CYHSM commited on
Commit
4e4e79f
·
verified ·
1 Parent(s): 5953d69

Upload registry, index and card

Browse files
Files changed (4) hide show
  1. README.md +19 -6
  2. datasets.h5 +2 -2
  3. index.parquet +2 -2
  4. labels.csv +21 -0
README.md CHANGED
@@ -78,12 +78,25 @@ with no valid gaze sample — mask them rather than dropping them, or the block
78
  and the gaze go out of alignment.
79
 
80
  **Units differ by dataset — read `label_units` from the file attributes.**
81
- `dsL01`–`dsL06` are in degrees of visual angle. `dsL07`–`dsL09` are in native
82
- screen pixels, centred on the screen middle: those datasets do not document
83
- enough geometry to reach degrees (a viewing distance without a physical screen
84
- size is not enough), and a converted-looking number that was actually guessed
85
- would be worse than an honest pixel. Pearson correlation is invariant to the
86
- difference; R² across datasets is not, and was already not comparable.
 
 
 
 
 
 
 
 
 
 
 
 
 
87
 
88
  ## Loading
89
 
 
78
  and the gaze go out of alignment.
79
 
80
  **Units differ by dataset — read `label_units` from the file attributes.**
81
+
82
+ | folder | units |
83
+ |---|---|
84
+ | `dsL01`–`dsL06` | degrees of visual angle |
85
+ | `dsL07_deepmreye_calib` | screen pixels |
86
+ | `dsL08_studyforrest_movie` | degrees of visual angle |
87
+ | `dsL09_fearlearning` | screen pixels |
88
+ | `dsL12_rest` | screen pixels |
89
+
90
+ The source papers were checked for each. Only studyforrest documents enough to
91
+ convert (63 cm viewing distance, 26.5 cm screen at 1280 px, movie subtending
92
+ 23.75°×13.5° → 0.018555 °/px), and that conversion is applied. The rest give a
93
+ viewing distance without a physical screen size, which does not determine
94
+ degrees — so they stay in pixels, centred on the screen middle, rather than
95
+ carrying a factor that was actually guessed.
96
+
97
+ Pearson correlation is invariant to the difference. **Training is not**: if you
98
+ fit one readout over several datasets pooled, standardise the target per dataset
99
+ first, or the largest-scale dataset dominates the loss.
100
 
101
  ## Loading
102
 
datasets.h5 CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:937b08d1cf5ed41ed901d758b4ad23ee8c0686a339ee3698152c50eaeab0d947
3
- size 6554376
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5b1091eb175b3248733670a95b782ddde2f1279cfc59411bd6e65d62172a55b0
3
+ size 6591208
index.parquet CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:4f3774f836720078eb064e77a39f2ab5e4d7c592f1f5f06748fec9b4f3bedd78
3
- size 100530
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b4fe6c6f61cdbece9e1c09e1a047141e6e5af44b7427ae2205a8e2f3c7ebb97b
3
+ size 101008
labels.csv CHANGED
@@ -6021,3 +6021,24 @@ timestamp,dataset,scope,subject,label
6021
  2026-08-03T18:13:21.433584+00:00,dsL07_deepmreye_calib,subject,sub-16,1
6022
  2026-08-03T18:13:21.433584+00:00,dsL07_deepmreye_calib,subject,sub-17,1
6023
  2026-08-03T18:13:21.433832+00:00,dsL10_visseq,dataset,,-99
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6021
  2026-08-03T18:13:21.433584+00:00,dsL07_deepmreye_calib,subject,sub-16,1
6022
  2026-08-03T18:13:21.433584+00:00,dsL07_deepmreye_calib,subject,sub-17,1
6023
  2026-08-03T18:13:21.433832+00:00,dsL10_visseq,dataset,,-99
6024
+ 2026-08-04T08:29:33.726853+00:00,dsL11_backtothefuture,subject,sub-01,1
6025
+ 2026-08-04T08:57:01.651529+00:00,dsL12_rest,subject,sub-01,1
6026
+ 2026-08-04T08:57:01.651529+00:00,dsL12_rest,subject,sub-02,1
6027
+ 2026-08-04T08:57:01.651529+00:00,dsL12_rest,subject,sub-03,1
6028
+ 2026-08-04T08:57:01.651529+00:00,dsL12_rest,subject,sub-04,1
6029
+ 2026-08-04T08:57:01.651529+00:00,dsL12_rest,subject,sub-05,1
6030
+ 2026-08-04T08:57:01.651529+00:00,dsL12_rest,subject,sub-06,1
6031
+ 2026-08-04T08:57:01.651529+00:00,dsL12_rest,subject,sub-07,1
6032
+ 2026-08-04T08:57:01.651529+00:00,dsL12_rest,subject,sub-08,1
6033
+ 2026-08-04T08:57:01.651529+00:00,dsL12_rest,subject,sub-09,1
6034
+ 2026-08-04T08:57:01.651529+00:00,dsL12_rest,subject,sub-11,1
6035
+ 2026-08-04T08:57:01.651529+00:00,dsL12_rest,subject,sub-12,1
6036
+ 2026-08-04T08:57:01.651529+00:00,dsL12_rest,subject,sub-13,1
6037
+ 2026-08-04T08:57:01.651529+00:00,dsL12_rest,subject,sub-14,1
6038
+ 2026-08-04T08:57:01.651529+00:00,dsL12_rest,subject,sub-17,1
6039
+ 2026-08-04T08:57:01.651529+00:00,dsL12_rest,subject,sub-20,1
6040
+ 2026-08-04T08:57:01.651529+00:00,dsL12_rest,subject,sub-21,1
6041
+ 2026-08-04T08:57:01.651529+00:00,dsL12_rest,subject,sub-22,1
6042
+ 2026-08-04T08:57:01.651529+00:00,dsL12_rest,subject,sub-23,1
6043
+ 2026-08-04T08:57:01.651529+00:00,dsL12_rest,subject,sub-24,1
6044
+ 2026-08-04T08:57:01.651529+00:00,dsL12_rest,subject,sub-25,1