Upload training_logs/phase5b_subseq_track2__seed1.json with huggingface_hub

training_logs/phase5b_subseq_track2__seed1.json

@@ -0,0 +1,96 @@
+{
+  "loss": [
+    17265.51953125,
+    6290.14990234375,
+    5699.16748046875,
+    5454.59326171875,
+    5321.16943359375,
+    5182.515625,
+    5067.52978515625,
+    5077.9296875,
+    5035.59326171875,
+    4954.22119140625,
+    4938.888671875,
+    4913.4189453125,
+    4926.28271484375,
+    4898.189453125,
+    4879.82421875,
+    4825.37255859375,
+    4854.55419921875,
+    4866.50390625
+  ],
+  "l0": [
+    500.0,
+    497.40380859375,
+    497.3095703125,
+    496.5732421875,
+    496.7822265625,
+    497.09375,
+    496.07421875,
+    496.8115234375,
+    496.672119140625,
+    496.0234375,
+    494.92138671875,
+    495.0595703125,
+    495.5595703125,
+    495.54443359375,
+    495.055908203125,
+    494.145751953125,
+    493.975341796875,
+    494.615478515625
+  ],
+  "steps_logged": [
+    0,
+    200,
+    400,
+    600,
+    800,
+    1000,
+    1200,
+    1400,
+    1600,
+    1800,
+    2000,
+    2200,
+    2400,
+    2600,
+    2800,
+    3000,
+    3200,
+    3400
+  ],
+  "final_step": 3400,
+  "converged": true,
+  "plateau_last": 0.01792446679831454,
+  "elapsed_s": 2074.7629668712616,
+  "T_max": 10,
+  "t_sample": 5,
+  "row": 12,
+  "arch_id": "phase5b_subseq_track2",
+  "arch": "phase5b_subseq_track2",
+  "group": 2,
+  "src_class": "SubseqTXCBareAntidead",
+  "src_module": "src.architectures.phase5b_subseq_sampling_txcdr",
+  "T": null,
+  "n_layers": null,
+  "k_win": 500,
+  "k_pos": 100,
+  "shifts": null,
+  "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_track2__seed1"
+}