Publish 2026_W42_Cosmos_flimsy-thistle-cat

README.md

@@ -9,12 +9,12 @@ tags:
   - international-brain-lab
 ---
 
-# Ephys Atlas region classifier (2026_W39, Cosmos)
+# Ephys Atlas region classifier (2026_W42, Cosmos)
 
 Predicts the **brain region of each Neuropixels recording channel** from electrophysiological
 features alone -- no histology required. Trained by the
 [International Brain Laboratory](https://www.internationalbrainlab.com/) on the Ephys Atlas
-feature release `2026_W39`.
+feature release `2026_W42`.
 
 > **What you can and cannot do without IBL access.** The *model* runs for anyone. Computing
 > the *input features* from raw Neuropixels data needs `ibllib` / `ibl-neuropixel`, and the
@@ -27,7 +27,7 @@ feature release `2026_W39`.
 import pandas as pd
 from ephysatlas import load_pretrained
 
-model = load_pretrained("int-brain-lab/ea-decoder-channel-xgboost", revision="2026_W39")
+model = load_pretrained("int-brain-lab/ea-decoder-channel-xgboost", revision="2026_W42")
 df = pd.read_parquet("example/features_sample.parquet")   # or your own features
 out = model.predict(df)
 print(out[["predicted_acronym", "prediction_probability", "fold_agreement"]].head())
@@ -57,7 +57,7 @@ two modes disagree on a small fraction of channels, so pick one per analysis.
 - **48 features**, listed in `ephysatlas_model.json` under `inputs.features`. Every
   one must be present; `predict` raises and names anything missing.
 - Indexed by `(pid, channel)`, one row per recording channel.
-- **Must be the denoised aggregated features of vintage `2026_W39`** -- that is, the
+- **Must be the denoised aggregated features of vintage `2026_W42`** -- that is, the
   `raw_ephys_features_denoised.pqt` table produced by the Ephys Atlas aggregation pipeline,
   as loaded by `ephysatlas.data.read_features_from_disk`. Units are baked into that table by
   the pipeline (RMS features in dB, `spike_count` in log2), so feeding raw features, or
@@ -66,8 +66,8 @@ two modes disagree on a small fraction of channels, so pick one per analysis.
 
 ## Performance
 
-Pooled out-of-fold accuracy: **0.5946** over 13 Cosmos regions,
-811 insertions. Splits are by insertion (`pid`), so no channel from a test
+Pooled out-of-fold accuracy: **0.6124** over 13 Cosmos regions,
+755 insertions. Splits are by insertion (`pid`), so no channel from a test
 insertion appears in training. See `confusion_matrix.png`.
 
 Note this figure scores each channel with the one fold that held it out, whereas the default
@@ -85,7 +85,7 @@ typically the marginally better estimator, so treat the number as slightly conse
 
 ## Reproducibility
 
-**Pin the revision.** `revision="2026_W39"` is an immutable tag. Omitting `revision` resolves
+**Pin the revision.** `revision="2026_W42"` is an immutable tag. Omitting `revision` resolves
 to `main`, which tracks whichever model is currently recommended and *will* change when a new
 feature vintage is published — fine for a first look, not for anything you publish or re-run.
 
@@ -101,5 +101,5 @@ which the feature transformer relies on).
 
 ## Citation
 
-Please cite the International Brain Laboratory Ephys Atlas. Model id `2026_W39_Cosmos_comfortable-cerulean-serval`,
-feature vintage `2026_W39`.
+Please cite the International Brain Laboratory Ephys Atlas. Model id `2026_W42_Cosmos_flimsy-thistle-cat`,
+feature vintage `2026_W42`.

