Upload folder using huggingface_hub
Browse files- config.json +64 -0
- experiment_cfg/metadata.json +307 -0
- model-00001-of-00002.safetensors +3 -0
- model-00002-of-00002.safetensors +3 -0
- model.safetensors.index.json +0 -0
- trainer_state.json +3534 -0
config.json
ADDED
|
@@ -0,0 +1,64 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"action_dim": 32,
|
| 3 |
+
"action_head_cfg": {
|
| 4 |
+
"action_dim": 32,
|
| 5 |
+
"action_horizon": 16,
|
| 6 |
+
"add_pos_embed": true,
|
| 7 |
+
"backbone_embedding_dim": 2048,
|
| 8 |
+
"diffusion_model_cfg": {
|
| 9 |
+
"attention_head_dim": 48,
|
| 10 |
+
"cross_attention_dim": 2048,
|
| 11 |
+
"dropout": 0.2,
|
| 12 |
+
"final_dropout": true,
|
| 13 |
+
"interleave_self_attention": true,
|
| 14 |
+
"norm_type": "ada_norm",
|
| 15 |
+
"num_attention_heads": 32,
|
| 16 |
+
"num_layers": 16,
|
| 17 |
+
"output_dim": 1024,
|
| 18 |
+
"positional_embeddings": null
|
| 19 |
+
},
|
| 20 |
+
"hidden_size": 1024,
|
| 21 |
+
"input_embedding_dim": 1536,
|
| 22 |
+
"max_action_dim": 32,
|
| 23 |
+
"max_state_dim": 64,
|
| 24 |
+
"model_dtype": "float32",
|
| 25 |
+
"noise_beta_alpha": 1.5,
|
| 26 |
+
"noise_beta_beta": 1.0,
|
| 27 |
+
"noise_s": 0.999,
|
| 28 |
+
"num_inference_timesteps": 4,
|
| 29 |
+
"num_target_vision_tokens": 32,
|
| 30 |
+
"num_timestep_buckets": 1000,
|
| 31 |
+
"tune_diffusion_model": true,
|
| 32 |
+
"tune_projector": true,
|
| 33 |
+
"use_vlln": true,
|
| 34 |
+
"vl_self_attention_cfg": {
|
| 35 |
+
"attention_head_dim": 64,
|
| 36 |
+
"dropout": 0.2,
|
| 37 |
+
"final_dropout": true,
|
| 38 |
+
"num_attention_heads": 32,
|
| 39 |
+
"num_layers": 4,
|
| 40 |
+
"positional_embeddings": null
|
| 41 |
+
}
|
| 42 |
+
},
|
| 43 |
+
"action_horizon": 16,
|
| 44 |
+
"architectures": [
|
| 45 |
+
"GR00T_N1_5"
|
| 46 |
+
],
|
| 47 |
+
"attn_implementation": null,
|
| 48 |
+
"backbone_cfg": {
|
| 49 |
+
"eagle_path": "NVEagle/eagle_er-qwen3_1_7B-Siglip2_400M_stage1_5_128gpu_er_v7_1mlp_nops",
|
| 50 |
+
"load_bf16": false,
|
| 51 |
+
"project_to_dim": null,
|
| 52 |
+
"reproject_vision": false,
|
| 53 |
+
"select_layer": 12,
|
| 54 |
+
"tune_llm": false,
|
| 55 |
+
"tune_visual": true,
|
| 56 |
+
"use_flash_attention": true
|
| 57 |
+
},
|
| 58 |
+
"compute_dtype": "bfloat16",
|
| 59 |
+
"hidden_size": 2048,
|
| 60 |
+
"model_dtype": "float32",
|
| 61 |
+
"model_type": "gr00t_n1_5",
|
| 62 |
+
"torch_dtype": "bfloat16",
|
| 63 |
+
"transformers_version": "4.51.3"
|
| 64 |
+
}
|
experiment_cfg/metadata.json
ADDED
|
@@ -0,0 +1,307 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"new_embodiment": {
