Upload training_logs/it_mlc_contrastive_alpha100_batchtopk__seed42.json with huggingface_hub
Browse files
training_logs/it_mlc_contrastive_alpha100_batchtopk__seed42.json
ADDED
|
@@ -0,0 +1,99 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"loss": [
|
| 3 |
+
157.5962371826172,
|
| 4 |
+
19.078102111816406,
|
| 5 |
+
17.303621292114258,
|
| 6 |
+
16.6531925201416,
|
| 7 |
+
16.436206817626953,
|
| 8 |
+
16.142433166503906,
|
| 9 |
+
15.9730863571167,
|
| 10 |
+
15.844514846801758,
|
| 11 |
+
15.75390911102295,
|
| 12 |
+
15.598241806030273,
|
| 13 |
+
15.517053604125977,
|
| 14 |
+
15.446672439575195,
|
| 15 |
+
15.358577728271484,
|
| 16 |
+
15.298223495483398,
|
| 17 |
+
15.256269454956055,
|
| 18 |
+
15.288846015930176,
|
| 19 |
+
15.203807830810547,
|
| 20 |
+
15.149807929992676
|
| 21 |
+
],
|
| 22 |
+
"l0": [
|
| 23 |
+
500.0,
|
| 24 |
+
500.0,
|
| 25 |
+
500.0,
|
| 26 |
+
500.0,
|
| 27 |
+
500.0,
|
| 28 |
+
500.0,
|
| 29 |
+
500.0,
|
| 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 |
+
],
|
| 42 |
+
"steps_logged": [
|
| 43 |
+
0,
|
| 44 |
+
200,
|
| 45 |
+
400,
|
| 46 |
+
600,
|
| 47 |
+
800,
|
| 48 |
+
1000,
|
| 49 |
+
1200,
|
| 50 |
+
1400,
|
| 51 |
+
1600,
|
| 52 |
+
1800,
|
| 53 |
+
2000,
|
| 54 |
+
2200,
|
| 55 |
+
2400,
|
| 56 |
+
2600,
|
| 57 |
+
2800,
|
| 58 |
+
3000,
|
| 59 |
+
3200,
|
| 60 |
+
3400
|
| 61 |
+
],
|
| 62 |
+
"final_step": 3400,
|
| 63 |
+
"converged": true,
|
| 64 |
+
"plateau_last": 0.019021697259006043,
|
| 65 |
+
"elapsed_s": 3835.5307714939117,
|
| 66 |
+
"h": 3686,
|
| 67 |
+
"alpha": 1.0,
|
| 68 |
+
"row": 5,
|
| 69 |
+
"arch_id": "mlc_contrastive_alpha100_batchtopk",
|
| 70 |
+
"arch": "mlc_contrastive_alpha100_batchtopk",
|
| 71 |
+
"group": 1,
|
| 72 |
+
"src_class": "MLCContrastive",
|
| 73 |
+
"src_module": "src.architectures.mlc_contrastive",
|
| 74 |
+
"T": null,
|
| 75 |
+
"T_max": null,
|
| 76 |
+
"t_sample": null,
|
| 77 |
+
"n_layers": 5,
|
| 78 |
+
"k_win": 500,
|
| 79 |
+
"k_pos": 100,
|
| 80 |
+
"shifts": [
|
| 81 |
+
1
|
| 82 |
+
],
|
| 83 |
+
"gamma": null,
|
| 84 |
+
"n_scales": null,
|
| 85 |
+
"seed": 42,
|
| 86 |
+
"d_in": 2304,
|
| 87 |
+
"d_sae": 18432,
|
| 88 |
+
"subject_model": "google/gemma-2-2b-it",
|
| 89 |
+
"anchor_layer": 13,
|
| 90 |
+
"mlc_layers": [
|
| 91 |
+
11,
|
| 92 |
+
12,
|
| 93 |
+
13,
|
| 94 |
+
14,
|
| 95 |
+
15
|
| 96 |
+
],
|
| 97 |
+
"phase": "phase7_unification",
|
| 98 |
+
"run_id": "it_mlc_contrastive_alpha100_batchtopk__seed42"
|
| 99 |
+
}
|