summaryrefslogtreecommitdiff
path: root/results/online_shallow_3seed
diff options
context:
space:
mode:
authorYurenHao0426 <Blackhao0426@gmail.com>2026-06-14 04:06:32 -0500
committerYurenHao0426 <Blackhao0426@gmail.com>2026-06-14 04:06:32 -0500
commitaa73718eb6427d7da3b9cb416275802d90c4b2ed (patch)
treeb68b0a664fb650744ef934a1c22abd740a7b62a6 /results/online_shallow_3seed
parent827c658fa9a750f3c6ebdb87703762f10f69f6ff (diff)
Add new experiment scripts, figures, and paper assets; untrack pyc/build artifactsHEADmaster
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Diffstat (limited to 'results/online_shallow_3seed')
-rw-r--r--results/online_shallow_3seed/scan_s123.json62
-rw-r--r--results/online_shallow_3seed/scan_s42.json62
-rw-r--r--results/online_shallow_3seed/scan_s456.json62
3 files changed, 186 insertions, 0 deletions
diff --git a/results/online_shallow_3seed/scan_s123.json b/results/online_shallow_3seed/scan_s123.json
new file mode 100644
index 0000000..ad5d91d
--- /dev/null
+++ b/results/online_shallow_3seed/scan_s123.json
@@ -0,0 +1,62 @@
+[
+ {
+ "method": "dfa",
+ "L": 4,
+ "d_hidden": 256,
+ "seed": 123,
+ "warmup_ratio": 0.2,
+ "term_grad_weight": 1.0,
+ "test_acc": 0.3106,
+ "mean_gamma": 0.11738517042249441,
+ "mean_rho": 0.005885639227926731,
+ "mean_nudge": -4.444736987352371e-07,
+ "per_layer_gamma": [
+ 0.4517223834991455,
+ 0.012923447415232658,
+ 0.004009386524558067,
+ 0.0008854642510414124
+ ],
+ "per_layer_rho": [
+ 0.023542556911706924,
+ 0.0,
+ 0.0,
+ 0.0
+ ],
+ "per_layer_nudge": [
+ -1.7816200852394104e-06,
+ 3.725290298461914e-09,
+ 0.0,
+ 0.0
+ ]
+ },
+ {
+ "method": "cb_eT",
+ "L": 4,
+ "d_hidden": 256,
+ "seed": 123,
+ "warmup_ratio": 0.2,
+ "term_grad_weight": 1.0,
+ "test_acc": 0.2753,
+ "mean_gamma": 0.17617796920239925,
+ "mean_rho": 0.0019302130676805973,
+ "mean_nudge": -7.422640919685364e-07,
+ "per_layer_gamma": [
+ 0.5222728252410889,
+ 0.043591007590293884,
+ 0.0659613385796547,
+ 0.07288670539855957
+ ],
+ "per_layer_rho": [
+ 0.022736016660928726,
+ 0.0,
+ -0.015015164390206337,
+ 0.0
+ ],
+ "per_layer_nudge": [
+ -2.9690563678741455e-06,
+ 0.0,
+ 0.0,
+ 0.0
+ ]
+ }
+] \ No newline at end of file
diff --git a/results/online_shallow_3seed/scan_s42.json b/results/online_shallow_3seed/scan_s42.json
new file mode 100644
index 0000000..624919b
--- /dev/null
+++ b/results/online_shallow_3seed/scan_s42.json
@@ -0,0 +1,62 @@
+[
+ {
+ "method": "dfa",
+ "L": 4,
+ "d_hidden": 256,
+ "seed": 42,
+ "warmup_ratio": 0.2,
+ "term_grad_weight": 1.0,
+ "test_acc": 0.3116,
+ "mean_gamma": 0.10076353600015864,
+ "mean_rho": -0.00476757250726223,
+ "mean_nudge": -3.260793164372444e-07,
+ "per_layer_gamma": [
+ 0.42987197637557983,
+ 0.002452872460708022,
+ -0.014306485652923584,
+ -0.014964219182729721
+ ],
+ "per_layer_rho": [
+ -0.01907029002904892,
+ 0.0,
+ 0.0,
+ 0.0
+ ],
+ "per_layer_nudge": [
+ -1.3024546205997467e-06,
+ -1.862645149230957e-09,
+ 0.0,
+ 0.0
+ ]
+ },
+ {
+ "method": "cb_eT",
+ "L": 4,
+ "d_hidden": 256,
+ "seed": 42,
+ "warmup_ratio": 0.2,
+ "term_grad_weight": 1.0,
+ "test_acc": 0.283,
+ "mean_gamma": 0.17921950668096542,
+ "mean_rho": 0.008699589408934116,
+ "mean_nudge": -8.21426510810852e-07,
+ "per_layer_gamma": [
+ 0.5069771409034729,
+ 0.05348020792007446,
+ 0.07490736246109009,
+ 0.08151331543922424
+ ],
+ "per_layer_rho": [
+ 0.034798357635736465,
+ 0.0,
+ 0.0,
+ 0.0
+ ],
+ "per_layer_nudge": [
+ -3.285706043243408e-06,
+ 0.0,
+ 0.0,
+ 0.0
+ ]
+ }
+] \ No newline at end of file
diff --git a/results/online_shallow_3seed/scan_s456.json b/results/online_shallow_3seed/scan_s456.json
new file mode 100644
index 0000000..09327b1
--- /dev/null
+++ b/results/online_shallow_3seed/scan_s456.json
@@ -0,0 +1,62 @@
+[
+ {
+ "method": "dfa",
+ "L": 4,
+ "d_hidden": 256,
+ "seed": 456,
+ "warmup_ratio": 0.2,
+ "term_grad_weight": 1.0,
+ "test_acc": 0.297,
+ "mean_gamma": 0.09892195643624291,
+ "mean_rho": -0.004888533148914576,
+ "mean_nudge": -3.1944364309310913e-07,
+ "per_layer_gamma": [
+ 0.3969520926475525,
+ -0.0020016406197100878,
+ -0.0031630932353436947,
+ 0.003900466952472925
+ ],
+ "per_layer_rho": [
+ -0.019554132595658302,
+ 0.0,
+ 0.0,
+ 0.0
+ ],
+ "per_layer_nudge": [
+ -1.2777745723724365e-06,
+ 0.0,
+ 0.0,
+ 0.0
+ ]
+ },
+ {
+ "method": "cb_eT",
+ "L": 4,
+ "d_hidden": 256,
+ "seed": 456,
+ "warmup_ratio": 0.2,
+ "term_grad_weight": 1.0,
+ "test_acc": 0.2893,
+ "mean_gamma": 0.23417379707098007,
+ "mean_rho": -0.002338360995054245,
+ "mean_nudge": -5.409820005297661e-07,
+ "per_layer_gamma": [
+ 0.41382449865341187,
+ 0.12546014785766602,
+ 0.1961553394794464,
+ 0.201255202293396
+ ],
+ "per_layer_rho": [
+ -0.00935344398021698,
+ 0.0,
+ 0.0,
+ 0.0
+ ],
+ "per_layer_nudge": [
+ -2.1639280021190643e-06,
+ 0.0,
+ 0.0,
+ 0.0
+ ]
+ }
+] \ No newline at end of file