|
| 3 |
+
"statistics": {
|
| 4 |
+
"state": {
|
| 5 |
+
"left_arm": {
|
| 6 |
+
"max": [
|
| 7 |
+
0.4827006161212921,
|
| 8 |
+
0.8112002015113831,
|
| 9 |
+
0.4532553553581238,
|
| 10 |
+
-0.8231125473976135,
|
| 11 |
+
1.1653939485549927,
|
| 12 |
+
0.7899761199951172,
|
| 13 |
+
0.19781433045864105,
|
| 14 |
+
1.0499905347824097
|
| 15 |
+
],
|
| 16 |
+
"min": [
|
| 17 |
+
-1.0513520240783691,
|
| 18 |
+
-0.000107858024421148,
|
| 19 |
+
-1.1072344779968262,
|
| 20 |
+
-2.296549081802368,
|
| 21 |
+
-0.07028748095035553,
|
| 22 |
+
-0.5859926342964172,
|
| 23 |
+
-1.47036874294281,
|
| 24 |
+
0.0
|
| 25 |
+
],
|
| 26 |
+
"mean": [
|
| 27 |
+
0.011497579514980316,
|
| 28 |
+
0.19215743243694305,
|
| 29 |
+
-0.1451243907213211,
|
| 30 |
+
-1.4894176721572876,
|
| 31 |
+
0.3010796308517456,
|
| 32 |
+
-0.09618806838989258,
|
| 33 |
+
-0.3399161398410797,
|
| 34 |
+
0.2933500409126282
|
| 35 |
+
],
|
| 36 |
+
"std": [
|
| 37 |
+
0.3714962303638458,
|
| 38 |
+
0.19363699853420258,
|
| 39 |
+
0.2132025510072708,
|
| 40 |
+
0.216729074716568,
|
| 41 |
+
0.26094645261764526,
|
| 42 |
+
0.3879685401916504,
|
| 43 |
+
0.44094032049179077,
|
| 44 |
+
0.32030728459358215
|
| 45 |
+
],
|
| 46 |
+
"q01": [
|
| 47 |
+
-0.9140344530344009,
|
| 48 |
+
0.0,
|
| 49 |
+
-0.7611510992050171,
|
| 50 |
+
-2.0928124308586122,
|
| 51 |
+
-0.008116316050291061,
|
| 52 |
+
-0.5690907031297684,
|
| 53 |
+
-1.2905553519725799,
|
| 54 |
+
0.029702702537178993
|
| 55 |
+
],
|
| 56 |
+
"q99": [
|
| 57 |
+
0.3913862168788912,
|
| 58 |
+
0.7450544595718385,
|
| 59 |
+
0.3400661617517477,
|
| 60 |
+
-1.013698261976242,
|
| 61 |
+
0.9407364845275887,
|
| 62 |
+
0.6568368136882784,
|
| 63 |
+
0.18469616919755938,
|
| 64 |
+
1.042564868927002
|
| 65 |
+
]
|
| 66 |
+
},
|
| 67 |
+
"right_arm": {
|
| 68 |
+
"max": [
|
| 69 |
+
0.6417672038078308,
|
| 70 |
+
-0.0014980281703174114,
|
| 71 |
+
0.792744517326355,
|
| 72 |
+
-1.0384330749511719,
|
| 73 |
+
0.6077679991722107,
|
| 74 |
+
0.9873323440551758,
|
| 75 |
+
1.360622525215149,
|
| 76 |
+
1.0603865385055542
|
| 77 |
+
],
|
| 78 |
+
"min": [
|
| 79 |
+
-0.9047250747680664,
|
| 80 |
+
-0.8373377919197083,
|
| 81 |
+
-0.6043045520782471,
|
| 82 |
+
-2.3454806804656982,
|
| 83 |
+
-0.7402056455612183,
|
| 84 |
+
-0.7786630392074585,
|
| 85 |
+
-0.33405861258506775,
|
| 86 |
+
0.0
|
| 87 |
+
],
|
| 88 |
+
"mean": [
|
| 89 |
+
0.09296321868896484,
|
| 90 |
+
-0.21980760991573334,
|
| 91 |
+
0.15327434241771698,
|
| 92 |
+
-1.646316647529602,
|
| 93 |
+
-0.2041722536087036,
|
| 94 |
+
-0.03187763690948486,
|
| 95 |
+
