Upload training_logs/spatial_matry_h8_t10_kpos20_shifts2_pref3686_9216_18432_sub1_5_10_indep_uniform_contr__seed2.json with huggingface_hub

training_logs/spatial_matry_h8_t10_kpos20_shifts2_pref3686_9216_18432_sub1_5_10_indep_uniform_contr__seed2.json

@@ -0,0 +1,114 @@
+{
+  "loss": [
+    86365.5390625,
+    38214.13671875,
+    35153.27734375,
+    33679.21484375,
+    32901.421875,
+    32410.71484375,
+    31843.916015625,
+    31754.787109375,
+    31482.669921875,
+    31360.029296875,
+    31397.2734375,
+    31117.8359375,
+    30917.80078125,
+    31052.041015625,
+    30872.521484375,
+    30697.71875,
+    30805.38671875
+  ],
+  "l0": [
+    200.0,
+    200.0,
+    200.0,
+    200.0,
+    200.0,
+    200.0,
+    200.0,
+    200.0,
+    200.0,
+    200.0,
+    200.0,
+    200.0,
+    200.0,
+    200.0,
+    200.0,
+    200.0,
+    200.0
+  ],
+  "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": 0.01761238136726913,
+  "elapsed_s": 12385.501829624176,
+  "shifts": [
+    2
+  ],
+  "level_prefix_sizes": [
+    3686,
+    9216,
+    18432
+  ],
+  "level_subset_sizes": [
+    1,
+    5,
+    10
+  ],
+  "nested": false,
+  "subset_mode": "uniform",
+  "enable_contrastive": true,
+  "matryoshka_h_size": 3686,
+  "final_step_wall_s": 12447.54460477829,
+  "row": -1,
+  "arch_id": "spatial_matry_h8_t10_kpos20_shifts2_pref3686_9216_18432_sub1_5_10_indep_uniform_contr",
+  "arch": "spatial_matry_h8_t10_kpos20_shifts2_pref3686_9216_18432_sub1_5_10_indep_uniform_contr",
+  "group": "deadzone_escape_phase2",
+  "src_class": "SpatialMatryoshkaH8",
+  "src_module": "src.architectures.spatial_matryoshka_h8",
+  "T": 10,
+  "T_max": null,
+  "t_sample": null,
+  "n_layers": null,
+  "k_win": 200,
+  "k_pos": 20,
+  "sigma_range": null,
+  "n_gaussians": 1,
+  "alpha": 1.0,
+  "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",
+  "purpose": "Han's deadzone-escape: low-rank features must reconstruct any single position (position-flexible); deeper features add compositional cross-position info. Tests whether T=10/20 with this loss escapes the T=2-5 deadzone.",
+  "recipe": "SpatialMatryoshkaH8 \u2014 H8 stack at T=10 k_pos=20 with random-subset Matryoshka decoder loss (prefixes=(3686, 9216, 18432), subsets=(1, 5, 10), indep, sample=uniform, contrastive=True)",
+  "run_id": "spatial_matry_h8_t10_kpos20_shifts2_pref3686_9216_18432_sub1_5_10_indep_uniform_contr__seed2"
+}