16,13506.0,0.00718989773054546,-4.935078331196592
18
+ 17,182525.0,0.09716689495541316,-2.331325212407093
19
+ 18,21025.0,0.011192625483838169,-4.49250015649023
20
+ 19,2964.0,0.001577880710301847,-6.451672654915401
21
+ 20,1.0,5.323484177806501e-07,-14.445967641331379
22
+ 21,31892.0,0.016977655739860494,-4.075857167708993
23
+ 22,34899.0,0.01857842743212691,-3.985754186846503
24
+ 23,1499.0,0.0007979902782531945,-7.133414143228782
25
+ 24,5963.0,0.0031743936152260166,-5.752638652208273
26
+ 25,1.0,5.323484177806501e-07,-14.445967641331379
27
+ 26,4503.0,0.0023971649252662675,-6.033468521029808
28
+ 27,1.0,5.323484177806501e-07,-14.445967641331379
29
+ 28,1.0,5.323484177806501e-07,-14.445967641331379
30
+ 29,47891.0,0.025494698075933114,-3.6692847670266815
31
+ 30,7635.0,0.004064480169755263,-5.505469423678647
32
+ 31,5930.0,0.003156826117439255,-5.758188149339608
33
+ 32,4512.0,0.002401956061026293,-6.031471848153484
34
+ 33,12108.0,0.0064456746424881115,-5.04434597118977
35
+ 34,5979.0,0.003182911189910507,-5.749959032450475
36
+ 35,4456.0,0.002372144549630577,-6.043960859724259
37
+ 36,1.0,5.323484177806501e-07,-14.445967641331379
38
+ 37,1.0,5.323484177806501e-07,-14.445967641331379
39
+ 38,4527.0,0.002409941287293003,-6.02815289389542
40
+ 39,3051.0,0.0016241950226487634,-6.42274295661471
41
+ 40,4447.0,0.002367353413870551,-6.045982650820683
42
+ 41,1537.0,0.0008182195181288592,-7.1083798977927835
43
+ 42,1548.0,0.0008240753507244463,-7.101248587181707
44
+ 43,1.0,5.323484177806501e-07,-14.445967641331379
45
+ 44,1589.0,0.000845901635853453,-7.075107474794663
46
+ 45,9140.0,0.004865664538515142,-5.325551976883183
47
+ 46,1.0,5.323484177806501e-07,-14.445967641331379
48
+ 47,1478.0,0.0007868109614798008,-7.147522539823232
49
+ 48,1.0,5.323484177806501e-07,-14.445967641331379
50
+ 49,1470.0,0.0007825521741375557,-7.152949961558597
51
+ 50,1.0,5.323484177806501e-07,-14.445967641331379
52
+ 51,3019.0,0.0016071598732797826,-6.4332867116245405
53
+ 52,10481.0,0.005579543766758993,-5.188648268161202
54
+ 53,2965.0,0.0015784130587196275,-6.451335329899553
55
+ 54,18210.0,0.009694064687785638,-4.636241468609725
56
+ 55,6071.0,0.0032318872443463267,-5.734689026200945
57
+ 56,1.0,5.323484177806501e-07,-14.445967641331379
58
+ 57,1.0,5.323484177806501e-07,-14.445967641331379
59
+ 58,1494.0,0.0007953285361642912,-7.136755275638617
60
+ 59,1.0,5.323484177806501e-07,-14.445967641331379
61
+ 60,5957.0,0.0031711995247193326,-5.753645363702692
62
+ 61,1.0,5.323484177806501e-07,-14.445967641331379
63
+ 62,7560.0,0.004024554038421715,-5.5153411721578
64
+ 63,30038.0,0.015990681773295167,-4.1357491155658535
65
+ 64,1.0,5.323484177806501e-07,-14.445967641331379
66
+ 65,4511.0,0.0024014237126085127,-6.031693503922983
67
+ 66,1.0,5.323484177806501e-07,-14.445967641331379
68
+ 67,1.0,5.323484177806501e-07,-14.445967641331379
69
+ 68,12198.0,0.00649358600008837,-5.036940358474977
70
+ 69,6001.0,0.003194622855101681,-5.746286240341866
71
+ 70,4564.0,0.002429638178750887,-6.020012930349413
72
+ 71,51016.0,0.027158286881497646,-3.6060730533362717
73
+ 72,5857.0,0.0031179646829412678,-5.770574835241597
74
+ 73,59895.0,0.03188500848297204,-3.445619333165948