0.24834179878234863,
|
| 96 |
+
0.325268417596817
|
| 97 |
+
],
|
| 98 |
+
"std": [
|
| 99 |
+
0.3301715552806854,
|
| 100 |
+
0.172244131565094,
|
| 101 |
+
0.2095300257205963,
|
| 102 |
+
0.24973280727863312,
|
| 103 |
+
0.1718880832195282,
|
| 104 |
+
0.4755953550338745,
|
| 105 |
+
0.4145625829696655,
|
| 106 |
+
0.33061647415161133
|
| 107 |
+
],
|
| 108 |
+
"q01": [
|
| 109 |
+
-0.7729669392108918,
|
| 110 |
+
-0.7347564280033112,
|
| 111 |
+
-0.4285223290324211,
|
| 112 |
+
-2.2316149473190308,
|
| 113 |
+
-0.6489248216152191,
|
| 114 |
+
-0.731025755405426,
|
| 115 |
+
-0.25769850611686707,
|
| 116 |
+
0.06831621378660202
|
| 117 |
+
],
|
| 118 |
+
"q99": [
|
| 119 |
+
0.5628942608833314,
|
| 120 |
+
-0.052740177512168744,
|
| 121 |
+
0.5677921921014794,
|
| 122 |
+
-1.2287228107452393,
|
| 123 |
+
0.12717000618577062,
|
| 124 |
+
0.8326435863971711,
|
| 125 |
+
1.1078305482864381,
|
| 126 |
+
1.0574162006378174
|
| 127 |
+
]
|
| 128 |
+
}
|
| 129 |
+
},
|
| 130 |
+
"action": {
|
| 131 |
+
"left_arm": {
|
| 132 |
+
"max": [
|
| 133 |
+
0.4832039475440979,
|
| 134 |
+
0.8114758133888245,
|
| 135 |
+
0.45405831933021545,
|
| 136 |
+
-0.8222137093544006,
|
| 137 |
+
1.165825366973877,
|
| 138 |
+
0.7930680513381958,
|
| 139 |
+
0.1978835165500641,
|
| 140 |
+
1.1789709329605103
|
| 141 |
+
],
|
| 142 |
+
"min": [
|
| 143 |
+
-1.052310824394226,
|
| 144 |
+
-0.08130098134279251,
|
| 145 |
+
-1.1090681552886963,
|
| 146 |
+
-2.297903299331665,
|
| 147 |
+
-0.07056311517953873,
|
| 148 |
+
-0.5859806537628174,
|
| 149 |
+
-1.4710875749588013,
|
| 150 |
+
-0.11570879071950912
|
| 151 |
+
],
|
| 152 |
+
"mean": [
|
| 153 |
+
0.011788768693804741,
|
| 154 |
+
0.1899206042289734,
|
| 155 |
+
-0.14491689205169678,
|
| 156 |
+
-1.489245057106018,
|
| 157 |
+
0.3011428117752075,
|
| 158 |
+
-0.09616442769765854,
|
| 159 |
+
-0.3399713933467865,
|
| 160 |
+
0.29545462131500244
|
| 161 |
+
],
|
| 162 |
+
"std": [
|
| 163 |
+
0.371960312128067,
|
| 164 |
+
0.19602903723716736,
|
| 165 |
+
0.21364881098270416,
|
| 166 |
+
0.21710477769374847,
|
| 167 |
+
0.26112067699432373,
|
| 168 |
+
0.38814306259155273,
|
| 169 |
+
0.4409879148006439,
|
| 170 |
+
0.3344240188598633
|
| 171 |
+
],
|
| 172 |
+
"q01": [
|
| 173 |
+
-0.9142525792121887,
|
| 174 |
+
-0.032213594764471054,
|
| 175 |
+
-0.7639224529266357,
|
| 176 |
+
-2.093883752822876,
|
| 177 |
+
-0.00920388475060463,
|
| 178 |
+
-0.5691068768501282,
|
| 179 |
+
-1.2916117906570435,
|
| 180 |
+
0.01007763110101223
|
| 181 |
+
],
|
| 182 |
+
"q99": [
|
| 183 |
+
0.3957670331001282,
|
| 184 |
+
0.7455146908760071,
|
