Upload training_logs/spatial_matry_h8_t10_kpos20_shifts2_pref3686_9216_18432_sub1_5_10_nested_uniform_contr__seed42.json with huggingface_hub

training_logs/spatial_matry_h8_t10_kpos20_shifts2_pref3686_9216_18432_sub1_5_10_nested_uniform_contr__seed42.json

@@ -0,0 +1,111 @@
+{
+  "loss": [
+    85536.9453125,
+    38466.1328125,
+    35270.8046875,
+    33850.078125,
+    32935.3984375,
+    32410.20703125,
+    31999.58984375,
+    31649.3984375,
+    31516.763671875,
+    31348.26171875,
+    31197.630859375,
+    30997.08984375,
+    31125.05859375,
+    30974.0,
+    30723.126953125,
+    30741.302734375
+  ],
+  "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
+  ],
+  "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": 0.019979922317185896,
+  "elapsed_s": 4568.830448389053,
+  "shifts": [
+    2
+  ],
+  "level_prefix_sizes": [
+    3686,
+    9216,
+    18432
+  ],
+  "level_subset_sizes": [
+    1,
+    5,
+    10
+  ],
+  "nested": true,
+  "subset_mode": "uniform",
+  "enable_contrastive": true,
+  "matryoshka_h_size": 3686,
+  "final_step_wall_s": 4593.298326253891,
+  "row": -1,
+  "arch_id": "spatial_matry_h8_t10_kpos20_shifts2_pref3686_9216_18432_sub1_5_10_nested_uniform_contr",
+  "arch": "spatial_matry_h8_t10_kpos20_shifts2_pref3686_9216_18432_sub1_5_10_nested_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": 42,
+  "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), nested, sample=uniform, contrastive=True)",
+  "run_id": "spatial_matry_h8_t10_kpos20_shifts2_pref3686_9216_18432_sub1_5_10_nested_uniform_contr__seed42"
+}