checksums.json

@@ -3,53 +3,53 @@
   "files": [
     {
       "path": "ephysatlas_model.json",
-      "hash": "bf98daff2073309d99ffba9b87af3adaa59d619e",
-      "bytes": 2830
+      "hash": "d1ccd4a224778c7ad6729187517a1c8ea9f02888",
+      "bytes": 2804
     },
     {
       "path": "example/expected_predictions.parquet",
-      "hash": "9e492b6eff2aeec7cf7221959bb80e6fc82d9f50",
-      "bytes": 85371
+      "hash": "2c4d503430409f3ff57a167be21fd114d0a36d7f",
+      "bytes": 85461
     },
     {
       "path": "example/features_sample.parquet",
-      "hash": "ae608b7c51acd4fd8e4aaa30adb326f8b796d7cd",
-      "bytes": 258470
+      "hash": "41a3a354c0c5631eaa39f955df92e30888f46cc3",
+      "bytes": 258149
     },
     {
       "path": "folds/FOLD00/model.ubj",
-      "hash": "c8a5b78ae465c044941894586d50952b03bca006",
-      "bytes": 5011871
+      "hash": "c33f3245eb9acb8ea385f15d0c4d37d86214399c",
+      "bytes": 4917864
     },
     {
       "path": "folds/FOLD01/model.ubj",
-      "hash": "664fde50893806d25af17199fde1bc43dcfa212a",
-      "bytes": 4974940
+      "hash": "2a05e06a4ff73ee342b1b55c2d9e192ba3446d32",
+      "bytes": 4999514
     },
     {
       "path": "folds/FOLD02/model.ubj",
-      "hash": "fb21b5290f8cc07aa1283710ef81aa1ce88d05c4",
-      "bytes": 4997608
+      "hash": "97b32e110f66629d25537b63fca9ae2b5b203f40",
+      "bytes": 4932620
     },
     {
       "path": "folds/FOLD03/model.ubj",
-      "hash": "afb097d38ae7a7995bfd562823a72d5d878ff0e2",
-      "bytes": 4982063
+      "hash": "4b862114e97d0cba9866689ac7ef3e554eda2557",
+      "bytes": 4957364
     },
     {
       "path": "folds/FOLD04/model.ubj",
-      "hash": "00c6e2559e7713211cb680ae24582662d080eded",
-      "bytes": 4968939
+      "hash": "67d48be4d794672ba7397e096e351eff20fda748",
+      "bytes": 4938268
     },
     {
       "path": "model.ubj",
-      "hash": "33e4a0c9baf46f77ce8ec0283b4bdb2b24c23582",
-      "bytes": 5074753
+      "hash": "bf6e97b1243ea71bd1db56ea96e84cf4178c0eb1",
+      "bytes": 5043048
     },
     {
       "path": "split.json",
-      "hash": "69640ef12e8c08ecf466f3d37ad7c28cd1af5cd0",
-      "bytes": 231942
+      "hash": "2a9e51b7162a17e8c5d1476f3b1cdd59492ed7f4",
+      "bytes": 216038
     }
   ]
 }

ephysatlas_model.json

@@ -1,23 +1,24 @@
 {
-  "model_id": "2026_W39_Cosmos_comfortable-cerulean-serval",
+  "model_id": "2026_W42_Cosmos_flimsy-thistle-cat",
   "task": "region-classification",
   "model_class": "xgboost.sklearn.XGBClassifier",
-  "vintage": "2026_W39",
+  "vintage": "2026_W42",
   "training": {
     "random_seed": 12345,
-    "training_size": 811,
+    "training_size": 755,
     "testing_size": 0,
-    "hash_training": "05a75d4702030cce0e44c86d9d3fecc91f67a9702e1c1493716c7d841344b38b",
+    "hash_training": "92f04caebcd3fbad1736b69ef0c513b7b9d3b004f4daf2e025eaccfc1c5435e4",
     "hash_testing": null
   },
   "environment": {
-    "python": "3.11.14",
-    "xgboost": "3.2.0",
+    "python": "3.12.12",
+    "xgboost": "3.4.1",
     "scikit_learn": "1.8.0",
-    "numpy": "2.3.5",
-    "pandas": "3.0.5",
-    "torch": "2.14.0",
-    "ephysatlas": "0.9.0"
+    "numpy": "2.5.4",
+    "pandas": "2.3.3",
+    "torch": "2.14.1",
+    "ephysatlas": "0.9.0",
+    "iblatlas": "1.3.0"
   },
   "method": "xgboost",
   "granularity": "channel",
@@ -64,13 +65,13 @@
       "psd_beta_csd",
       "psd_gamma_csd",
       "psd_lfp_csd",
-      "rms_lf_csd_diff1",
-      "psd_delta_csd_diff1",
-      "psd_theta_csd_diff1",
-      "psd_alpha_csd_diff1",
-      "psd_beta_csd_diff1",
-      "psd_gamma_csd_diff1",
-      "psd_lfp_csd_diff1",
+      "rms_lf_bpr",
+      "psd_delta_bpr",
+      "psd_theta_bpr",
+      "psd_alpha_bpr",
+      "psd_beta_bpr",
+      "psd_gamma_bpr",
+      "psd_lfp_bpr",
       "rms_ap",
       "cor_ratio",
       "channel_labels",
@@ -88,7 +89,7 @@
       "tip_val",
       "outside"
     ],
-    "feature_order_sha256": "b1ff4d2e78a689518c73eb480649b09789d3d6d613ae2b4585850b813e24094f"
+    "feature_order_sha256": "7493de89a612e7e348ca2f077898dc599603bf1ba6c73803507ffef66c3f23bb"
   },
   "outputs": {
     "kind": "categorical",
@@ -131,6 +132,6 @@
       "void_fluid"
     ],
     "region_map": "Cosmos",
-    "accuracy": 0.5945880856966708
+    "accuracy": 0.6123792770419426
   }
 }