| 185 |
+
0.34207770228385925,
|
| 186 |
+
-1.0124273300170898,
|
| 187 |
+
0.9418641924858093,
|
| 188 |
+
0.6565437912940979,
|
| 189 |
+
0.18561168015003204,
|
| 190 |
+
1.124131202697765
|
| 191 |
+
]
|
| 192 |
+
},
|
| 193 |
+
"right_arm": {
|
| 194 |
+
"max": [
|
| 195 |
+
0.6442719101905823,
|
| 196 |
+
-2.0694557179012918e-13,
|
| 197 |
+
0.7930680513381958,
|
| 198 |
+
-1.0369709730148315,
|
| 199 |
+
0.6089903712272644,
|
| 200 |
+
0.9894176125526428,
|
| 201 |
+
1.3959225416183472,
|
| 202 |
+
1.155961275100708
|
| 203 |
+
],
|
| 204 |
+
"min": [
|
| 205 |
+
-0.905048668384552,
|
| 206 |
+
-0.8375535011291504,
|
| 207 |
+
-0.6059224009513855,
|
| 208 |
+
-2.3469905853271484,
|
| 209 |
+
-0.7409127354621887,
|
| 210 |
+
-0.7792622447013855,
|
| 211 |
+
-0.3344078063964844,
|
| 212 |
+
-0.0788932517170906
|
| 213 |
+
],
|
| 214 |
+
"mean": [
|
| 215 |
+
0.09317154437303543,
|
| 216 |
+
-0.21982984244823456,
|
| 217 |
+
0.15331701934337616,
|
| 218 |
+
-1.6462136507034302,
|
| 219 |
+
-0.20444351434707642,
|
| 220 |
+
-0.03180892765522003,
|
| 221 |
+
0.2483728677034378,
|
| 222 |
+
0.326642245054245
|
| 223 |
+
],
|
| 224 |
+
"std": [
|
| 225 |
+
0.33065080642700195,
|
| 226 |
+
0.17234687507152557,
|
| 227 |
+
0.20994938910007477,
|
| 228 |
+
0.2501552402973175,
|
| 229 |
+
0.1719447821378708,
|
| 230 |
+
0.4758480191230774,
|
| 231 |
+
0.4146299362182617,
|
| 232 |
+
0.34059736132621765
|
| 233 |
+
],
|
| 234 |
+
"q01": [
|
| 235 |
+
-0.7731263041496277,
|
| 236 |
+
-0.7342399001121521,
|
| 237 |
+
-0.429514616727829,
|
| 238 |
+
-2.2314050436019897,
|
| 239 |
+
-0.6519418358802795,
|
| 240 |
+
-0.731708824634552,
|
| 241 |
+
-0.25770875811576843,
|
| 242 |
+
0.05149511247873306
|
| 243 |
+
],
|
| 244 |
+
"q99": [
|
| 245 |
+
0.5675728917121887,
|
| 246 |
+
-0.05215534567832947,
|
| 247 |
+
0.5660389065742493,
|
| 248 |
+
-1.228718638420105,
|
| 249 |
+
0.1298514693975471,
|
| 250 |
+
0.832951545715332,
|
| 251 |
+
1.1085312604904196,
|
| 252 |
+
1.1222137212753296
|
| 253 |
+
]
|
| 254 |
+
}
|
| 255 |
+
}
|
| 256 |
+
},
|
| 257 |
+
"modalities": {
|
| 258 |
+
"video": {
|
| 259 |
+
"cam_head": {
|
| 260 |
+
"resolution": [
|
| 261 |
+
224,
|
| 262 |
+
224
|
| 263 |
+
],
|
| 264 |
+
"channels": 3,
|
| 265 |
+
"fps": 10.0
|
| 266 |
+
}
|
| 267 |
+
},
|
| 268 |
+
"state": {
|
| 269 |
+
"left_arm": {
|
| 270 |
+
"absolute": true,
|
| 271 |
+
"rotation_type": null,
|
| 272 |
+
"shape": [
|
| 273 |
+
8
|
| 274 |
+
],
|
| 275 |
+
"continuous": true
|
| 276 |
+
},
|
| 277 |
+
"right_arm": {
|
| 278 |
+
"absolute": true,
|
| 279 |
+
"rotation_type": null,