75
+ 74,9102.0,0.004845435298639477,-5.329718192754792
76
+ 75,112691.0,0.05999087554811924,-2.8135628025225023
77
+ 76,31764.0,0.01690951514238457,-4.079878789316278
78
+ 77,61349.0,0.0326590430824251,-3.4216334912436674
79
+ 78,1.0,5.323484177806501e-07,-14.445967641331379
80
+ 79,1.0,5.323484177806501e-07,-14.445967641331379
81
+ 80,18028.0,0.00959717727574956,-4.64628625752084
82
+ 81,3008.0,0.0016013040406841954,-6.436936956261649
83
+ 82,1526.0,0.000812363685533272,-7.115562429486977
84
+ 83,1.0,5.323484177806501e-07,-14.445967641331379
85
+ 84,1.0,5.323484177806501e-07,-14.445967641331379
86
+ 85,13454.0,0.0071622156128208665,-4.938935902745828
87
+ 86,6032.0,0.0032111256560528813,-5.741133731643587
88
+ 87,1517.0,0.0008075725497732462,-7.121477661982847
89
+ 88,1.0,5.323484177806501e-07,-14.445967641331379
90
+ 89,8945.0,0.004761856597047915,-5.347117645388559
91
+ 90,1.0,5.323484177806501e-07,-14.445967641331379
92
+ 91,1500.0,0.0007985226266709752,-7.132747254241077
93
+ 92,1.0,5.323484177806501e-07,-14.445967641331379
94
+ 93,30192.0,0.01607266342963339,-4.130635373723048
95
+ 94,7591.0,0.004041056839372915,-5.51124902731461
96
+ 95,3011.0,0.0016029010859375374,-6.435940112849646
97
+ 96,1.0,5.323484177806501e-07,-14.445967641331379
98
+ 97,25513.0,0.013581805182837726,-4.299024236169163
99
+ 98,1.0,5.323484177806501e-07,-14.445967641331379
100
+ 99,161279.0,0.08585662047124547,-2.455076577882286
reviewer_subset/ext_group_priors.json ADDED
@@ -0,0 +1,601 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "prior_h5": "external/datasets/rruff_conditioned_2346k_v1_trainready.hdf5",
3
+ "smoothing": 1.0,
4
+ "num_classes": 99,
5
+ "rows": [
6
+ {
7
+ "ext_group": 1,
8
+ "count_smoothed": 5950.0,
9
+ "probability": 0.0031674730857948682,
10
+ "log_prior": -5.7548211427917035
11
+ },
12
+ {
13
+ "ext_group": 2,
14
+ "count_smoothed": 82459.0,
15
+ "probability": 0.04389691818177462,
16
+ "log_prior": -3.1259111622359455
17
+ },
18
+ {
19
+ "ext_group": 3,
20
+ "count_smoothed": 28565.0,
21
+ "probability": 0.01520653255390427,
22
+ "log_prior": -4.186030170172816
23
+ },
24
+ {
25
+ "ext_group": 4,
26
+ "count_smoothed": 230649.0,
27
+ "probability": 0.12278563021268916,
28
+ "log_prior": -2.0973152879281907
29
+ },
30
+ {
31
+ "ext_group": 5,
32
+ "count_smoothed": 168237.0,
33
+ "probability": 0.08956070076216323,
34
+ "log_prior": -2.412838662782828
35
+ },
36
+ {
37
+ "ext_group": 6,
38
+ "count_smoothed": 148099.0,
39
+ "probability": 0.0788402683248965,
40
+ "log_prior": -2.5403313932927483
41
+ },
42
+ {
43
+ "ext_group": 7,
44
+ "count_smoothed": 1.0,
45
+ "probability": 5.323484177806501e-07,
46
+ "log_prior": -14.445967641331379
47
+ },
48
+ {
49
+ "ext_group": 8,
50
+ "count_smoothed": 1.0,
51
+ "probability": 5.323484177806501e-07,
52
+ "log_prior": -14.445967641331379
53
+ },
54
+ {
55
+ "ext_group": 9,
56
+ "count_smoothed": 1505.0,
57
+ "probability": 0.0008011843687598784,
58
+ "log_prior": -7.129419464148403
59
+ },
60
+ {
61
+ "ext_group": 10,
62
+ "count_smoothed": 33151.0,
63
+ "probability": 0.017647882397846332,
64
+ "log_prior": -4.0371394802598335
65
+ },
66
+ {
67
+ "ext_group": 11,
68
