Add files using upload-large-folder tool

README.md

@@ -55,10 +55,10 @@ paper_test = bidi.filter(
 
 | Scope | Rows | Held-out `test` rows |
 |---|---:|---:|
-| Complete release | 186,370 | 37,240 |
-| Corrected paper categories | 167,234 | 33,450 |
+| Complete release | 184,978 | 36,808 |
+| Corrected paper categories | 165,972 | 33,012 |
 
-Labels are balanced: 93,185 `feature` and 93,185 `random` rows.
+Labels are balanced: 92,489 `feature` and 92,489 `random` rows.
 
 Rows on `chr2`, `chr3`, `chr16`, and `chr22` are marked `test`; remaining
 included chromosomes are `train`. The Hugging Face split `all` loads the
@@ -66,7 +66,7 @@ whole physical file.
 
 ## Categories
 
-The complete release contains 93,185 source features across:
+The complete release contains 92,489 source features across:
 
 ```text
 repeats, UCNE, vista_enhancer, promoters, UTR5, UTR3,
@@ -84,8 +84,8 @@ view. `repeat_class` is populated on the 19,974 repeat-derived rows
 | `causal` | ROI end-anchored after strand orientation for autoregressive models. |
 | `bidi` | ROI centered for bidirectional/masked models. |
 
-All `bidi` sequences are 2,048 bp. In `causal`, 165,782 rows are 2,048 bp and
-20,590 long-feature rows are truncated to 1,000 bp. Every released random
+All 184,978 `bidi` sequences are 2,048 bp. In `causal`, 165,012 rows are
+2,048 bp and 19,966 long-feature rows are truncated to 1,000 bp. Every released random
 control context contains only uppercase `A/C/G/T`.
 
 ## Evaluation
@@ -121,8 +121,9 @@ uses UCSC `rmsk` field 9 as `source_strand`, field 11 as `repeat_class`, and
 rebuilds repeat-derived controls and sequence orientation. Labels remain
 unchanged.
 
-VISTA accepts only the 2,889 hg38 source rows and excludes 1,750 rows from
-other assemblies without liftover. All ambiguous feature/control groups and
+VISTA requires hg38 and normalized positive expression: 1,367 raw records
+collapse to 1,240 unique elements; 1,522 non-positive hg38 and 1,750 non-hg38
+rows are excluded without liftover. All ambiguous feature/control groups and
 all conflicting-label exact-input groups are removed. Same-label duplicates
 remaining within a config share `context_group_id`; independent checks find
 zero exact inputs crossing train/test and zero conflicting-label groups.

functional-random-gamba-bidi.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9ef5fd5efdf841bbcc001986261dff3a8a099b7ff96165e89f2d942f49d50354
-size 135874144
+oid sha256:7a94f395626580c672f7088631bab5ed4c4c64ff1237a52a5dd8599a5ae470b7
+size 134892470

functional-random-gamba-causal.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:94d210adb8d12adfb2a6a2d1e9acdb0063f50111b9d314f83a82bf3487c3fb6c
-size 128905181
+oid sha256:eb1da69ad900714efce9260eafdd27a0093d221259f4ba4dbd158b198ee34434
+size 128128138