Upload folder using huggingface_hub
Browse files- optimizer.pt +3 -0
- pytorch_model.bin +1 -1
- rng_state.pth +3 -0
- scheduler.pt +3 -0
- trainer_state.json +597 -0
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8eea091ae0ea2a79b4e438a32407d90f9deb9ac787787ec42e07f81e8d5c5c18
|
| 3 |
+
size 4520032097
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 2276530250
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7bd9e897a8e77b48d7e4d168454b46c70594862ee0f1a8d2983f0c8bd8fafe44
|
| 3 |
size 2276530250
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:af0afd1672ec86a039c699898e5ac47a0e763960c21a7ef096da5ceaf59011d6
|
| 3 |
+
size 14244
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ca384772c61308bd01372abf5bad1a0e621b2de503b7c92b489d09b8e7ec4b76
|
| 3 |
+
size 1064
|
trainer_state.json
ADDED
|
@@ -0,0 +1,597 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 2.039255671679837,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 8000,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.02549069589599796,
|
| 13 |
+
"grad_norm": 3183.12646484375,
|
| 14 |
+
"learning_rate": 2.5e-06,
|
| 15 |
+
"loss": 227.1212,
|
| 16 |
+
"step": 100
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.05098139179199592,
|
| 20 |
+
"grad_norm": 363.3997802734375,
|
| 21 |
+
"learning_rate": 5e-06,
|
| 22 |
+
"loss": 88.4222,
|
| 23 |
+
"step": 200
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.07647208768799388,
|
| 27 |
+
"grad_norm": 263.5315246582031,
|
| 28 |
+
"learning_rate": 7.500000000000001e-06,
|
| 29 |
+
"loss": 67.8251,
|
| 30 |
+
"step": 300
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.10196278358399184,
|
| 34 |
+
"grad_norm": 328.4805908203125,
|
| 35 |
+
"learning_rate": 1e-05,
|
| 36 |
+
"loss": 63.1532,
|
| 37 |
+
"step": 400
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.1274534794799898,
|
| 41 |
+
"grad_norm": 496.2308044433594,
|
| 42 |
+
"learning_rate": 9.995728791936505e-06,
|
| 43 |
+
"loss": 60.2136,
|
| 44 |
+
"step": 500
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.15294417537598776,
|
| 48 |
+
"grad_norm": 405.20025634765625,
|
| 49 |
+
"learning_rate": 9.98292246503335e-06,
|
| 50 |
+
"loss": 59.0975,
|
| 51 |
+
"step": 600
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.17843487127198573,
|
| 55 |
+
"grad_norm": 451.58197021484375,
|
| 56 |
+
"learning_rate": 9.961602898685225e-06,
|
| 57 |
+
"loss": 55.2574,
|
| 58 |
+
"step": 700
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.20392556716798368,
|
| 62 |
+
"grad_norm": 127.22090148925781,
|
| 63 |
+
"learning_rate": 9.931806517013612e-06,
|
| 64 |
+
"loss": 56.7531,
|
| 65 |
+
"step": 800
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.22941626306398163,
|
| 69 |
+
"grad_norm": 431.8385925292969,
|
| 70 |
+
"learning_rate": 9.893584226636773e-06,
|
| 71 |
+
"loss": 57.5419,
|
| 72 |
+
"step": 900
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.2549069589599796,
|
| 76 |
+
"grad_norm": 249.0337677001953,
|
| 77 |
+
"learning_rate": 9.847001329696653e-06,
|
| 78 |
+
"loss": 57.2545,
|
| 79 |
+
"step": 1000
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.28039765485597756,
|
| 83 |
+
"grad_norm": 726.9508056640625,
|
| 84 |
+
"learning_rate": 9.792137412291265e-06,