example/expected_predictions.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:75bf524095dd94ef5a970cbdc16798344f3dbc74a925353fbd3e5e878f03de2b
-size 85371
+oid sha256:ba3024a451d903a2d018f1259923532111feac7aec16eb7b32d90e4245250f56
+size 85461

example/features_sample.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:020655a5f746c30862c2a27be2e9be2b910b149ef3b97fb17064538487aa481d
-size 258470
+oid sha256:e44d1e4c3e5988e631efd4616df0673cc6b359891bcdea4c413572de7ab84759
+size 258149

folds/FOLD00/meta.yaml

@@ -1,75 +0,0 @@
-ACCURACY: 0.6049155773420479
-CLASSES:
-- 0
-- 313
-- 315
-- 512
-- 549
-- 623
-- 698
-- 703
-- 997
-- 1065
-- 1089
-- 1097
-- 2000
-FEATURES:
-- rms_lf
-- psd_lfp
-- psd_delta
-- psd_theta
-- psd_alpha
-- psd_beta
-- psd_gamma
-- psd_residual_lfp
-- psd_residual_delta
-- psd_residual_theta
-- psd_residual_alpha
-- psd_residual_beta
-- psd_residual_gamma
-- aperiodic_offset
-- aperiodic_exponent
-- decay_fit_error
-- decay_fit_r_squared
-- decay_n_peaks
-- rms_lf_csd
-- psd_delta_csd
-- psd_theta_csd
-- psd_alpha_csd
-- psd_beta_csd
-- psd_gamma_csd
-- psd_lfp_csd
-- rms_lf_csd_diff1
-- psd_delta_csd_diff1
-- psd_theta_csd_diff1
-- psd_alpha_csd_diff1
-- psd_beta_csd_diff1
-- psd_gamma_csd_diff1
-- psd_lfp_csd_diff1
-- rms_ap
-- cor_ratio
-- channel_labels
-- alpha_mean
-- alpha_std
-- depolarisation_slope
-- peak_time_secs
-- peak_val
-- polarity
-- recovery_slope
-- recovery_time_secs
-- repolarisation_slope
-- spike_count
-- tip_time_secs
-- tip_val
-- trough_time_secs
-- trough_val
-- outside
-MODEL_CLASS: xgboost.sklearn.XGBClassifier
-RANDOM_SEED: 12345
-REGION_MAP: Cosmos
-TRAINING:
-  hash_testing: 467345c78a5d2ec2b04d9b767973bf8e3aa4e476f3309dd1700ef0e068a89971
-  hash_training: 056b72fe0fa9de10a3e9f515381060424408cd65adc4cfbbc185afe29b0cf982
-  testing_size: 153
-  training_size: 612
-VINTAGE: 2026_W32

