Upload results.json with huggingface_hub
Browse files- results.json +94 -0
results.json
ADDED
|
@@ -0,0 +1,94 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"d_sae": 16384,
|
| 3 |
+
"actual_l0": 19.6816084,
|
| 4 |
+
"effective_l0": 12,
|
| 5 |
+
"cross_entropy": 90.97328370361328,
|
| 6 |
+
"sae_type": "saebench",
|
| 7 |
+
"sae_release": "canrager/saebench_gemma-2-2b_width-2pow14_date-0107",
|
| 8 |
+
"num_tokens": 10000000,
|
| 9 |
+
"num_epochs": 1,
|
| 10 |
+
"lr": 0.0005,
|
| 11 |
+
"loaded_dag": null,
|
| 12 |
+
"allow_negative_influence": false,
|
| 13 |
+
"history": {
|
| 14 |
+
"train_loss": [
|
| 15 |
+
123.50254767281669,
|
| 16 |
+
116.86768455505371,
|
| 17 |
+
111.47526424952916,
|
| 18 |
+
107.15069541931152,
|
| 19 |
+
103.60219999040876,
|
| 20 |
+
100.70831909179688,
|
| 21 |
+
98.08268596104213,
|
| 22 |
+
95.82541280473981,
|
| 23 |
+
93.8609597342355,
|
| 24 |
+
91.96091542925153,
|
| 25 |
+
91.88661466326032
|
| 26 |
+
],
|
| 27 |
+
"train_steps": [
|
| 28 |
+
140,
|
| 29 |
+
280,
|
| 30 |
+
420,
|
| 31 |
+
560,
|
| 32 |
+
700,
|
| 33 |
+
840,
|
| 34 |
+
980,
|
| 35 |
+
1120,
|
| 36 |
+
1260,
|
| 37 |
+
1400,
|
| 38 |
+
1407
|
| 39 |
+
],
|
| 40 |
+
"val_loss": [
|
| 41 |
+
119.9688098022461,
|
| 42 |
+
113.95728345947266,
|
| 43 |
+
109.10677463378906,
|
| 44 |
+
105.2270860961914,
|
| 45 |
+
102.0273933959961,
|
| 46 |
+
99.294438671875,
|
| 47 |
+
96.89552971191407,
|
| 48 |
+
94.75061934814453,
|
| 49 |
+
92.81334992675781,
|
| 50 |
+
91.05568310546874,
|
| 51 |
+
90.97328370361328
|
| 52 |
+
],
|
| 53 |
+
"val_cross_entropy": [
|
| 54 |
+
119.9688098022461,
|
| 55 |
+
113.95728345947266,
|
| 56 |
+
109.10677463378906,
|
| 57 |
+
105.2270860961914,
|
| 58 |
+
102.0273933959961,
|
| 59 |
+
99.294438671875,
|
| 60 |
+
96.89552971191407,
|
| 61 |
+
94.75061934814453,
|
| 62 |
+
92.81334992675781,
|
| 63 |
+
91.05568310546874,
|
| 64 |
+
90.97328370361328
|
| 65 |
+
],
|
| 66 |
+
"effective_l0": [
|
| 67 |
+
16,
|
| 68 |
+
15,
|
| 69 |
+
14,
|
| 70 |
+
14,
|
| 71 |
+
13,
|
| 72 |
+
13,
|
| 73 |
+
13,
|
| 74 |
+
12,
|
| 75 |
+
12,
|
| 76 |
+
12,
|
| 77 |
+
12
|
| 78 |
+
],
|
| 79 |
+
"val_steps": [
|
| 80 |
+
140,
|
| 81 |
+
280,
|
| 82 |
+
420,
|
| 83 |
+
560,
|
| 84 |
+
700,
|
| 85 |
+
840,
|
| 86 |
+
980,
|
| 87 |
+
1120,
|
| 88 |
+
1260,
|
| 89 |
+
1400,
|
| 90 |
+
1407
|
| 91 |
+
]
|
| 92 |
+
},
|
| 93 |
+
"sae_id": "gemma-2-2b_matryoshka_batch_top_k_width-2pow14_date-0107/resid_post_layer_12/trainer_0"
|
| 94 |
+
}
|