ericrcwu commited on
Commit
6f08cc1
·
verified ·
1 Parent(s): 3ba80ea

Upload folder using huggingface_hub

Browse files
.gitattributes CHANGED
@@ -38,3 +38,4 @@ bl_a0ccbcff_smokefull/tokenizer.json filter=lfs diff=lfs merge=lfs -text
38
  bl_63cd1bf7_smokefull/tokenizer.json filter=lfs diff=lfs merge=lfs -text
39
  p6_twostage_d0f9c744_smokefull/tokenizer.json filter=lfs diff=lfs merge=lfs -text
40
  distill_r1_smokefull/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 
 
38
  bl_63cd1bf7_smokefull/tokenizer.json filter=lfs diff=lfs merge=lfs -text
39
  p6_twostage_d0f9c744_smokefull/tokenizer.json filter=lfs diff=lfs merge=lfs -text
40
  distill_r1_smokefull/tokenizer.json filter=lfs diff=lfs merge=lfs -text
41
+ distill_r1_distilled_corpus/active_rows.jsonl filter=lfs diff=lfs merge=lfs -text
distill_r1_distilled_corpus/active_manifest.json ADDED
@@ -0,0 +1,186 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "acceptance": {
3
+ "criteria": {
4
+ "canonical_domain": {
5
+ "detail": "0 supported rows off-domain",
6
+ "status": "pass"
7
+ },
8
+ "expert_source_capped": {
9
+ "detail": "ppr10k+fivek fraction=0.49 vs cap 0.50",
10
+ "status": "pass",
11
+ "waived": false
12
+ },
13
+ "generic_input_support": {
14
+ "detail": "all 2761 supported rows have a paired input image",
15
+ "status": "pass"
16
+ },
17
+ "manifests_present": {
18
+ "detail": "model_clients.yaml=present; active_set_version=active_set_pending_freeze",
19
+ "status": "pass"
20
+ },
21
+ "no_dominance": {
22
+ "detail": "max_family_fraction=0.32",
23
+ "family_counts": {
24
+ "controlled_procedural": 39,
25
+ "fivek_derived": 538,
26
+ "fresh_luts": 197,
27
+ "gmic_rawtherapee": 191,
28
+ "ppr10k_derived": 956,
29
+ "scraped_web": 799,
30
+ "smaller_public_packs": 41,
31
+ "unsupported_teacher": 272
32
+ },
33
+ "status": "pass"
34
+ },
35
+ "no_leakage": {
36
+ "detail": "leakage_status=pass",
37
+ "status": "pass"
38
+ },
39
+ "provenance_and_behavior": {
40
+ "detail": "0 supported rows missing measured_behavior",
41
+ "status": "pass"
42
+ },
43
+ "representability_and_recon": {
44
+ "detail": "all 2761 supported rows: representability tier + 64 materialized tokens",
45
+ "status": "pass"
46
+ },
47
+ "scale": {
48
+ "detail": "n=3033 (scale gate relaxed for demo pool)",
49
+ "status": "pass"
50
+ },
51
+ "tags_backed_by_checks": {
52
+ "detail": "0 tag/behavior mismatches",
53
+ "examples": [],
54
+ "status": "pass"
55
+ },
56
+ "unmentioned_behavior_handled": {
57
+ "detail": "reverse tag<->behavior coverage",
58
+ "status": "pass"
59
+ },
60
+ "unsupported_coverage": {
61
+ "detail": "272 unsupported rows, 16 categories, mixed present",
62
+ "status": "pass"
63
+ }
64
+ },
65
+ "overall": "pass"
66
+ },
67
+ "active_set_version": "active_set_v2_portable_unsup",
68
+ "behavior_v2_remeasure": {
69
+ "adr": "0022",
70
+ "behavior_vector_version": "behavior_v2",
71
+ "note": "measured_behavior recomputed at behavior_v2 from luts/canonical_residual",
72
+ "supported_remeasured": 2761
73
+ },
74
+ "behavior_vector_version": "behavior_v2",
75
+ "counts": {
76
+ "supported": 2761,
77
+ "total": 3033,
78
+ "unsupported": 272,
79
+ "unsupported_fraction": 0.0897
80
+ },
81
+ "distillation": {
82
+ "counts": {
83
+ "holdout": 120,
84
+ "kept_gold": 527,
85
+ "replaced": 2114,
86
+ "unsupported": 272
87
+ },
88
+ "harvest_adapter": "models/sft_adapters/p6_twostage_d0f9c744_smokefull",