folds/FOLD00/model.ubj

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:999fc52438951497bebbe0d7a335f77db9881d94488dd06580a90ecc297959ae
-size 5011871
+oid sha256:34e918c0371091a560f1a2b13e53d9f359d512df75e78c46c4f43c8c0e348dae
+size 4917864

folds/FOLD01/meta.yaml

@@ -1,75 +0,0 @@
-ACCURACY: 0.6026177832244008
-CLASSES:
-- 0
-- 313
-- 315
-- 512
-- 549
-- 623
-- 698
-- 703
-- 997
-- 1065
-- 1089
-- 1097
-- 2000
-FEATURES:
-- rms_lf
-- psd_lfp
-- psd_delta
-- psd_theta
-- psd_alpha
-- psd_beta
-- psd_gamma
-- psd_residual_lfp
-- psd_residual_delta
-- psd_residual_theta
-- psd_residual_alpha
-- psd_residual_beta
-- psd_residual_gamma
-- aperiodic_offset
-- aperiodic_exponent
-- decay_fit_error
-- decay_fit_r_squared
-- decay_n_peaks
-- rms_lf_csd
-- psd_delta_csd
-- psd_theta_csd
-- psd_alpha_csd
-- psd_beta_csd
-- psd_gamma_csd
-- psd_lfp_csd
-- rms_lf_csd_diff1
-- psd_delta_csd_diff1
-- psd_theta_csd_diff1
-- psd_alpha_csd_diff1
-- psd_beta_csd_diff1
-- psd_gamma_csd_diff1
-- psd_lfp_csd_diff1
-- rms_ap
-- cor_ratio
-- channel_labels
-- alpha_mean
-- alpha_std
-- depolarisation_slope
-- peak_time_secs
-- peak_val
-- polarity
-- recovery_slope
-- recovery_time_secs
-- repolarisation_slope
-- spike_count
-- tip_time_secs
-- tip_val
-- trough_time_secs
-- trough_val
-- outside
-MODEL_CLASS: xgboost.sklearn.XGBClassifier
-RANDOM_SEED: 12345
-REGION_MAP: Cosmos
-TRAINING:
-  hash_testing: 50b5cc7b97983ebcf6f85dff467a16ddd4bc17f3593932154a1baf706cc98624
-  hash_training: 033a58ccd28370fa442503948aaa7035e5ed3285d4127fce396802f9b048654f
-  testing_size: 153
-  training_size: 612
-VINTAGE: 2026_W32

folds/FOLD01/model.ubj

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fcc90566edc818b6510ca160d6a69c4f418ccdacf1a68225cf5fc3418070104e
-size 4974940
+oid sha256:e5abc6211fadc70d741c3bcd42c79fe704c47cea8872c5fcd21e87fc450e9faa
+size 4999514

folds/FOLD02/meta.yaml

@@ -1,75 +0,0 @@
-ACCURACY: 0.6063282952069716
-CLASSES:
-- 0
-- 313
-- 315
-- 512
-- 549
-- 623
-- 698
-- 703
-- 997
-- 1065
-- 1089
-- 1097
-- 2000
-FEATURES:
-- rms_lf
-- psd_lfp
-- psd_delta
-- psd_theta
-- psd_alpha
-- psd_beta
-- psd_gamma
-- psd_residual_lfp
-- psd_residual_delta
-- psd_residual_theta
-- psd_residual_alpha
-- psd_residual_beta
-- psd_residual_gamma
-- aperiodic_offset
-- aperiodic_exponent
-- decay_fit_error
-- decay_fit_r_squared
-- decay_n_peaks
-- rms_lf_csd
-- psd_delta_csd
-- psd_theta_csd
-- psd_alpha_csd
-- psd_beta_csd
-- psd_gamma_csd
-- psd_lfp_csd
-- rms_lf_csd_diff1
-- psd_delta_csd_diff1
-- psd_theta_csd_diff1
-- psd_alpha_csd_diff1
-- psd_beta_csd_diff1
-- psd_gamma_csd_diff1
-- psd_lfp_csd_diff1
-- rms_ap
-- cor_ratio
-- channel_labels
-- alpha_mean
-- alpha_std
-- depolarisation_slope
-- peak_time_secs
-- peak_val
-- polarity
-- recovery_slope
-- recovery_time_secs
-- repolarisation_slope
-- spike_count
-- tip_time_secs
-- tip_val
-- trough_time_secs
-- trough_val
-- outside
-MODEL_CLASS: xgboost.sklearn.XGBClassifier
-RANDOM_SEED: 12345
-REGION_MAP: Cosmos
-TRAINING:
-  hash_testing: b4f523886d2def517530873ea3f754bd1d6331ebafffed4566bd0d4e20598405
-  hash_training: dafb953e4cc24630af37bfe50bdc3465dbd360b346e42c57b82dd70118a768e8
-  testing_size: 153
-  training_size: 612
-VINTAGE: 2026_W32