+ "count_smoothed": 5892.0,
69
+ "probability": 0.0031365968775635905,
70
+ "log_prior": -5.764616863748858
71
+ },
72
+ {
73
+ "ext_group": 12,
74
+ "count_smoothed": 19579.0,
75
+ "probability": 0.010422849671727349,
76
+ "log_prior": -4.563754799074106
77
+ },
78
+ {
79
+ "ext_group": 13,
80
+ "count_smoothed": 1.0,
81
+ "probability": 5.323484177806501e-07,
82
+ "log_prior": -14.445967641331379
83
+ },
84
+ {
85
+ "ext_group": 14,
86
+ "count_smoothed": 10481.0,
87
+ "probability": 0.005579543766758993,
88
+ "log_prior": -5.188648268161202
89
+ },
90
+ {
91
+ "ext_group": 15,
92
+ "count_smoothed": 1460.0,
93
+ "probability": 0.0007772286899597491,
94
+ "log_prior": -7.159775926628997
95
+ },
96
+ {
97
+ "ext_group": 16,
98
+ "count_smoothed": 13506.0,
99
+ "probability": 0.00718989773054546,
100
+ "log_prior": -4.935078331196592
101
+ },
102
+ {
103
+ "ext_group": 17,
104
+ "count_smoothed": 182525.0,
105
+ "probability": 0.09716689495541316,
106
+ "log_prior": -2.331325212407093
107
+ },
108
+ {
109
+ "ext_group": 18,
110
+ "count_smoothed": 21025.0,
111
+ "probability": 0.011192625483838169,
112
+ "log_prior": -4.49250015649023
113
+ },
114
+ {
115
+ "ext_group": 19,
116
+ "count_smoothed": 2964.0,
117
+ "probability": 0.001577880710301847,
118
+ "log_prior": -6.451672654915401
119
+ },
120
+ {
121
+ "ext_group": 20,
122
+ "count_smoothed": 1.0,
123
+ "probability": 5.323484177806501e-07,
124
+ "log_prior": -14.445967641331379
125
+ },
126
+ {
127
+ "ext_group": 21,
128
+ "count_smoothed": 31892.0,
129
+ "probability": 0.016977655739860494,
130
+ "log_prior": -4.075857167708993
131
+ },
132
+ {
133
+ "ext_group": 22,
134
+ "count_smoothed": 34899.0,
135
+ "probability": 0.01857842743212691,
136
+ "log_prior": -3.985754186846503
137
+ },
138
+ {
139
+ "ext_group": 23,
140
+ "count_smoothed": 1499.0,
141
+ "probability": 0.0007979902782531945,
142
+ "log_prior": -7.133414143228782
143
+ },
144
+ {
145
+ "ext_group": 24,
146
+ "count_smoothed": 5963.0,
147
+ "probability": 0.0031743936152260166,
148
+ "log_prior": -5.752638652208273
149
+ },
150
+ {
151
+ "ext_group": 25,
152
+ "count_smoothed": 1.0,
153
+ "probability": 5.323484177806501e-07,
154
+ "log_prior": -14.445967641331379
155
+ },
156
+ {
157
+ "ext_group": 26,
158
+ "count_smoothed": 4503.0,
159
+ "probability": 0.0023971649252662675,
160
+ "log_prior": -6.033468521029808
161
+ },
162
+ {
163
+ "ext_group": 27,
164
+ "count_smoothed": 1.0,
165
+ "probability": 5.323484177806501e-07,
166
+ "log_prior": -14.445967641331379
167
+ },
168
+ {
169
+ "ext_group": 28,
170
+ "count_smoothed": 1.0,
171
+ "probability": 5.323484177806501e-07,
172
+ "log_prior": -14.445967641331379
173
+ },
174
+ {
175
+ "ext_group": 29,
176
+ "count_smoothed": 47891.0,
177
+ "probability": 0.025494698075933114,
178
+ "log_prior": -3.6692847670266815
179
+ },
180
+ {
181
+ "ext_group": 30,
182
+ "count_smoothed": 7635.0,
183
+ "probability": 0.004064480169755263,
184
+ "log_prior": -5.505469423678647
185
+ },
186
+ {
187
+ "ext_group": 31,
188
+ "count_smoothed": 5930.0,
189
+ "probability": 0.003156826117439255,
190
+ "log_prior": -5.758188149339608
191
+ },
192
+ {
193
+ "ext_group": 32,
194
+ "count_smoothed": 4512.0,
