Upload training_logs/txc_bare_antidead_t10__seed1.json with huggingface_hub
Browse files
training_logs/txc_bare_antidead_t10__seed1.json
ADDED
|
@@ -0,0 +1,102 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"loss": [
|
| 3 |
+
18078.052734375,
|
| 4 |
+
6580.3515625,
|
| 5 |
+
5834.27587890625,
|
| 6 |
+
5468.43359375,
|
| 7 |
+
5193.6572265625,
|
| 8 |
+
5057.0078125,
|
| 9 |
+
4964.65478515625,
|
| 10 |
+
4855.59912109375,
|
| 11 |
+
4828.58203125,
|
| 12 |
+
4738.08740234375,
|
| 13 |
+
4721.66748046875,
|
| 14 |
+
4689.2666015625,
|
| 15 |
+
4666.70068359375,
|
| 16 |
+
4659.56396484375,
|
| 17 |
+
4595.5849609375,
|
| 18 |
+
4575.4833984375,
|
| 19 |
+
4596.00537109375,
|
| 20 |
+
4561.62646484375,
|
| 21 |
+
4562.41552734375,
|
| 22 |
+
4571.14697265625
|
| 23 |
+
],
|
| 24 |
+
"l0": [
|
| 25 |
+
500.0,
|
| 26 |
+
496.53076171875,
|
| 27 |
+
495.14404296875,
|
| 28 |
+
495.56884765625,
|
| 29 |
+
493.7509765625,
|
| 30 |
+
493.21337890625,
|
| 31 |
+
493.2578125,
|
| 32 |
+
493.10302734375,
|
| 33 |
+
494.2763671875,
|
| 34 |
+
492.20166015625,
|
| 35 |
+
492.78369140625,
|
| 36 |
+
493.066650390625,
|
| 37 |
+
493.89013671875,
|
| 38 |
+
494.232666015625,
|
| 39 |
+
490.362548828125,
|
| 40 |
+
491.499267578125,
|
| 41 |
+
493.197998046875,
|
| 42 |
+
491.828857421875,
|
| 43 |
+
492.11083984375,
|
| 44 |
+
491.9853515625
|
| 45 |
+
],
|
| 46 |
+
"steps_logged": [
|
| 47 |
+
0,
|
| 48 |
+
200,
|
| 49 |
+
400,
|
| 50 |
+
600,
|
| 51 |
+
800,
|
| 52 |
+
1000,
|
| 53 |
+
1200,
|
| 54 |
+
1400,
|
| 55 |
+
1600,
|
| 56 |
+
1800,
|
| 57 |
+
2000,
|
| 58 |
+
2200,
|
| 59 |
+
2400,
|
| 60 |
+
2600,
|
| 61 |
+
2800,
|
| 62 |
+
3000,
|
| 63 |
+
3200,
|
| 64 |
+
3400,
|
| 65 |
+
3600,
|
| 66 |
+
3800
|
| 67 |
+
],
|
| 68 |
+
"final_step": 3800,
|
| 69 |
+
"converged": true,
|
| 70 |
+
"plateau_last": 0.019976440153727675,
|
| 71 |
+
"elapsed_s": 2170.5987079143524,
|
| 72 |
+
"row": 10,
|
| 73 |
+
"arch_id": "txc_bare_antidead_t10",
|
| 74 |
+
"arch": "txc_bare_antidead_t10",
|
| 75 |
+
"group": 2,
|
| 76 |
+
"src_class": "TXCBareAntidead",
|
| 77 |
+
"src_module": "src.architectures.txc_bare_antidead",
|
| 78 |
+
"T": 10,
|
| 79 |
+
"T_max": null,
|
| 80 |
+
"t_sample": null,
|
| 81 |
+
"n_layers": null,
|
| 82 |
+
"k_win": 500,
|
| 83 |
+
"k_pos": 50,
|
| 84 |
+
"shifts": null,
|
| 85 |
+
"alpha": null,
|
| 86 |
+
"gamma": null,
|
| 87 |
+
"n_scales": null,
|
| 88 |
+
"seed": 1,
|
| 89 |
+
"d_in": 2304,
|
| 90 |
+
"d_sae": 18432,
|
| 91 |
+
"subject_model": "google/gemma-2-2b",
|
| 92 |
+
"anchor_layer": 12,
|
| 93 |
+
"mlc_layers": [
|
| 94 |
+
10,
|
| 95 |
+
11,
|
| 96 |
+
12,
|
| 97 |
+
13,
|
| 98 |
+
14
|
| 99 |
+
],
|
| 100 |
+
"phase": "phase7_unification",
|
| 101 |
+
"run_id": "txc_bare_antidead_t10__seed1"
|
| 102 |
+
}
|