Upload training_logs/mlc_contrastive_alpha100_batchtopk__seed1.json with huggingface_hub

training_logs/mlc_contrastive_alpha100_batchtopk__seed1.json

@@ -0,0 +1,93 @@
+{
+  "loss": [
+    113.1244125366211,
+    15.078754425048828,
+    13.990482330322266,
+    13.722837448120117,
+    13.43752384185791,
+    13.301484107971191,
+    13.143375396728516,
+    13.06921100616455,
+    12.994958877563477,
+    12.895394325256348,
+    12.889880180358887,
+    12.839088439941406,
+    12.78973388671875,
+    12.753411293029785,
+    12.682119369506836,
+    12.63943099975586
+  ],
+  "l0": [
+    500.0,
+    490.70361328125,
+    488.095703125,
+    490.61767578125,
+    487.789794921875,
+    490.43359375,
+    488.31787109375,
+    489.744873046875,
+    490.695556640625,
+    489.99658203125,
+    490.54736328125,
+    490.084228515625,
+    491.463623046875,
+    490.50732421875,
+    488.458740234375,
+    489.27197265625
+  ],
+  "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": 0.019833510861077906,
+  "elapsed_s": 1356.9366526603699,
+  "h": 3686,
+  "alpha": 1.0,
+  "row": 5,
+  "arch_id": "mlc_contrastive_alpha100_batchtopk",
+  "arch": "mlc_contrastive_alpha100_batchtopk",
+  "group": 1,
+  "src_class": "MLCContrastive",
+  "src_module": "src.architectures.mlc_contrastive",
+  "T": null,
+  "T_max": null,
+  "t_sample": null,
+  "n_layers": 5,
+  "k_win": 500,
+  "k_pos": 100,
+  "shifts": [
+    1
+  ],
+  "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": "mlc_contrastive_alpha100_batchtopk__seed1"
+}