195
+ "probability": 0.002401956061026293,
196
+ "log_prior": -6.031471848153484
197
+ },
198
+ {
199
+ "ext_group": 33,
200
+ "count_smoothed": 12108.0,
201
+ "probability": 0.0064456746424881115,
202
+ "log_prior": -5.04434597118977
203
+ },
204
+ {
205
+ "ext_group": 34,
206
+ "count_smoothed": 5979.0,
207
+ "probability": 0.003182911189910507,
208
+ "log_prior": -5.749959032450475
209
+ },
210
+ {
211
+ "ext_group": 35,
212
+ "count_smoothed": 4456.0,
213
+ "probability": 0.002372144549630577,
214
+ "log_prior": -6.043960859724259
215
+ },
216
+ {
217
+ "ext_group": 36,
218
+ "count_smoothed": 1.0,
219
+ "probability": 5.323484177806501e-07,
220
+ "log_prior": -14.445967641331379
221
+ },
222
+ {
223
+ "ext_group": 37,
224
+ "count_smoothed": 1.0,
225
+ "probability": 5.323484177806501e-07,
226
+ "log_prior": -14.445967641331379
227
+ },
228
+ {
229
+ "ext_group": 38,
230
+ "count_smoothed": 4527.0,
231
+ "probability": 0.002409941287293003,
232
+ "log_prior": -6.02815289389542
233
+ },
234
+ {
235
+ "ext_group": 39,
236
+ "count_smoothed": 3051.0,
237
+ "probability": 0.0016241950226487634,
238
+ "log_prior": -6.42274295661471
239
+ },
240
+ {
241
+ "ext_group": 40,
242
+ "count_smoothed": 4447.0,
243
+ "probability": 0.002367353413870551,
244
+ "log_prior": -6.045982650820683
245
+ },
246
+ {
247
+ "ext_group": 41,
248
+ "count_smoothed": 1537.0,
249
+ "probability": 0.0008182195181288592,
250
+ "log_prior": -7.1083798977927835
251
+ },
252
+ {
253
+ "ext_group": 42,
254
+ "count_smoothed": 1548.0,
255
+ "probability": 0.0008240753507244463,
256
+ "log_prior": -7.101248587181707
257
+ },
258
+ {
259
+ "ext_group": 43,
260
+ "count_smoothed": 1.0,
261
+ "probability": 5.323484177806501e-07,
262
+ "log_prior": -14.445967641331379
263
+ },
264
+ {
265
+ "ext_group": 44,
266
+ "count_smoothed": 1589.0,
267
+ "probability": 0.000845901635853453,
268
+ "log_prior": -7.075107474794663
269
+ },
270
+ {
271
+ "ext_group": 45,
272
+ "count_smoothed": 9140.0,
273
+ "probability": 0.004865664538515142,
274
+ "log_prior": -5.325551976883183
275
+ },
276
+ {
277
+ "ext_group": 46,
278
+ "count_smoothed": 1.0,
279
+ "probability": 5.323484177806501e-07,
280
+ "log_prior": -14.445967641331379
281
+ },
282
+ {
283
+ "ext_group": 47,
284
+ "count_smoothed": 1478.0,
285
+ "probability": 0.0007868109614798008,
286
+ "log_prior": -7.147522539823232
287
+ },
288
+ {
289
+ "ext_group": 48,
290
+ "count_smoothed": 1.0,
291
+ "probability": 5.323484177806501e-07,
292
+ "log_prior": -14.445967641331379
293
+ },
294
+ {
295
+ "ext_group": 49,
296
+ "count_smoothed": 1470.0,
297
+ "probability": 0.0007825521741375557,
298
+ "log_prior": -7.152949961558597
299
+ },
300
+ {
301
+ "ext_group": 50,
302
+ "count_smoothed": 1.0,
303
+ "probability": 5.323484177806501e-07,
304
+ "log_prior": -14.445967641331379
305
+ },
306
+ {
307
+ "ext_group": 51,
308
+ "count_smoothed": 3019.0,
309
+ "probability": 0.0016071598732797826,
310
+ "log_prior": -6.4332867116245405
311
+ },
312
+ {
313
+ "ext_group": 52,
314
+ "count_smoothed": 10481.0,
315
+ "probability": 0.005579543766758993,
316
+ "log_prior": -5.188648268161202
317
+ },
318
+ {
319
+ "ext_group": 53,