|
| 85 |
+
"loss": 56.1048,
|
| 86 |
+
"step": 1100
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.3058883507519755,
|
| 90 |
+
"grad_norm": 811.3701171875,
|
| 91 |
+
"learning_rate": 9.729086208503174e-06,
|
| 92 |
+
"loss": 53.1675,
|
| 93 |
+
"step": 1200
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.33137904664797346,
|
| 97 |
+
"grad_norm": 754.3577270507812,
|
| 98 |
+
"learning_rate": 9.657955440256396e-06,
|
| 99 |
+
"loss": 54.9404,
|
| 100 |
+
"step": 1300
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.35686974254397147,
|
| 104 |
+
"grad_norm": 267.7255554199219,
|
| 105 |
+
"learning_rate": 9.578866633275289e-06,
|
| 106 |
+
"loss": 55.7077,
|
| 107 |
+
"step": 1400
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.3823604384399694,
|
| 111 |
+
"grad_norm": 244.13253784179688,
|
| 112 |
+
"learning_rate": 9.491954909459895e-06,
|
| 113 |
+
"loss": 56.8997,
|
| 114 |
+
"step": 1500
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.40785113433596737,
|
| 118 |
+
"grad_norm": 302.5712890625,
|
| 119 |
+
"learning_rate": 9.397368756032445e-06,
|
| 120 |
+
"loss": 50.8917,
|
| 121 |
+
"step": 1600
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.4333418302319653,
|
| 125 |
+
"grad_norm": 660.6898193359375,
|
| 126 |
+
"learning_rate": 9.295269771849426e-06,
|
| 127 |
+
"loss": 49.8636,
|
| 128 |
+
"step": 1700
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.45883252612796327,
|
| 132 |
+
"grad_norm": 235.12942504882812,
|
| 133 |
+
"learning_rate": 9.185832391312644e-06,
|
| 134 |
+
"loss": 53.0454,
|
| 135 |
+
"step": 1800
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.4843232220239613,
|
| 139 |
+
"grad_norm": 231.6936492919922,
|
| 140 |
+
"learning_rate": 9.069243586350976e-06,
|
| 141 |
+
"loss": 56.1398,
|
| 142 |
+
"step": 1900
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.5098139179199592,
|
| 146 |
+
"grad_norm": 460.98486328125,
|
| 147 |
+
"learning_rate": 8.94570254698197e-06,
|
| 148 |
+
"loss": 52.8867,
|
| 149 |
+
"step": 2000
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.5353046138159572,
|
| 153 |
+
"grad_norm": 567.8645629882812,
|
| 154 |
+
"learning_rate": 8.815420340999034e-06,
|
| 155 |
+
"loss": 56.6194,
|
| 156 |
+
"step": 2100
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.5607953097119551,
|
| 160 |
+
"grad_norm": 158.99224853515625,
|
| 161 |
+
"learning_rate": 8.67861955336566e-06,
|
| 162 |
+
"loss": 50.311,
|
| 163 |
+
"step": 2200
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.5862860056079531,
|
| 167 |
+
"grad_norm": 463.69635009765625,
|
| 168 |
+
"learning_rate": 8.535533905932739e-06,
|
| 169 |
+
"loss": 48.126,
|
| 170 |
+
"step": 2300
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.611776701503951,
|
| 174 |
+
"grad_norm": 346.60333251953125,
|
| 175 |
+
"learning_rate": 8.386407858128707e-06,
|
| 176 |
+
"loss": 51.602,
|
| 177 |
+
"step": 2400
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.637267397399949,
|
| 181 |
+
"grad_norm": 430.1127624511719,
|
| 182 |
+
"learning_rate": 8.231496189304704e-06,
|
| 183 |
+
"loss": 50.3868,
|
| 184 |
+
"step": 2500
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.6627580932959469,
|
| 188 |
+
