Upload training_logs/phase57_partB_h8_bare_multidistance_t9__seed2.json with huggingface_hub

training_logs/phase57_partB_h8_bare_multidistance_t9__seed2.json

@@ -0,0 +1,98 @@
+{
+  "loss": [
+    125489.5390625,
+    51330.390625,
+    47167.515625,
+    44767.375,
+    43692.6875,
+    43005.29296875,
+    42384.7421875,
+    42023.90625,
+    41666.9296875,
+    41155.05859375,
+    40768.59375,
+    40673.7109375,
+    40660.3671875,
+    40635.84375,
+    40189.2421875,
+    40298.28515625,
+    40418.390625
+  ],
+  "l0": [
+    500.0,
+    494.5947265625,
+    495.097900390625,
+    493.616943359375,
+    494.564208984375,
+    495.037109375,
+    493.1884765625,
+    494.356689453125,
+    493.68359375,
+    493.552001953125,
+    492.6875,
+    492.47216796875,
+    492.65478515625,
+    492.886474609375,
+    491.590087890625,
+    493.7158203125,
+    492.57373046875
+  ],
+  "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": 0.01980758147084849,
+  "elapsed_s": 5632.007280349731,
+  "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": 2,
+  "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__seed2"
+}