Ground truth to top-1000; card regenerated

README.md

@@ -34,7 +34,8 @@ respectively), see [Encoding](#encoding).
 | `query_lens.npy` | int32 | `[50]` | true vectors per query, before padding |
 | `queries_ids.npy` | `<U2` | `[50]` | original query ids |
 | `qrels.test.tsv` | text | 66,336 rows | TREC qrels, `qid \t 0 \t docid \t relevance`, no header |
-| `gt_top100.tsv` | text | 5,000 rows | exact MaxSim top-100, see below |
+| `gt_top1000.tsv` | text | 50,000 rows | exact MaxSim top-1000, see below |
+| `gt_top100.tsv` | text | 5,000 rows | first 100 ranks of `gt_top1000.tsv`, same format |
 
 All positional indices (the `gt_top*.tsv` files, and the row order of every `.npy` file) refer to the
 order of `doc_ids.npy` and `queries_ids.npy`. Reordering either file invalidates the ground truth.
@@ -74,9 +75,9 @@ order of `doc_ids.npy` and `queries_ids.npy`. Reordering either file invalidates
 | query padding | rows at or beyond `query_lens[i]` in `queries.npy[i]` are exactly zero |
 | token_ids | tokenizer id of each kept document token (after the skiplist above), aligned 1:1 with `documents.npy` |
 
-## Ground truth: `gt_top100.tsv`
+## Ground truth: `gt_top1000.tsv` and `gt_top100.tsv`
 
-Exact brute-force MaxSim top-100 per query over the full corpus, from the vectors in this repo.
+Exact brute-force MaxSim top-1000 per query over the full corpus, from the vectors in this repo. `gt_top100.tsv` holds the first 100 ranks per query of the same lists (the original layout of these exports).
 
 No header; tab-separated `qidx  docidx  rank  score`:
 
@@ -89,12 +90,12 @@ No header; tab-separated `qidx  docidx  rank  score`:
 
 ## Retrieval quality
 
-Sanity check of the vectors, not a leaderboard number: `gt_top100.tsv` (exact MaxSim over the full
+Sanity check of the vectors, not a leaderboard number: `gt_top1000.tsv` (exact MaxSim over the full
 corpus) scored against `qrels.test.tsv` with ir_measures.
 
-| nDCG@10 | MRR@10 | Success@5 | Recall@100 | Recall@1000 | MAP@100 |
+| nDCG@10 | MRR@10 | Success@5 | Recall@100 | Recall@1000 | MAP@1000 |
 |---|---|---|---|---|---|
-| 0.8390 | 0.9750 | 1.0000 | 0.1604 | n/a (gt is top-100) | 0.1311 |
+| 0.8390 | 0.9750 | 1.0000 | 0.1604 | 0.5419 | 0.3246 |
 
 ## Loading
 
@@ -142,8 +143,12 @@ Checks run by the exporter on the files exactly as written here:
 - ✅ query vectors unit-norm — norm range [1.000000, 1.000000]
 - ✅ all vectors finite
 - ✅ gt_top100.tsv has k rows per query — 5000 rows, k=100
-- ✅ gt rows grouped by qidx with ranks 1..k and descending scores
-- ✅ gt indices in range
+- ✅ gt_top100.tsv rows grouped by qidx with ranks 1..k and descending scores
+- ✅ gt_top100.tsv indices in range
+- ✅ gt_top1000.tsv has k rows per query — 50000 rows, k=1000
+- ✅ gt_top1000.tsv rows grouped by qidx with ranks 1..k and descending scores
+- ✅ gt_top1000.tsv indices in range
+- ✅ gt_top100.tsv is the first 100 ranks of gt_top1000.tsv
 
 ## Provenance
 
@@ -151,4 +156,4 @@ Checks run by the exporter on the files exactly as written here:
 |---|---|
 | exported | 2026-09-25 |
 | hardware | Tesla V100S-PCIE-32GB |
-| revised | 2026-09-28: card regenerated; every other file unchanged |
+| revised | 2026-09-29: ground truth extended to top-1000 (`gt_top1000.tsv`, exact MaxSim over this repo's vectors on Tesla V100S-PCIE-32GB); `gt_top100.tsv` rewritten as its first 100 ranks: 32 rows differ from the previous file, 26 with a different document at that rank, scores moving by at most 0.000002 |