"grad_norm": 430.7168884277344,
|
| 189 |
+
"learning_rate": 8.071063563448341e-06,
|
| 190 |
+
"loss": 49.5458,
|
| 191 |
+
"step": 2600
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.688248789191945,
|
| 195 |
+
"grad_norm": 409.49114990234375,
|
| 196 |
+
"learning_rate": 7.905384077009693e-06,
|
| 197 |
+
"loss": 51.5311,
|
| 198 |
+
"step": 2700
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.7137394850879429,
|
| 202 |
+
"grad_norm": 443.4747619628906,
|
| 203 |
+
"learning_rate": 7.734740790612137e-06,
|
| 204 |
+
"loss": 50.3823,
|
| 205 |
+
"step": 2800
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.7392301809839409,
|
| 209 |
+
"grad_norm": 459.6910705566406,
|
| 210 |
+
"learning_rate": 7.559425245448006e-06,
|
| 211 |
+
"loss": 46.766,
|
| 212 |
+
"step": 2900
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.7647208768799388,
|
| 216 |
+
"grad_norm": 579.1948852539062,
|
| 217 |
+
"learning_rate": 7.379736965185369e-06,
|
| 218 |
+
"loss": 51.7392,
|
| 219 |
+
"step": 3000
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.7902115727759368,
|
| 223 |
+
"grad_norm": 329.1972961425781,
|
| 224 |
+
"learning_rate": 7.195982944236853e-06,
|
| 225 |
+
"loss": 51.0259,
|
| 226 |
+
"step": 3100
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.8157022686719347,
|
| 230 |
+
"grad_norm": 488.2768859863281,
|
| 231 |
+
"learning_rate": 7.008477123264849e-06,
|
| 232 |
+
"loss": 53.4051,
|
| 233 |
+
"step": 3200
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.8411929645679327,
|
| 237 |
+
"grad_norm": 463.906494140625,
|
| 238 |
+
"learning_rate": 6.817539852819149e-06,
|
| 239 |
+
"loss": 43.2941,
|
| 240 |
+
"step": 3300
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.8666836604639306,
|
| 244 |
+
"grad_norm": 263.90185546875,
|
| 245 |
+
"learning_rate": 6.6234973460234184e-06,
|
| 246 |
+
"loss": 50.0815,
|
| 247 |
+
"step": 3400
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.8921743563599286,
|
| 251 |
+
"grad_norm": 186.7496337890625,
|
| 252 |
+
"learning_rate": 6.426681121245527e-06,
|
| 253 |
+
"loss": 49.7881,
|
| 254 |
+
"step": 3500
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.9176650522559265,
|
| 258 |
+
"grad_norm": 398.3317565917969,
|
| 259 |
+
"learning_rate": 6.227427435703997e-06,
|
| 260 |
+
"loss": 50.899,
|
| 261 |
+
"step": 3600
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.9431557481519246,
|
| 265 |
+
"grad_norm": 238.25408935546875,
|
| 266 |
+
"learning_rate": 6.026076710978172e-06,
|
| 267 |
+
"loss": 44.1347,
|
| 268 |
+
"step": 3700
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.9686464440479226,
|
| 272 |
+
"grad_norm": 567.3057861328125,
|
| 273 |
+
"learning_rate": 5.82297295140367e-06,
|
| 274 |
+
"loss": 50.0442,
|
| 275 |
+
"step": 3800
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.9941371399439205,
|
| 279 |
+
"grad_norm": 226.54409790039062,
|
| 280 |
+
"learning_rate": 5.61846315634674e-06,
|
| 281 |
+
"loss": 41.8183,
|
| 282 |
+
"step": 3900
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 1.0,
|
| 286 |
+
"eval_loss": 2458.4501953125,
|
| 287 |
+
"eval_runtime": 28.4173,
|
| 288 |
+
"eval_samples_per_second": 122.707,
|
| 289 |
+
"eval_steps_per_second": 15.343,
|
| 290 |
+
