Upload training_logs/agentic_txc_02__seed2.json with huggingface_hub
Browse files
training_logs/agentic_txc_02__seed2.json
ADDED
|
@@ -0,0 +1,109 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"loss": [
|
| 3 |
+
88588.8515625,
|
| 4 |
+
13272.6826171875,
|
| 5 |
+
11230.08203125,
|
| 6 |
+
10447.7978515625,
|
| 7 |
+
10027.3544921875,
|
| 8 |
+
9824.888671875,
|
| 9 |
+
9686.921875,
|
| 10 |
+
9631.7236328125,
|
| 11 |
+
9532.1923828125,
|
| 12 |
+
9331.6337890625,
|
| 13 |
+
9233.6923828125,
|
| 14 |
+
9184.7578125,
|
| 15 |
+
9060.9970703125,
|
| 16 |
+
9093.2607421875,
|
| 17 |
+
9082.064453125,
|
| 18 |
+
8907.6015625,
|
| 19 |
+
8955.919921875,
|
| 20 |
+
8842.6884765625,
|
| 21 |
+
8937.6474609375,
|
| 22 |
+
8909.7255859375,
|
| 23 |
+
8823.203125
|
| 24 |
+
],
|
| 25 |
+
"l0": [
|
| 26 |
+
500.0,
|
| 27 |
+
496.3740234375,
|
| 28 |
+
495.755859375,
|
| 29 |
+
495.64453125,
|
| 30 |
+
495.14404296875,
|
| 31 |
+
494.255615234375,
|
| 32 |
+
494.2236328125,
|
| 33 |
+
494.0458984375,
|
| 34 |
+
493.953125,
|
| 35 |
+
493.642578125,
|
| 36 |
+
494.06494140625,
|
| 37 |
+
492.32861328125,
|
| 38 |
+
491.549072265625,
|
| 39 |
+
493.762939453125,
|
| 40 |
+
493.201171875,
|
| 41 |
+
493.247802734375,
|
| 42 |
+
492.772705078125,
|
| 43 |
+
492.216064453125,
|
| 44 |
+
493.815185546875,
|
| 45 |
+
492.09521484375,
|
| 46 |
+
491.924072265625
|
| 47 |
+
],
|
| 48 |
+
"steps_logged": [
|
| 49 |
+
0,
|
| 50 |
+
200,
|
| 51 |
+
400,
|
| 52 |
+
600,
|
| 53 |
+
800,
|
| 54 |
+
1000,
|
| 55 |
+
1200,
|
| 56 |
+
1400,
|
| 57 |
+
1600,
|
| 58 |
+
1800,
|
| 59 |
+
2000,
|
| 60 |
+
2200,
|
| 61 |
+
2400,
|
| 62 |
+
2600,
|
| 63 |
+
2800,
|
| 64 |
+
3000,
|
| 65 |
+
3200,
|
| 66 |
+
3400,
|
| 67 |
+
3600,
|
| 68 |
+
3800,
|
| 69 |
+
4000
|
| 70 |
+
],
|
| 71 |
+
"final_step": 4000,
|
| 72 |
+
"converged": true,
|
| 73 |
+
"plateau_last": 0.01896143984788191,
|
| 74 |
+
"elapsed_s": 3024.120314836502,
|
| 75 |
+
"n_scales": 3,
|
| 76 |
+
"gamma": 0.5,
|
| 77 |
+
"row": 8,
|
| 78 |
+
"arch_id": "agentic_txc_02",
|
| 79 |
+
"arch": "agentic_txc_02",
|
| 80 |
+
"group": 2,
|
| 81 |
+
"src_class": "MatryoshkaTXCDRContrastiveMultiscale",
|
| 82 |
+
"src_module": "src.architectures.matryoshka_txcdr_contrastive_multiscale",
|
| 83 |
+
"T": 5,
|
| 84 |
+
"T_max": null,
|
| 85 |
+
"t_sample": null,
|
| 86 |
+
"n_layers": null,
|
| 87 |
+
"k_win": 500,
|
| 88 |
+
"k_pos": 100,
|
| 89 |
+
"shifts": [
|
| 90 |
+
1,
|
| 91 |
+
2,
|
| 92 |
+
3
|
| 93 |
+
],
|
| 94 |
+
"alpha": null,
|
| 95 |
+
"seed": 2,
|
| 96 |
+
"d_in": 2304,
|
| 97 |
+
"d_sae": 18432,
|
| 98 |
+
"subject_model": "google/gemma-2-2b",
|
| 99 |
+
"anchor_layer": 12,
|
| 100 |
+
"mlc_layers": [
|
| 101 |
+
10,
|
| 102 |
+
11,
|
| 103 |
+
12,
|
| 104 |
+
13,
|
| 105 |
+
14
|
| 106 |
+
],
|
| 107 |
+
"phase": "phase7_unification",
|
| 108 |
+
"run_id": "agentic_txc_02__seed2"
|
| 109 |
+
}
|