320
+ "count_smoothed": 2965.0,
321
+ "probability": 0.0015784130587196275,
322
+ "log_prior": -6.451335329899553
323
+ },
324
+ {
325
+ "ext_group": 54,
326
+ "count_smoothed": 18210.0,
327
+ "probability": 0.009694064687785638,
328
+ "log_prior": -4.636241468609725
329
+ },
330
+ {
331
+ "ext_group": 55,
332
+ "count_smoothed": 6071.0,
333
+ "probability": 0.0032318872443463267,
334
+ "log_prior": -5.734689026200945
335
+ },
336
+ {
337
+ "ext_group": 56,
338
+ "count_smoothed": 1.0,
339
+ "probability": 5.323484177806501e-07,
340
+ "log_prior": -14.445967641331379
341
+ },
342
+ {
343
+ "ext_group": 57,
344
+ "count_smoothed": 1.0,
345
+ "probability": 5.323484177806501e-07,
346
+ "log_prior": -14.445967641331379
347
+ },
348
+ {
349
+ "ext_group": 58,
350
+ "count_smoothed": 1494.0,
351
+ "probability": 0.0007953285361642912,
352
+ "log_prior": -7.136755275638617
353
+ },
354
+ {
355
+ "ext_group": 59,
356
+ "count_smoothed": 1.0,
357
+ "probability": 5.323484177806501e-07,
358
+ "log_prior": -14.445967641331379
359
+ },
360
+ {
361
+ "ext_group": 60,
362
+ "count_smoothed": 5957.0,
363
+ "probability": 0.0031711995247193326,
364
+ "log_prior": -5.753645363702692
365
+ },
366
+ {
367
+ "ext_group": 61,
368
+ "count_smoothed": 1.0,
369
+ "probability": 5.323484177806501e-07,
370
+ "log_prior": -14.445967641331379
371
+ },
372
+ {
373
+ "ext_group": 62,
374
+ "count_smoothed": 7560.0,
375
+ "probability": 0.004024554038421715,
376
+ "log_prior": -5.5153411721578
377
+ },
378
+ {
379
+ "ext_group": 63,
380
+ "count_smoothed": 30038.0,
381
+ "probability": 0.015990681773295167,
382
+ "log_prior": -4.1357491155658535
383
+ },
384
+ {
385
+ "ext_group": 64,
386
+ "count_smoothed": 1.0,
387
+ "probability": 5.323484177806501e-07,
388
+ "log_prior": -14.445967641331379
389
+ },
390
+ {
391
+ "ext_group": 65,
392
+ "count_smoothed": 4511.0,
393
+ "probability": 0.0024014237126085127,
394
+ "log_prior": -6.031693503922983
395
+ },
396
+ {
397
+ "ext_group": 66,
398
+ "count_smoothed": 1.0,
399
+ "probability": 5.323484177806501e-07,
400
+ "log_prior": -14.445967641331379
401
+ },
402
+ {
403
+ "ext_group": 67,
404
+ "count_smoothed": 1.0,
405
+ "probability": 5.323484177806501e-07,
406
+ "log_prior": -14.445967641331379
407
+ },
408
+ {
409
+ "ext_group": 68,
410
+ "count_smoothed": 12198.0,
411
+ "probability": 0.00649358600008837,
412
+ "log_prior": -5.036940358474977
413
+ },
414
+ {
415
+ "ext_group": 69,
416
+ "count_smoothed": 6001.0,
417
+ "probability": 0.003194622855101681,
418
+ "log_prior": -5.746286240341866
419
+ },
420
+ {
421
+ "ext_group": 70,
422
+ "count_smoothed": 4564.0,
423
+ "probability": 0.002429638178750887,
424
+ "log_prior": -6.020012930349413
425
+ },
426
+ {
427
+ "ext_group": 71,
428
+ "count_smoothed": 51016.0,
429
+ "probability": 0.027158286881497646,
430
+ "log_prior": -3.6060730533362717
431
+ },
432
+ {
433
+ "ext_group": 72,
434
+ "count_smoothed": 5857.0,
435
+ "probability": 0.0031179646829412678,
436
+ "log_prior": -5.770574835241597
437
+ },
438
+ {
439
+ "ext_group": 73,
440
+ "count_smoothed": 59895.0,
441
+ "probability": 0.03188500848297204,
442
+ "log_prior": -3.445619333165948
443
+ },
