Upload training_logs/txc_contrastive_h8_t2_kpos20_shifts2__seed2.json with huggingface_hub
Browse files
training_logs/txc_contrastive_h8_t2_kpos20_shifts2__seed2.json
ADDED
|
@@ -0,0 +1,76 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"loss": [
|
| 3 |
+
88813.3515625,
|
| 4 |
+
33919.0859375,
|
| 5 |
+
28151.41015625,
|
| 6 |
+
26574.19140625,
|
| 7 |
+
25957.111328125,
|
| 8 |
+
25535.26171875,
|
| 9 |
+
24991.052734375,
|
| 10 |
+
24703.8984375,
|
| 11 |
+
24675.486328125,
|
| 12 |
+
24477.263671875,
|
| 13 |
+
24300.0625,
|
| 14 |
+
24341.671875,
|
| 15 |
+
24063.431640625,
|
| 16 |
+
24102.466796875,
|
| 17 |
+
24171.744140625,
|
| 18 |
+
24071.5390625
|
| 19 |
+
],
|
| 20 |
+
"l0": [
|
| 21 |
+
40.0,
|
| 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 |
+
],
|
| 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": 0.01946369211085733,
|
| 59 |
+
"elapsed_s": 1918.7778370380402,
|
| 60 |
+
"shifts": [
|
| 61 |
+
2
|
| 62 |
+
],
|
| 63 |
+
"matryoshka_h_size": 3686,
|
| 64 |
+
"alpha": 1.0,
|
| 65 |
+
"arch_id": "txc_contrastive_h8_t2_kpos20_shifts2",
|
| 66 |
+
"src_class": "TXCContrastiveMergeH8",
|
| 67 |
+
"src_module": "src.architectures.txc_contrastive_merge_h8",
|
| 68 |
+
"T": 2,
|
| 69 |
+
"k_pos": 20,
|
| 70 |
+
"k_win": 40,
|
| 71 |
+
"d_sae": 18432,
|
| 72 |
+
"d_in": 2304,
|
| 73 |
+
"subject_model": "google/gemma-2-2b",
|
| 74 |
+
"anchor_layer": 12,
|
| 75 |
+
"seed": 2
|
| 76 |
+
}
|