Upload training_logs/txc_contrastive_h8_t2_kpos20_shifts2__seed42.json with huggingface_hub
Browse files
training_logs/txc_contrastive_h8_t2_kpos20_shifts2__seed42.json
ADDED
|
@@ -0,0 +1,79 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"loss": [
|
| 3 |
+
99038.65625,
|
| 4 |
+
35986.73828125,
|
| 5 |
+
28336.90234375,
|
| 6 |
+
26902.771484375,
|
| 7 |
+
26052.306640625,
|
| 8 |
+
25504.376953125,
|
| 9 |
+
25081.302734375,
|
| 10 |
+
24938.330078125,
|
| 11 |
+
24620.857421875,
|
| 12 |
+
24403.29296875,
|
| 13 |
+
24505.50390625,
|
| 14 |
+
24293.412109375,
|
| 15 |
+
24252.876953125,
|
| 16 |
+
23968.390625,
|
| 17 |
+
24089.78125,
|
| 18 |
+
24107.2890625,
|
| 19 |
+
24068.662109375
|
| 20 |
+
],
|
| 21 |
+
"l0": [
|
| 22 |
+
40.0,
|
| 23 |
+
40.0,
|
| 24 |
+
40.0,
|
| 25 |
+
40.0,
|
| 26 |
+
40.0,
|
| 27 |
+
40.0,
|
| 28 |
+
40.0,
|
| 29 |
+
40.0,
|
| 30 |
+
40.0,
|
| 31 |
+
40.0,
|
| 32 |
+
40.0,
|
| 33 |
+
40.0,
|
| 34 |
+
40.0,
|
| 35 |
+
40.0,
|
| 36 |
+
40.0,
|
| 37 |
+
40.0,
|
| 38 |
+
40.0
|
| 39 |
+
],
|
| 40 |
+
"steps_logged": [
|
| 41 |
+
0,
|
| 42 |
+
200,
|
| 43 |
+
400,
|
| 44 |
+
600,
|
| 45 |
+
800,
|
| 46 |
+
1000,
|
| 47 |
+
1200,
|
| 48 |
+
1400,
|
| 49 |
+
1600,
|
| 50 |
+
1800,
|
| 51 |
+
2000,
|
| 52 |
+
2200,
|
| 53 |
+
2400,
|
| 54 |
+
2600,
|
| 55 |
+
2800,
|
| 56 |
+
3000,
|
| 57 |
+
3200
|
| 58 |
+
],
|
| 59 |
+
"final_step": 3200,
|
| 60 |
+
"converged": true,
|
| 61 |
+
"plateau_last": 0.01852696816351773,
|
| 62 |
+
"elapsed_s": 2039.5590670108795,
|
| 63 |
+
"shifts": [
|
| 64 |
+
2
|
| 65 |
+
],
|
| 66 |
+
"matryoshka_h_size": 3686,
|
| 67 |
+
"alpha": 1.0,
|
| 68 |
+
"arch_id": "txc_contrastive_h8_t2_kpos20_shifts2",
|
| 69 |
+
"src_class": "TXCContrastiveMergeH8",
|
| 70 |
+
"src_module": "src.architectures.txc_contrastive_merge_h8",
|
| 71 |
+
"T": 2,
|
| 72 |
+
"k_pos": 20,
|
| 73 |
+
"k_win": 40,
|
| 74 |
+
"d_sae": 18432,
|
| 75 |
+
"d_in": 2304,
|
| 76 |
+
"subject_model": "google/gemma-2-2b",
|
| 77 |
+
"anchor_layer": 12,
|
| 78 |
+
"seed": 42
|
| 79 |
+
}
|