Upload training_logs/phase57_partB_h8_bare_multidistance_t9__seed1.json with huggingface_hub

training_logs/phase57_partB_h8_bare_multidistance_t9__seed1.json

@@ -0,0 +1,98 @@
+{
+  "loss": [
+    127962.5546875,
+    51273.875,
+    47019.99609375,
+    45007.99609375,
+    43787.14453125,
+    42891.7578125,
+    42378.12890625,
+    41683.0,
+    41553.15234375,
+    40835.6328125,
+    40923.5625,
+    40681.88671875,
+    40999.2734375,
+    40610.6953125,
+    40525.515625,
+    40043.76953125,
+    40232.875
+  ],
+  "l0": [
+    500.0,
+    495.394287109375,
+    494.80126953125,
+    494.4013671875,
+    492.4873046875,
+    493.312744140625,
+    494.326171875,
+    492.109375,
+    494.06298828125,
+    491.5498046875,
+    492.344970703125,
+    491.951904296875,
+    494.55859375,
+    492.784912109375,
+    493.404052734375,
+    492.083740234375,
+    492.326171875
+  ],
+  "steps_logged": [
+    0,
+    200,
+    400,
+    600,
+    800,
+    1000,
+    1200,
+    1400,
+    1600,
+    1800,
+    2000,
+    2200,
+    2400,
+    2600,
+    2800,
+    3000,
+    3200
+  ],
+  "final_step": 3200,
+  "converged": true,
+  "plateau_last": 0.015874899712026075,
+  "elapsed_s": 5466.467396020889,
+  "shifts": [
+    1,
+    2,
+    4
+  ],
+  "matryoshka_h_size": 3686,
+  "alpha": null,
+  "row": 36,
+  "arch_id": "phase57_partB_h8_bare_multidistance_t9",
+  "arch": "phase57_partB_h8_bare_multidistance_t9",
+  "group": 4,
+  "src_class": "TXCBareMultiDistanceContrastiveAntidead",
+  "src_module": "src.architectures.txc_bare_multidistance_contrastive_antidead",
+  "T": 9,
+  "T_max": null,
+  "t_sample": null,
+  "n_layers": null,
+  "k_win": 500,
+  "k_pos": 56,
+  "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": "phase57_partB_h8_bare_multidistance_t9__seed1"
+}