444
+ {
445
+ "ext_group": 74,
446
+ "count_smoothed": 9102.0,
447
+ "probability": 0.004845435298639477,
448
+ "log_prior": -5.329718192754792
449
+ },
450
+ {
451
+ "ext_group": 75,
452
+ "count_smoothed": 112691.0,
453
+ "probability": 0.05999087554811924,
454
+ "log_prior": -2.8135628025225023
455
+ },
456
+ {
457
+ "ext_group": 76,
458
+ "count_smoothed": 31764.0,
459
+ "probability": 0.01690951514238457,
460
+ "log_prior": -4.079878789316278
461
+ },
462
+ {
463
+ "ext_group": 77,
464
+ "count_smoothed": 61349.0,
465
+ "probability": 0.0326590430824251,
466
+ "log_prior": -3.4216334912436674
467
+ },
468
+ {
469
+ "ext_group": 78,
470
+ "count_smoothed": 1.0,
471
+ "probability": 5.323484177806501e-07,
472
+ "log_prior": -14.445967641331379
473
+ },
474
+ {
475
+ "ext_group": 79,
476
+ "count_smoothed": 1.0,
477
+ "probability": 5.323484177806501e-07,
478
+ "log_prior": -14.445967641331379
479
+ },
480
+ {
481
+ "ext_group": 80,
482
+ "count_smoothed": 18028.0,
483
+ "probability": 0.00959717727574956,
484
+ "log_prior": -4.64628625752084
485
+ },
486
+ {
487
+ "ext_group": 81,
488
+ "count_smoothed": 3008.0,
489
+ "probability": 0.0016013040406841954,
490
+ "log_prior": -6.436936956261649
491
+ },
492
+ {
493
+ "ext_group": 82,
494
+ "count_smoothed": 1526.0,
495
+ "probability": 0.000812363685533272,
496
+ "log_prior": -7.115562429486977
497
+ },
498
+ {
499
+ "ext_group": 83,
500
+ "count_smoothed": 1.0,
501
+ "probability": 5.323484177806501e-07,
502
+ "log_prior": -14.445967641331379
503
+ },
504
+ {
505
+ "ext_group": 84,
506
+ "count_smoothed": 1.0,
507
+ "probability": 5.323484177806501e-07,
508
+ "log_prior": -14.445967641331379
509
+ },
510
+ {
511
+ "ext_group": 85,
512
+ "count_smoothed": 13454.0,
513
+ "probability": 0.0071622156128208665,
514
+ "log_prior": -4.938935902745828
515
+ },
516
+ {
517
+ "ext_group": 86,
518
+ "count_smoothed": 6032.0,
519
+ "probability": 0.0032111256560528813,
520
+ "log_prior": -5.741133731643587
521
+ },
522
+ {
523
+ "ext_group": 87,
524
+ "count_smoothed": 1517.0,
525
+ "probability": 0.0008075725497732462,
526
+ "log_prior": -7.121477661982847
527
+ },
528
+ {
529
+ "ext_group": 88,
530
+ "count_smoothed": 1.0,
531
+ "probability": 5.323484177806501e-07,
532
+ "log_prior": -14.445967641331379
533
+ },
534
+ {
535
+ "ext_group": 89,
536
+ "count_smoothed": 8945.0,
537
+ "probability": 0.004761856597047915,
538
+ "log_prior": -5.347117645388559
539
+ },
540
+ {
541
+ "ext_group": 90,
542
+ "count_smoothed": 1.0,
543
+ "probability": 5.323484177806501e-07,
544
+ "log_prior": -14.445967641331379
545
+ },
546
+ {
547
+ "ext_group": 91,
548
+ "count_smoothed": 1500.0,
549
+ "probability": 0.0007985226266709752,
550
+ "log_prior": -7.132747254241077
551
+ },
552
+ {
553
+ "ext_group": 92,
554
+ "count_smoothed": 1.0,
555
+ "probability": 5.323484177806501e-07,
556
+ "log_prior": -14.445967641331379
557
+ },
558
+ {
559
+ "ext_group": 93,
560
+ "count_smoothed": 30192.0,
561
+ "probability": 0.01607266342963339,
562
+ "log_prior": -4.130635373723048
563
+ },
564
+ {
565
+ "ext_group": 94,
566
+ "count_smoothed": 7591.0,
567
+ "probability": 0.004041056839372915,
568
+ "log_prior": -5.51124902731461