folds/FOLD02/model.ubj

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:19f5a7ef02346c67705576ed2c47725831971d5c3cba1b67ba959088da243635
-size 4997608
+oid sha256:fe757606d33199c59593c5f4cc5b2539558786f464d78cefda14367b96d4fa88
+size 4932620

folds/FOLD03/meta.yaml

@@ -1,75 +0,0 @@
-ACCURACY: 0.5737847222222222
-CLASSES:
-- 0
-- 313
-- 315
-- 512
-- 549
-- 623
-- 698
-- 703
-- 997
-- 1065
-- 1089
-- 1097
-- 2000
-FEATURES:
-- rms_lf
-- psd_lfp
-- psd_delta
-- psd_theta
-- psd_alpha
-- psd_beta
-- psd_gamma
-- psd_residual_lfp
-- psd_residual_delta
-- psd_residual_theta
-- psd_residual_alpha
-- psd_residual_beta
-- psd_residual_gamma
-- aperiodic_offset
-- aperiodic_exponent
-- decay_fit_error
-- decay_fit_r_squared
-- decay_n_peaks
-- rms_lf_csd
-- psd_delta_csd
-- psd_theta_csd
-- psd_alpha_csd
-- psd_beta_csd
-- psd_gamma_csd
-- psd_lfp_csd
-- rms_lf_csd_diff1
-- psd_delta_csd_diff1
-- psd_theta_csd_diff1
-- psd_alpha_csd_diff1
-- psd_beta_csd_diff1
-- psd_gamma_csd_diff1
-- psd_lfp_csd_diff1
-- rms_ap
-- cor_ratio
-- channel_labels
-- alpha_mean
-- alpha_std
-- depolarisation_slope
-- peak_time_secs
-- peak_val
-- polarity
-- recovery_slope
-- recovery_time_secs
-- repolarisation_slope
-- spike_count
-- tip_time_secs
-- tip_val
-- trough_time_secs
-- trough_val
-- outside
-MODEL_CLASS: xgboost.sklearn.XGBClassifier
-RANDOM_SEED: 12345
-REGION_MAP: Cosmos
-TRAINING:
-  hash_testing: 96473b4de53f65aea04b072a378bce2e93e0c8698c6c04564d6c458acb035882
-  hash_training: cfe6ea9328439235a9dd44481aa474099ceed22e661fc82b8fc6a700dcec499d
-  testing_size: 153
-  training_size: 612
-VINTAGE: 2026_W32

folds/FOLD03/model.ubj

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:53b55f8415feb1dca256a19a51670b08d8264ff448ad02cf903a73c236031183
-size 4982063
+oid sha256:3b6db366be2cea1e9bcae0d4cef01febd85eef93cb88779398b2ec9eaf54c265
+size 4957364