89
+ "mean_best_of_N": 0.28834919704144385,
90
+ "n": 16,
91
+ "source_rows": "data/active_sft/active_rows.jsonl",
92
+ "tau": 0.18,
93
+ "temperature": 1.0,
94
+ "top_p": 0.9
95
+ },
96
+ "image_pairing": {
97
+ "generic_assigned": 1267,
98
+ "paths_normalized": 1494,
99
+ "pool_free_after": 2387,
100
+ "seed": 20260709
101
+ },
102
+ "materialization": {
103
+ "admission_gate_pass": true,
104
+ "materialized_rows": 2761,
105
+ "recon_overall": {
106
+ "max_deltae": 9.8467,
107
+ "mean_deltae": 1.3003,
108
+ "mean_psnr": 36.6728,
109
+ "p95_deltae": 2.8024
110
+ },
111
+ "tokenizer_version": "vq_v2_srgbres_17to4_cb256_t64__w91cffdd2c82f",
112
+ "unresolved_rows": 0
113
+ },
114
+ "portable_unsupported_migration": {
115
+ "adr": "0023",
116
+ "audit_finding": "F2",
117
+ "files": {
118
+ "data/active_sft/active_rows.jsonl": {
119
+ "changed": 272,
120
+ "rows": 3033,
121
+ "unresolved": 0
122
+ },
123
+ "data/active_sft/unsupported_eval_rows.jsonl": {
124
+ "changed": 235,
125
+ "rows": 235,
126
+ "unresolved": 0
127
+ },
128
+ "data/active_sft/unsupported_rows.jsonl": {
129
+ "changed": 507,
130
+ "rows": 507,
131
+ "unresolved": 0
132
+ }
133
+ },
134
+ "note": "unsupported rows: absolute -> corpus-relative image_path; route/refuse_kind backfilled",
135
+ "prior_active_set_version": "active_set_pending_freeze"
136
+ },
137
+ "regenerated_from": "active_rows.jsonl (post instruction-gen + unsupported append + dedupe)",
138
+ "regeneration_notes": [
139
+ "Regenerated OUTSIDE the pipeline from final rows; not a fresh selection.",
140
+ "per_bucket/effective_size/target_met omitted: selector bucket state is not persisted per-row.",
141
+ "leakage_report + tokenizer reconstruction remain pending; re-run at the Stage 9 freeze."
142
+ ],
143
+ "supported_allowed_override": {
144
+ "count": 4,
145
+ "dropped_tags": {
146
+ "darker": 2,
147
+ "more_contrast": 1
148
+ }
149
+ },
150
+ "supported_empty_gold_tags": 5,
151
+ "supported_instruction_status": {
152
+ "teacher_generated": 2761
153
+ },
154
+ "supported_per_family": {
155
+ "controlled_procedural": 39,
156
+ "fivek_derived": 538,
157
+ "fresh_luts": 197,
158
+ "gmic_rawtherapee": 191,
159
+ "ppr10k_derived": 956,
160
+ "scraped_web": 799,
161
+ "smaller_public_packs": 41
162
+ },
163
+ "unsupported_breakdown": {
164
+ "by_category": {
165
+ "content_generation": 18,
166
+ "content_removal": 18,
167
+ "content_replacement": 18,
168
+ "geometry": 18,
169
+ "inpainting": 13,
170
+ "local_region_edit": 13,
171
+ "mixed_partial_supported_plus_content_generation": 33,
172
+ "mixed_partial_supported_plus_local_edit": 12,
173
+ "mixed_partial_supported_plus_relighting": 17,
174
+ "mixed_partial_supported_plus_semantic_recolor": 13,
175
+ "mixed_partial_supported_plus_texture": 17,
176
+ "reference_style_transfer": 12,
177
+ "relighting": 18,
178
+ "selective_preservation": 18,
179
+ "semantic_object_recolor": 16,
180
+ "texture_detail": 18
181
+ },
182
+ "eval_reserved_rows_file": "data/active_sft/unsupported_eval_rows.jsonl",
183
+ "mixed": 92,
184
+ "train_rows": 272
185
+ }
186
+ }
distill_r1_distilled_corpus/active_rows.jsonl ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a669019b332ae1c08d9c11710a6c8473255bf872669953844f280ccd88f623e0
3
+ size 11875910
distill_r1_distilled_corpus/harvest_cache.jsonl ADDED
The diff for this file is too large to render. See raw diff