Upload training_logs/txc_contrastive_h8_t8_kpos20_shifts2__seed2.json with huggingface_hub
Browse files
training_logs/txc_contrastive_h8_t8_kpos20_shifts2__seed2.json
ADDED
|
@@ -0,0 +1,76 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"loss": [
|
| 3 |
+
63993.8203125,
|
| 4 |
+
37476.90234375,
|
| 5 |
+
35145.15234375,
|
| 6 |
+
34236.21875,
|
| 7 |
+
33460.609375,
|
| 8 |
+
33247.62109375,
|
| 9 |
+
32855.45703125,
|
| 10 |
+
32638.46484375,
|
| 11 |
+
32370.107421875,
|
| 12 |
+
32400.365234375,
|
| 13 |
+
31997.744140625,
|
| 14 |
+
32004.169921875,
|
| 15 |
+
31975.552734375,
|
| 16 |
+
31822.931640625,
|
| 17 |
+
31747.025390625,
|
| 18 |
+
31727.166015625
|
| 19 |
+
],
|
| 20 |
+
"l0": [
|
| 21 |
+
160.0,
|
| 22 |
+
160.0,
|
| 23 |
+
160.0,
|
| 24 |
+
160.0,
|
| 25 |
+
160.0,
|
| 26 |
+
160.0,
|
| 27 |
+
160.0,
|
| 28 |
+
160.0,
|
| 29 |
+
160.0,
|
| 30 |
+
160.0,
|
| 31 |
+
160.0,
|
| 32 |
+
160.0,
|
| 33 |
+
160.0,
|
| 34 |
+
160.0,
|
| 35 |
+
160.0,
|
| 36 |
+
160.0
|
| 37 |
+
],
|
| 38 |
+
"steps_logged": [
|
| 39 |
+
0,
|
| 40 |
+
200,
|
| 41 |
+
400,
|
| 42 |
+
600,
|
| 43 |
+
800,
|
| 44 |
+
1000,
|
| 45 |
+
1200,
|
| 46 |
+
1400,
|
| 47 |
+
1600,
|
| 48 |
+
1800,
|
| 49 |
+
2000,
|
| 50 |
+
2200,
|
| 51 |
+
2400,
|
| 52 |
+
2600,
|
| 53 |
+
2800,
|
| 54 |
+
3000
|
| 55 |
+
],
|
| 56 |
+
"final_step": 3000,
|
| 57 |
+
"converged": true,
|
| 58 |
+
"plateau_last": 0.018397963894626286,
|
| 59 |
+
"elapsed_s": 5539.234444141388,
|
| 60 |
+
"shifts": [
|
| 61 |
+
2
|
| 62 |
+
],
|
| 63 |
+
"matryoshka_h_size": 3686,
|
| 64 |
+
"alpha": 1.0,
|
| 65 |
+
"arch_id": "txc_contrastive_h8_t8_kpos20_shifts2",
|
| 66 |
+
"src_class": "TXCContrastiveMergeH8",
|
| 67 |
+
"src_module": "src.architectures.txc_contrastive_merge_h8",
|
| 68 |
+
"T": 8,
|
| 69 |
+
"k_pos": 20,
|
| 70 |
+
"k_win": 160,
|
| 71 |
+
"d_sae": 18432,
|
| 72 |
+
"d_in": 2304,
|
| 73 |
+
"subject_model": "google/gemma-2-2b",
|
| 74 |
+
"anchor_layer": 12,
|
| 75 |
+
"seed": 2
|
| 76 |
+
}
|