folds/FOLD04/meta.yaml

@@ -1,75 +0,0 @@
-ACCURACY: 0.5932904411764706
-CLASSES:
-- 0
-- 313
-- 315
-- 512
-- 549
-- 623
-- 698
-- 703
-- 997
-- 1065
-- 1089
-- 1097
-- 2000
-FEATURES:
-- rms_lf
-- psd_lfp
-- psd_delta
-- psd_theta
-- psd_alpha
-- psd_beta
-- psd_gamma
-- psd_residual_lfp
-- psd_residual_delta
-- psd_residual_theta
-- psd_residual_alpha
-- psd_residual_beta
-- psd_residual_gamma
-- aperiodic_offset
-- aperiodic_exponent
-- decay_fit_error
-- decay_fit_r_squared
-- decay_n_peaks
-- rms_lf_csd
-- psd_delta_csd
-- psd_theta_csd
-- psd_alpha_csd
-- psd_beta_csd
-- psd_gamma_csd
-- psd_lfp_csd
-- rms_lf_csd_diff1
-- psd_delta_csd_diff1
-- psd_theta_csd_diff1
-- psd_alpha_csd_diff1
-- psd_beta_csd_diff1
-- psd_gamma_csd_diff1
-- psd_lfp_csd_diff1
-- rms_ap
-- cor_ratio
-- channel_labels
-- alpha_mean
-- alpha_std
-- depolarisation_slope
-- peak_time_secs
-- peak_val
-- polarity
-- recovery_slope
-- recovery_time_secs
-- repolarisation_slope
-- spike_count
-- tip_time_secs
-- tip_val
-- trough_time_secs
-- trough_val
-- outside
-MODEL_CLASS: xgboost.sklearn.XGBClassifier
-RANDOM_SEED: 12345
-REGION_MAP: Cosmos
-TRAINING:
-  hash_testing: c19df30438105f7346e9caa36f8687d78f05f69cf93fab88296a02b3d2c6b321
-  hash_training: eca91fb309540a7d88a0f33dde4882fd3f6bef7d5c0b77f94e1acaf45d5a6c72
-  testing_size: 153
-  training_size: 612
-VINTAGE: 2026_W32

folds/FOLD04/model.ubj

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:04fad23202941f688e8920e56d691c3f41ce43ec9791a061be54d5821ced4130
-size 4968939
+oid sha256:7283c703797d3dceec0277d6616e044caf9db88dfdd2708318c4dd02f3b18f84
+size 4938268

meta.yaml

@@ -1,75 +0,0 @@
-ACCURACY: 0.5961873638344226
-CLASSES:
-- 0
-- 313
-- 315
-- 512
-- 549
-- 623
-- 698
-- 703
-- 997
-- 1065
-- 1089
-- 1097
-- 2000
-FEATURES:
-- rms_lf
-- psd_lfp
-- psd_delta
-- psd_theta
-- psd_alpha
-- psd_beta
-- psd_gamma
-- psd_residual_lfp
-- psd_residual_delta
-- psd_residual_theta
-- psd_residual_alpha
-- psd_residual_beta
-- psd_residual_gamma
-- aperiodic_offset
-- aperiodic_exponent
-- decay_fit_error
-- decay_fit_r_squared
-- decay_n_peaks
-- rms_lf_csd
-- psd_delta_csd
-- psd_theta_csd
-- psd_alpha_csd
-- psd_beta_csd
-- psd_gamma_csd
-- psd_lfp_csd
-- rms_lf_csd_diff1
-- psd_delta_csd_diff1
-- psd_theta_csd_diff1
-- psd_alpha_csd_diff1
-- psd_beta_csd_diff1
-- psd_gamma_csd_diff1
-- psd_lfp_csd_diff1
-- rms_ap
-- cor_ratio
-- channel_labels
-- alpha_mean
-- alpha_std
-- depolarisation_slope
-- peak_time_secs
-- peak_val
-- polarity
-- recovery_slope
-- recovery_time_secs
-- repolarisation_slope
-- spike_count
-- tip_time_secs
-- tip_val
-- trough_time_secs
-- trough_val
-- outside
-MODEL_CLASS: xgboost.sklearn.XGBClassifier
-RANDOM_SEED: 12345
-REGION_MAP: Cosmos
-TRAINING:
-  hash_testing: null
-  hash_training: 8ca9ba837f320718e6bb9c6daaca680f110f503748404124ad9d221803083d92
-  testing_size: 0
-  training_size: 765
-VINTAGE: 2026_W32

model.ubj

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:aa30724d94a82799b65fd13801f75d424704209ab34c2329d091647a21c774f0
-size 5074753
+oid sha256:043e0a984781a19fa10b4b18bfd3aca3f498a8a1a05a71055ff4061459947929
+size 5043048