"step": 3923
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 1.0196278358399185,
|
| 294 |
+
"grad_norm": 698.6807861328125,
|
| 295 |
+
"learning_rate": 5.412896727361663e-06,
|
| 296 |
+
"loss": 44.033,
|
| 297 |
+
"step": 4000
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 1.0451185317359164,
|
| 301 |
+
"grad_norm": 525.1095581054688,
|
| 302 |
+
"learning_rate": 5.206624871244066e-06,
|
| 303 |
+
"loss": 44.6573,
|
| 304 |
+
"step": 4100
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 1.0706092276319144,
|
| 308 |
+
"grad_norm": 500.8614501953125,
|
| 309 |
+
"learning_rate": 5e-06,
|
| 310 |
+
"loss": 42.8117,
|
| 311 |
+
"step": 4200
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 1.0960999235279123,
|
| 315 |
+
"grad_norm": 487.75689697265625,
|
| 316 |
+
"learning_rate": 4.793375128755934e-06,
|
| 317 |
+
"loss": 43.7467,
|
| 318 |
+
"step": 4300
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 1.1215906194239103,
|
| 322 |
+
"grad_norm": 385.0135192871094,
|
| 323 |
+
"learning_rate": 4.587103272638339e-06,
|
| 324 |
+
"loss": 43.3253,
|
| 325 |
+
"step": 4400
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 1.1470813153199082,
|
| 329 |
+
"grad_norm": 796.0384521484375,
|
| 330 |
+
"learning_rate": 4.381536843653262e-06,
|
| 331 |
+
"loss": 44.0489,
|
| 332 |
+
"step": 4500
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 1.1725720112159062,
|
| 336 |
+
"grad_norm": 428.2261962890625,
|
| 337 |
+
"learning_rate": 4.17702704859633e-06,
|
| 338 |
+
"loss": 48.0535,
|
| 339 |
+
"step": 4600
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 1.198062707111904,
|
| 343 |
+
"grad_norm": 385.2317199707031,
|
| 344 |
+
"learning_rate": 3.973923289021829e-06,
|
| 345 |
+
"loss": 43.4284,
|
| 346 |
+
"step": 4700
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 1.223553403007902,
|
| 350 |
+
"grad_norm": 159.48098754882812,
|
| 351 |
+
"learning_rate": 3.7725725642960047e-06,
|
| 352 |
+
"loss": 40.4862,
|
| 353 |
+
"step": 4800
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 1.2490440989039,
|
| 357 |
+
"grad_norm": 200.54164123535156,
|
| 358 |
+
"learning_rate": 3.573318878754475e-06,
|
| 359 |
+
"loss": 44.7052,
|
| 360 |
+
"step": 4900
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 1.274534794799898,
|
| 364 |
+
"grad_norm": 301.8349304199219,
|
| 365 |
+
"learning_rate": 3.3765026539765832e-06,
|
| 366 |
+
"loss": 41.2848,
|
| 367 |
+
"step": 5000
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 1.3000254906958961,
|
| 371 |
+
"grad_norm": 335.5228271484375,
|
| 372 |
+
"learning_rate": 3.1824601471808504e-06,
|
| 373 |
+
"loss": 43.5137,
|
| 374 |
+
"step": 5100
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 1.3255161865918939,
|
| 378 |
+
"grad_norm": 94.21492767333984,
|
| 379 |
+
"learning_rate": 2.991522876735154e-06,
|
| 380 |
+
"loss": 41.9114,
|
| 381 |
+
"step": 5200
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 1.351006882487892,
|
| 385 |
+
"grad_norm": 352.31048583984375,
|
| 386 |
+
"learning_rate": 2.804017055763149e-06,
|
| 387 |
+
"loss": 41.0477,
|
| 388 |
+
"step": 5300
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 1.3764975783838898,
|
| 392 |
+
"grad_norm": 228.1314697265625,
|
| 393 |
+