gt_top100.tsv

@@ -5,7 +5,7 @@
 0	140830	5	9.784604
 0	92160	6	9.782530
 0	146094	7	9.775735
-0	171112	8	9.773419
+0	171112	8	9.773420
 0	120397	9	9.771715
 0	145424	10	9.770369
 0	103419	11	9.765282
@@ -21,7 +21,7 @@
 0	94609	21	9.748045
 0	115116	22	9.746242
 0	113088	23	9.743753
-0	170998	24	9.743333
+0	170998	24	9.743332
 0	73424	25	9.741647
 0	80972	26	9.739978
 0	119070	27	9.737442
@@ -576,8 +576,8 @@
 5	120363	76	15.817274
 5	123404	77	15.817032
 5	103841	78	15.813909
-5	113833	79	15.812727
-5	58695	80	15.812727
+5	58695	79	15.812727
+5	113833	80	15.812727
 5	75620	81	15.812584
 5	95025	82	15.812558
 5	110161	83	15.812132
@@ -795,8 +795,8 @@
 7	127999	95	20.497002
 7	107508	96	20.496162
 7	131369	97	20.496044
-7	157523	98	20.495369
-7	68974	99	20.495369
+7	68974	98	20.495369
+7	157523	99	20.495369
 7	70336	100	20.494902
 8	121750	1	8.759260
 8	146451	2	8.746679
@@ -1294,7 +1294,7 @@
 12	92218	94	9.806202
 12	70867	95	9.806197
 12	92538	96	9.806149
-12	171323	97	9.806026
+12	171323	97	9.806028
 12	153678	98	9.805875
 12	96130	99	9.805294
 12	155794	100	9.805129
@@ -2329,8 +2329,8 @@
 23	123137	29	14.784794
 23	92366	30	14.784226
 23	72875	31	14.784121
-23	107442	32	14.783558
-23	51033	33	14.783558
+23	51033	32	14.783558
+23	107442	33	14.783558
 23	162919	34	14.782087
 23	146658	35	14.779297
 23	62795	36	14.779279
@@ -2377,7 +2377,7 @@
 23	141863	77	14.755318
 23	106067	78	14.752854
 23	96882	79	14.752674
-23	170954	80	14.752598
+23	170954	80	14.752599
 23	100370	81	14.751928
 23	93086	82	14.751436
 23	168555	83	14.751429
@@ -2398,8 +2398,8 @@
 23	87904	98	14.746175
 23	162225	99	14.746008
 23	153230	100	14.745858
-24	146663	1	16.656931
-24	96180	2	16.656931
+24	96180	1	16.656931
+24	146663	2	16.656931
 24	69001	3	16.636349
 24	162116	4	16.632050
 24	39864	5	16.623644
@@ -2562,12 +2562,12 @@
 25	82184	62	13.034206
 25	129509	63	13.033991
 25	102778	64	13.033990
-25	171179	65	13.033692
+25	171179	65	13.033691
 25	136259	66	13.033386
 25	102293	67	13.033348
 25	41090	68	13.032923
 25	154587	69	13.032598
-25	170996	70	13.032542
+25	170996	70	13.032541
 25	102902	71	13.032492
 25	161448	72	13.032341
 25	45815	73	13.032230
@@ -2616,8 +2616,8 @@
 26	99360	16	16.777071
 26	126100	17	16.775227
 26	105609	18	16.774487
-26	151491	19	16.770290
-26	101897	20	16.770290
+26	101897	19	16.770290
+26	151491	20	16.770290
 26	23653	21	16.769642
 26	70617	22	16.766392
 26	37208	23	16.765717
@@ -2706,8 +2706,8 @@
 27	75747	6	21.803362
 27	62429	7	21.800676
 27	105545	8	21.795357
-27	155357	9	21.795111
-27	91070	10	21.795111
+27	91070	9	21.795111
+27	155357	10	21.795111
 27	48525	11	21.794794
 27	90228	12	21.794308
 27	105356	13	21.791227
@@ -2735,8 +2735,8 @@
 27	71645	35	21.768303
 27	141063	36	21.767433
 27	100021	37	21.766346
-27	133813	38	21.765835
-27	25585	39	21.765835
+27	25585	38	21.765835
+27	133813	39	21.765835
 27	68492	40	21.765636
 27	86339	41	21.765400
 27	62055	42	21.764383
@@ -2833,8 +2833,8 @@
 28	170899	33	30.651924
 28	73230	34	30.648951
 28	92937	35	30.648048
-28	138058	36	30.647924
-28	97798	37	30.647924
+28	97798	36	30.647924
+28	138058	37	30.647924
 28	166876	38	30.645617
 28	31876	39	30.644648
 28	100033	40	30.644129
@@ -2965,8 +2965,8 @@
 29	58478	65	12.934286
 29	60048	66	12.933493
 29	167508	67	12.932112
-29	128572	68	12.932018
-29	91251	69	12.932018
+29	91251	68	12.932018
+29	128572	69	12.932018
 29	11810	70	12.931959
 29	29042	71	12.931899
 29	142932	72	12.931566
@@ -3526,8 +3526,8 @@
 35	148545	26	15.988437
 35	158679	27	15.987347
 35	116830	28	15.983826
-35	159607	29	15.982826
-35	87910	30	15.982826
+35	87910	29	15.982826
+35	159607	30	15.982826
 35	95156	31	15.980643
 35	67174	32	15.978757
 35	60719	33	15.978340
@@ -4410,8 +4410,8 @@
 44	114393	10	12.995048
 44	49352	11	12.994320
 44	89577	12	12.993188
-44	134131	13	12.991380
-44	90067	14	12.991380
+44	90067	13	12.991380
+44	134131	14	12.991380
 44	103033	15	12.989880
 44	72540	16	12.987570
 44	70836	17	12.987398
@@ -4555,8 +4555,8 @@
 45	115596	55	15.569304
 45	99455	56	15.569086
 45	88958	57	15.565956
-45	159476	58	15.565398
-45	120561	59	15.565398
+45	120561	58	15.565398
+45	159476	59	15.565398
 45	68562	60	15.563435
 45	117113	61	15.562384
 45	109610	62	15.561424
@@ -4579,8 +4579,8 @@
 45	95806	79	15.553474
 45	115717	80	15.553319
 45	94177	81	15.552975
-45	148188	82	15.552907
-45	112200	83	15.552907
+45	112200	82	15.552907
+45	148188	83	15.552907
 45	118455	84	15.552725
 45	89411	85	15.552633
 45	109183	86	15.551933