Upload training_logs/phase5b_subseq_h8__seed1.json with huggingface_hub

training_logs/phase5b_subseq_h8__seed1.json

@@ -0,0 +1,95 @@
+{
+  "loss": [
+    121757.96875,
+    50702.26953125,
+    47850.796875,
+    47004.0859375,
+    46143.71484375,
+    45698.10546875,
+    45365.57421875,
+    45648.64453125,
+    45268.5234375,
+    45107.01953125,
+    44737.34375,
+    45042.4375,
+    45003.08984375,
+    44817.30078125,
+    44881.87109375,
+    44662.80859375
+  ],
+  "l0": [
+    500.0,
+    495.823486328125,
+    495.682861328125,
+    495.972900390625,
+    495.685302734375,
+    495.725341796875,
+    496.040283203125,
+    497.181884765625,
+    496.80126953125,
+    496.099853515625,
+    494.629638671875,
+    495.306396484375,
+    494.967041015625,
+    495.694580078125,
+    495.9208984375,
+    494.676025390625
+  ],
+  "steps_logged": [
+    0,
+    200,
+    400,
+    600,
+    800,
+    1000,
+    1200,
+    1400,
+    1600,
+    1800,
+    2000,
+    2200,
+    2400,
+    2600,
+    2800,
+    3000
+  ],
+  "final_step": 3000,
+  "converged": true,
+  "plateau_last": 0.007604562277863013,
+  "elapsed_s": 5567.742310523987,
+  "T_max": 10,
+  "t_sample": 5,
+  "shifts": [
+    1,
+    2,
+    5
+  ],
+  "matryoshka_h_size": 3686,
+  "row": 13,
+  "arch_id": "phase5b_subseq_h8",
+  "arch": "phase5b_subseq_h8",
+  "group": 2,
+  "src_class": "SubseqH8",
+  "src_module": "src.architectures.phase5b_subseq_sampling_txcdr",
+  "T": null,
+  "n_layers": null,
+  "k_win": 500,
+  "k_pos": 100,
+  "alpha": null,
+  "gamma": null,
+  "n_scales": null,
+  "seed": 1,
+  "d_in": 2304,
+  "d_sae": 18432,
+  "subject_model": "google/gemma-2-2b",
+  "anchor_layer": 12,
+  "mlc_layers": [
+    10,
+    11,
+    12,
+    13,
+    14
+  ],
+  "phase": "phase7_unification",
+  "run_id": "phase5b_subseq_h8__seed1"
+}