"learning_rate": 2.6202630348146323e-06,
|
| 394 |
+
"loss": 38.6621,
|
| 395 |
+
"step": 5400
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 1.401988274279888,
|
| 399 |
+
"grad_norm": 170.14947509765625,
|
| 400 |
+
"learning_rate": 2.4405747545519966e-06,
|
| 401 |
+
"loss": 40.0131,
|
| 402 |
+
"step": 5500
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 1.4274789701758859,
|
| 406 |
+
"grad_norm": 166.94281005859375,
|
| 407 |
+
"learning_rate": 2.265259209387867e-06,
|
| 408 |
+
"loss": 42.6173,
|
| 409 |
+
"step": 5600
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 1.4529696660718838,
|
| 413 |
+
"grad_norm": 255.24964904785156,
|
| 414 |
+
"learning_rate": 2.094615922990309e-06,
|
| 415 |
+
"loss": 41.4291,
|
| 416 |
+
"step": 5700
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 1.4784603619678818,
|
| 420 |
+
"grad_norm": 296.5071716308594,
|
| 421 |
+
"learning_rate": 1.928936436551661e-06,
|
| 422 |
+
"loss": 39.6349,
|
| 423 |
+
"step": 5800
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 1.5039510578638797,
|
| 427 |
+
"grad_norm": 227.43650817871094,
|
| 428 |
+
"learning_rate": 1.7685038106952952e-06,
|
| 429 |
+
"loss": 40.9158,
|
| 430 |
+
"step": 5900
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 1.5294417537598777,
|
| 434 |
+
"grad_norm": 280.7147216796875,
|
| 435 |
+
"learning_rate": 1.6135921418712959e-06,
|
| 436 |
+
"loss": 45.4827,
|
| 437 |
+
"step": 6000
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 1.5549324496558756,
|
| 441 |
+
"grad_norm": 166.27865600585938,
|
| 442 |
+
"learning_rate": 1.4644660940672628e-06,
|
| 443 |
+
"loss": 40.6627,
|
| 444 |
+
"step": 6100
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 1.5804231455518736,
|
| 448 |
+
"grad_norm": 389.299560546875,
|
| 449 |
+
"learning_rate": 1.321380446634342e-06,
|
| 450 |
+
"loss": 38.7581,
|
| 451 |
+
"step": 6200
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 1.6059138414478715,
|
| 455 |
+
"grad_norm": 170.0509033203125,
|
| 456 |
+
"learning_rate": 1.1845796590009684e-06,
|
| 457 |
+
"loss": 40.3946,
|
| 458 |
+
"step": 6300
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 1.6314045373438695,
|
| 462 |
+
"grad_norm": 287.59423828125,
|
| 463 |
+
"learning_rate": 1.0542974530180327e-06,
|
| 464 |
+
"loss": 43.811,
|
| 465 |
+
"step": 6400
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 1.6568952332398674,
|
| 469 |
+
"grad_norm": 150.49293518066406,
|
| 470 |
+
"learning_rate": 9.307564136490255e-07,
|
| 471 |
+
"loss": 40.4753,
|
| 472 |
+
"step": 6500
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 1.6823859291358654,
|
| 476 |
+
"grad_norm": 264.41253662109375,
|
| 477 |
+
"learning_rate": 8.141676086873574e-07,
|
| 478 |
+
"loss": 43.6258,
|
| 479 |
+
"step": 6600
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 1.7078766250318633,
|
| 483 |
+
"grad_norm": 157.13479614257812,
|
| 484 |
+
"learning_rate": 7.047302281505735e-07,
|
| 485 |
+
"loss": 43.768,
|
| 486 |
+
"step": 6700
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 1.7333673209278613,
|
| 490 |
+
"grad_norm": 206.51609802246094,
|
| 491 |
+
"learning_rate": 6.026312439675553e-07,
|
| 492 |
+
"loss": 39.8553,
|
| 493 |
+
"step": 6800
