Upload training_logs/txc_bare_antidead_t6__seed42.json with huggingface_hub
Browse files
training_logs/txc_bare_antidead_t6__seed42.json
ADDED
|
@@ -0,0 +1,120 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"loss": [
|
| 3 |
+
13036.1240234375,
|
| 4 |
+
5839.79052734375,
|
| 5 |
+
5169.66357421875,
|
| 6 |
+
4787.41552734375,
|
| 7 |
+
4576.44482421875,
|
| 8 |
+
4405.8486328125,
|
| 9 |
+
4273.591796875,
|
| 10 |
+
4159.86865234375,
|
| 11 |
+
4090.8046875,
|
| 12 |
+
4027.442138671875,
|
| 13 |
+
3998.333984375,
|
| 14 |
+
3951.990234375,
|
| 15 |
+
3904.77294921875,
|
| 16 |
+
3890.715576171875,
|
| 17 |
+
3875.372802734375,
|
| 18 |
+
3821.909912109375,
|
| 19 |
+
3804.688232421875,
|
| 20 |
+
3804.86767578125,
|
| 21 |
+
3801.357421875,
|
| 22 |
+
3779.475341796875,
|
| 23 |
+
3757.30126953125,
|
| 24 |
+
3722.21484375,
|
| 25 |
+
3729.3564453125,
|
| 26 |
+
3717.6826171875,
|
| 27 |
+
3712.84912109375
|
| 28 |
+
],
|
| 29 |
+
"l0": [
|
| 30 |
+
500.0,
|
| 31 |
+
500.0,
|
| 32 |
+
500.0,
|
| 33 |
+
500.0,
|
| 34 |
+
500.0,
|
| 35 |
+
500.0,
|
| 36 |
+
500.0,
|
| 37 |
+
500.0,
|
| 38 |
+
500.0,
|
| 39 |
+
500.0,
|
| 40 |
+
500.0,
|
| 41 |
+
500.0,
|
| 42 |
+
500.0,
|
| 43 |
+
500.0,
|
| 44 |
+
500.0,
|
| 45 |
+
500.0,
|
| 46 |
+
500.0,
|
| 47 |
+
500.0,
|
| 48 |
+
500.0,
|
| 49 |
+
500.0,
|
| 50 |
+
500.0,
|
| 51 |
+
500.0,
|
| 52 |
+
500.0,
|
| 53 |
+
500.0,
|
| 54 |
+
500.0
|
| 55 |
+
],
|
| 56 |
+
"steps_logged": [
|
| 57 |
+
0,
|
| 58 |
+
200,
|
| 59 |
+
400,
|
| 60 |
+
600,
|
| 61 |
+
800,
|
| 62 |
+
1000,
|
| 63 |
+
1200,
|
| 64 |
+
1400,
|
| 65 |
+
1600,
|
| 66 |
+
1800,
|
| 67 |
+
2000,
|
| 68 |
+
2200,
|
| 69 |
+
2400,
|
| 70 |
+
2600,
|
| 71 |
+
2800,
|
| 72 |
+
3000,
|
| 73 |
+
3200,
|
| 74 |
+
3400,
|
| 75 |
+
3600,
|
| 76 |
+
3800,
|
| 77 |
+
4000,
|
| 78 |
+
4200,
|
| 79 |
+
4400,
|
| 80 |
+
4600,
|
| 81 |
+
4800
|
| 82 |
+
],
|
| 83 |
+
"final_step": 4800,
|
| 84 |
+
"converged": true,
|
| 85 |
+
"plateau_last": 0.01961331952904919,
|
| 86 |
+
"elapsed_s": 1407.0589573383331,
|
| 87 |
+
"T": 6,
|
| 88 |
+
"src_class": "TXCBareAntidead",
|
| 89 |
+
"n_seqs_used": 24000,
|
| 90 |
+
"ctx_used": 64,
|
| 91 |
+
"ctx_slice_direction": "last",
|
| 92 |
+
"row": 606,
|
| 93 |
+
"arch_id": "txc_bare_antidead_t6",
|
| 94 |
+
"arch": "txc_bare_antidead_t6",
|
| 95 |
+
"group": 99,
|
| 96 |
+
"src_module": "src.architectures.txc_bare_antidead",
|
| 97 |
+
"T_max": null,
|
| 98 |
+
"t_sample": null,
|
| 99 |
+
"n_layers": null,
|
| 100 |
+
"k_win": 500,
|
| 101 |
+
"k_pos": 83,
|
| 102 |
+
"shifts": null,
|
| 103 |
+
"alpha": null,
|
| 104 |
+
"gamma": null,
|
| 105 |
+
"n_scales": null,
|
| 106 |
+
"seed": 42,
|
| 107 |
+
"d_in": 2304,
|
| 108 |
+
"d_sae": 18432,
|
| 109 |
+
"subject_model": "google/gemma-2-2b",
|
| 110 |
+
"anchor_layer": 12,
|
| 111 |
+
"mlc_layers": [
|
| 112 |
+
10,
|
| 113 |
+
11,
|
| 114 |
+
12,
|
| 115 |
+
13,
|
| 116 |
+
14
|
| 117 |
+
],
|
| 118 |
+
"phase": "phase7_unification",
|
| 119 |
+
"run_id": "txc_bare_antidead_t6__seed42"
|
| 120 |
+
}
|