569
+ },
570
+ {
571
+ "ext_group": 95,
572
+ "count_smoothed": 3011.0,
573
+ "probability": 0.0016029010859375374,
574
+ "log_prior": -6.435940112849646
575
+ },
576
+ {
577
+ "ext_group": 96,
578
+ "count_smoothed": 1.0,
579
+ "probability": 5.323484177806501e-07,
580
+ "log_prior": -14.445967641331379
581
+ },
582
+ {
583
+ "ext_group": 97,
584
+ "count_smoothed": 25513.0,
585
+ "probability": 0.013581805182837726,
586
+ "log_prior": -4.299024236169163
587
+ },
588
+ {
589
+ "ext_group": 98,
590
+ "count_smoothed": 1.0,
591
+ "probability": 5.323484177806501e-07,
592
+ "log_prior": -14.445967641331379
593
+ },
594
+ {
595
+ "ext_group": 99,
596
+ "count_smoothed": 161279.0,
597
+ "probability": 0.08585662047124547,
598
+ "log_prior": -2.455076577882286
599
+ }
600
+ ]
601
+ }
reviewer_subset/failure_case_000_Actinolite__R050336-1__5330.csv ADDED
The diff for this file is too large to render. See raw diff
 
reviewer_subset/failure_case_000_Actinolite__R050336-1__5330.json ADDED
@@ -0,0 +1,27 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "tag": "failure_case",
3
+ "benchmark_index": 0,
4
+ "case_id": "Actinolite__R050336-1__5330",
5
+ "mineral": "Actinolite",
6
+ "severity": "recoverable",
7
+ "fit_bucket": "fail",
8
+ "rwp": 0.16142818331718445,
9
+ "true_eg": 5,
10
+ "true_space_group": 12,
11
+ "true_space_group_symbol": "C2/m",
12
+ "true_space_group_source": "rruff_fit_dif",
13
+ "paper_model_summary": {
14
+ "index": 0,
15
+ "true_eg": 5,
16
+ "pred_eg": 4,
17
+ "top5_eg": [
18
+ 4,
19
+ 5,
20
+ 75,
21
+ 99,
22
+ 17
23
+ ],
24
+ "correct": false
25
+ },
26
+ "csv_path": "assets/reviewer_examples/failure_case_000_Actinolite__R050336-1__5330.csv"
27
+ }
reviewer_subset/manifest.json ADDED
@@ -0,0 +1,20 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "benchmark_h5_basename": "RRUFF_usable_plus_recoverable_325_with_labels_maxnorm.hdf5",
3
+ "failure_json": "results/mixed2500k_compare_325_failure_modes_655279.json",
4
+ "model_key": null,
5
+ "metadata_csv": "reviewer_case_metadata.csv",
6
+ "examples": {
7
+ "correct_case": {
8
+ "csv_path": "assets/reviewer_examples/correct_case_015_Arsenopyrite__R050071-1__6130.csv",
9
+ "json_path": "assets/reviewer_examples/correct_case_015_Arsenopyrite__R050071-1__6130.json",
10
+ "case_id": "Arsenopyrite__R050071-1__6130",
11
+ "true_eg": 99
12
+ },
13
+ "failure_case": {
14
+ "csv_path": "assets/reviewer_examples/failure_case_000_Actinolite__R050336-1__5330.csv",
15
+ "json_path": "assets/reviewer_examples/failure_case_000_Actinolite__R050336-1__5330.json",
16
+ "case_id": "Actinolite__R050336-1__5330",
17
+ "true_eg": 5
18
+ }
19
+ }
20
+ }
reviewer_subset/reviewer_case_metadata.csv ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ case_id,dif_filename,space_group_symbol,space_group_number,mapped_extinction_group,space_group_source
2
+ Arsenopyrite__R050071-1__6130,Arsenopyrite__R050071-1__Powder__DIF_File__6130.txt,P-1,2,99,rruff_fit_dif
3
+ Actinolite__R050336-1__5330,Actinolite__R050336-1__Powder__DIF_File__5330.txt,C2/m,12,5,rruff_fit_dif
reviewer_subset/rruff325_precomputed_inference.json ADDED
The diff for this file is too large to render. See raw diff