Upload sp3_spine_text_33M.json with huggingface_hub
Browse files- sp3_spine_text_33M.json +153 -0
sp3_spine_text_33M.json
ADDED
|
@@ -0,0 +1,153 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"screen": "SP-2-SPINE-LM",
|
| 3 |
+
"seed": 0,
|
| 4 |
+
"steps": 4000,
|
| 5 |
+
"batch": 32,
|
| 6 |
+
"block": 512,
|
| 7 |
+
"lrs": [
|
| 8 |
+
0.001
|
| 9 |
+
],
|
| 10 |
+
"text_mb": 96.0,
|
| 11 |
+
"d_spine": 640,
|
| 12 |
+
"layers_spine": 8,
|
| 13 |
+
"n_spine": 33137920,
|
| 14 |
+
"tf_mult": 1.0,
|
| 15 |
+
"d_transformer": 672,
|
| 16 |
+
"layers_transformer": 6,
|
| 17 |
+
"n_transformer": 33257952,
|
| 18 |
+
"device": "cuda",
|
| 19 |
+
"thresholds": {
|
| 20 |
+
"pass": 1.05,
|
| 21 |
+
"concern": 1.15
|
| 22 |
+
},
|
| 23 |
+
"corpora": {
|
| 24 |
+
"text": {
|
| 25 |
+
"spine_lr0.001": {
|
| 26 |
+
"best_bpb": 1.5229665040969849,
|
| 27 |
+
"history": [
|
| 28 |
+
{
|
| 29 |
+
"step": 0,
|
| 30 |
+
"train_bits": 8.129907461277208,
|
| 31 |
+
"eval_bpb": 5.5101518630981445,
|
| 32 |
+
"best": 5.5101518630981445,
|
| 33 |
+
"kBps": 11.0
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"step": 250,
|
| 37 |
+
"train_bits": 2.1948707954659055,
|
| 38 |
+
"eval_bpb": 2.1676111221313477,
|
| 39 |
+
"best": 2.1676111221313477,
|
| 40 |
+
"kBps": 19.2
|
| 41 |
+
},
|
| 42 |
+
{
|
| 43 |
+
"step": 500,
|
| 44 |
+
"train_bits": 1.9935468724699963,
|
| 45 |
+
"eval_bpb": 1.9912030696868896,
|
| 46 |
+
"best": 1.9912030696868896,
|
| 47 |
+
"kBps": 19.1
|
| 48 |
+
},
|
| 49 |
+
{
|
| 50 |
+
"step": 750,
|
| 51 |
+
"train_bits": 1.8484287715852128,
|
| 52 |
+
"eval_bpb": 1.8803513050079346,
|
| 53 |
+
"best": 1.8803513050079346,
|
| 54 |
+
"kBps": 19.1
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"step": 1000,
|
| 58 |
+
"train_bits": 1.8020306041048408,
|
| 59 |
+
"eval_bpb": 1.8245058059692383,
|
| 60 |
+
"best": 1.8245058059692383,
|
| 61 |
+
"kBps": 19.0
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"step": 1250,
|
| 65 |
+
"train_bits": 1.7574234639505542,
|
| 66 |
+
"eval_bpb": 1.7559597492218018,
|
| 67 |
+
"best": 1.7559597492218018,
|
| 68 |
+
"kBps": 19.0
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"step": 1500,
|
| 72 |
+
"train_bits": 1.692556251576676,
|
| 73 |
+
"eval_bpb": 1.7225568294525146,
|
| 74 |
+
"best": 1.7225568294525146,
|
| 75 |
+
"kBps": 19.0
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"step": 1750,
|
| 79 |
+
"train_bits": 1.7061122680829843,
|
| 80 |
+
"eval_bpb": 1.6795181035995483,
|
| 81 |
+
"best": 1.6795181035995483,
|
| 82 |
+
"kBps": 19.0
|
| 83 |
+
},
|
| 84 |
+
{
|
| 85 |
+
"step": 2000,
|
| 86 |
+
"train_bits": 1.6381732735820456,
|
| 87 |
+
"eval_bpb": 1.6360663175582886,
|
| 88 |
+
"best": 1.6360663175582886,
|
| 89 |
+
"kBps": 19.0
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"step": 2250,
|
| 93 |
+
"train_bits": 1.5964389316277712,
|
| 94 |
+
"eval_bpb": 1.6126888990402222,
|
| 95 |
+
"best": 1.6126888990402222,
|
| 96 |
+
"kBps": 19.0
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"step": 2500,
|
| 100 |
+
"train_bits": 1.5226208821897382,
|
| 101 |
+
"eval_bpb": 1.5832231044769287,
|
| 102 |
+
"best": 1.5832231044769287,
|
| 103 |
+
"kBps": 19.0
|
| 104 |
+
},
|
| 105 |
+
{
|
| 106 |
+
"step": 2750,
|
| 107 |
+
"train_bits": 1.5009390293954639,
|
| 108 |
+
"eval_bpb": 1.5672656297683716,
|
| 109 |
+
"best": 1.5672656297683716,
|
| 110 |
+
"kBps": 19.0
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"step": 3000,
|
| 114 |
+
"train_bits": 1.5241976191328486,
|
| 115 |
+
"eval_bpb": 1.5502691268920898,
|
| 116 |
+
"best": 1.5502691268920898,
|
| 117 |
+
"kBps": 19.0
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"step": 3250,
|
| 121 |
+
"train_bits": 1.526238021302685,
|
| 122 |
+
"eval_bpb": 1.5373672246932983,
|
| 123 |
+
"best": 1.5373672246932983,
|
| 124 |
+
"kBps": 19.0
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"step": 3500,
|
| 128 |
+
"train_bits": 1.4832688439152155,
|
| 129 |
+
"eval_bpb": 1.531510591506958,
|
| 130 |
+
"best": 1.531510591506958,
|
| 131 |
+
"kBps": 19.0
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"step": 3750,
|
| 135 |
+
"train_bits": 1.4648457183894898,
|
| 136 |
+
"eval_bpb": 1.5232207775115967,
|
| 137 |
+
"best": 1.5232207775115967,
|
| 138 |
+
"kBps": 19.0
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"step": 3999,
|
| 142 |
+
"train_bits": 1.51629054477678,
|
| 143 |
+
"eval_bpb": 1.5229665040969849,
|
| 144 |
+
"best": 1.5229665040969849,
|
| 145 |
+
"kBps": 19.0
|
| 146 |
+
}
|
| 147 |
+
],
|
| 148 |
+
"wall_s": 3449.4
|
| 149 |
+
}
|
| 150 |
+
}
|
| 151 |
+
},
|
| 152 |
+
"partial": true
|
| 153 |
+
}
|