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 1.7588580168238592,
|
| 497 |
+
"grad_norm": 415.54351806640625,
|
| 498 |
+
"learning_rate": 5.080450905401057e-07,
|
| 499 |
+
"loss": 37.9228,
|
| 500 |
+
"step": 6900
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 1.7843487127198574,
|
| 504 |
+
"grad_norm": 189.60276794433594,
|
| 505 |
+
"learning_rate": 4.211333667247125e-07,
|
| 506 |
+
"loss": 41.446,
|
| 507 |
+
"step": 7000
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 1.8098394086158551,
|
| 511 |
+
"grad_norm": 612.336181640625,
|
| 512 |
+
"learning_rate": 3.420445597436056e-07,
|
| 513 |
+
"loss": 39.8801,
|
| 514 |
+
"step": 7100
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 1.8353301045118533,
|
| 518 |
+
"grad_norm": 338.9154052734375,
|
| 519 |
+
"learning_rate": 2.7091379149682683e-07,
|
| 520 |
+
"loss": 41.9962,
|
| 521 |
+
"step": 7200
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 1.860820800407851,
|
| 525 |
+
"grad_norm": 207.06942749023438,
|
| 526 |
+
"learning_rate": 2.0786258770873647e-07,
|
| 527 |
+
"loss": 40.2771,
|
| 528 |
+
"step": 7300
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 1.8863114963038492,
|
| 532 |
+
"grad_norm": 585.3238525390625,
|
| 533 |
+
"learning_rate": 1.5299867030334815e-07,
|
| 534 |
+
"loss": 37.347,
|
| 535 |
+
"step": 7400
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 1.911802192199847,
|
| 539 |
+
"grad_norm": 224.99607849121094,
|
| 540 |
+
"learning_rate": 1.0641577336322761e-07,
|
| 541 |
+
"loss": 45.2395,
|
| 542 |
+
"step": 7500
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 1.937292888095845,
|
| 546 |
+
"grad_norm": 304.6292724609375,
|
| 547 |
+
"learning_rate": 6.819348298638839e-08,
|
| 548 |
+
"loss": 51.0356,
|
| 549 |
+
"step": 7600
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 1.9627835839918428,
|
| 553 |
+
"grad_norm": 1007.5321044921875,
|
| 554 |
+
"learning_rate": 3.839710131477492e-08,
|
| 555 |
+
"loss": 44.2402,
|
| 556 |
+
"step": 7700
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 1.988274279887841,
|
| 560 |
+
"grad_norm": 436.2712097167969,
|
| 561 |
+
"learning_rate": 1.7077534966650767e-08,
|
| 562 |
+
"loss": 41.5961,
|
| 563 |
+
"step": 7800
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 2.0,
|
| 567 |
+
"eval_loss": 2133.657470703125,
|
| 568 |
+
"eval_runtime": 28.5718,
|
| 569 |
+
"eval_samples_per_second": 122.043,
|
| 570 |
+
"eval_steps_per_second": 15.26,
|
| 571 |
+
"step": 7846
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 2.0137649757838387,
|
| 575 |
+
"grad_norm": 354.5096740722656,
|
| 576 |
+
"learning_rate": 4.2712080634949024e-09,
|
| 577 |
+
"loss": 41.8052,
|
| 578 |
+
"step": 7900
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 2.039255671679837,
|
| 582 |
+
"grad_norm": 220.47647094726562,
|
| 583 |
+
"learning_rate": 0.0,
|
| 584 |
+
"loss": 39.9241,
|
| 585 |
+
"step": 8000
|
| 586 |
+
}
|
| 587 |
+
],
|
| 588 |
+
"logging_steps": 100,
|
| 589 |
+
"max_steps": 8000,
|
| 590 |
+
"num_input_tokens_seen": 0,
|
| 591 |
+
"num_train_epochs": 3,
|
| 592 |
+
"save_steps": 2000,
|
| 593 |
+
"total_flos": 0.0,
|
| 594 |
+
"train_batch_size": 8,
|
| 595 |
+
"trial_name": null,
|
| 596 |
+
"trial_params": null
|
| 597 |
+
}
|