|
| 280 |
+
"shape": [
|
| 281 |
+
8
|
| 282 |
+
],
|
| 283 |
+
"continuous": true
|
| 284 |
+
}
|
| 285 |
+
},
|
| 286 |
+
"action": {
|
| 287 |
+
"left_arm": {
|
| 288 |
+
"absolute": true,
|
| 289 |
+
"rotation_type": null,
|
| 290 |
+
"shape": [
|
| 291 |
+
8
|
| 292 |
+
],
|
| 293 |
+
"continuous": true
|
| 294 |
+
},
|
| 295 |
+
"right_arm": {
|
| 296 |
+
"absolute": true,
|
| 297 |
+
"rotation_type": null,
|
| 298 |
+
"shape": [
|
| 299 |
+
8
|
| 300 |
+
],
|
| 301 |
+
"continuous": true
|
| 302 |
+
}
|
| 303 |
+
}
|
| 304 |
+
},
|
| 305 |
+
"embodiment_tag": "new_embodiment"
|
| 306 |
+
}
|
| 307 |
+
}
|
model-00001-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ef1ece6c107df1c55502d127cd18c573d7ce47816010553954bf7b0393ea9cff
|
| 3 |
+
size 4999367032
|
model-00002-of-00002.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5367d19cf066726459e023ef6224839ec857c73213d090291674d152a3e7a823
|
| 3 |
+
size 2586705312
|
model.safetensors.index.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
trainer_state.json
ADDED
|
@@ -0,0 +1,3534 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 104.16666666666667,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 5000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.20833333333333334,
|
| 14 |
+
"grad_norm": 6.974559783935547,
|
| 15 |
+
"learning_rate": 9e-07,
|
| 16 |
+
"loss": 0.9968,
|
| 17 |
+
"step": 10
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.4166666666666667,
|
| 21 |
+
"grad_norm": 5.023812294006348,
|
| 22 |
+
"learning_rate": 1.9e-06,
|
| 23 |
+
"loss": 0.9334,
|
| 24 |
+
"step": 20
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.625,
|
| 28 |
+
"grad_norm": 2.8353774547576904,
|
| 29 |
+
"learning_rate": 2.9e-06,
|
| 30 |
+
"loss": 0.6922,
|
| 31 |
+
"step": 30
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.8333333333333334,
|
| 35 |
+
"grad_norm": 0.9118760228157043,
|
| 36 |
+
"learning_rate": 3.9e-06,
|
| 37 |
+
"loss": 0.4461,
|
| 38 |
+
"step": 40
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 1.0416666666666667,
|
| 42 |
+
"grad_norm": 0.5101693272590637,
|
| 43 |
+
"learning_rate": 4.9000000000000005e-06,
|
| 44 |
+
"loss": 0.2961,
|
| 45 |
+
"step": 50
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 1.25,
|
| 49 |
+
"grad_norm": 0.41475388407707214,
|
| 50 |
+
"learning_rate": 5.9e-06,
|
| 51 |
+
"loss": 0.2194,
|
| 52 |
+
"step": 60
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 1.4583333333333333,
|
| 56 |
+
"grad_norm": 0.3171979486942291,
|
| 57 |
+
"learning_rate": 6.900000000000001e-06,
|
| 58 |
+
"loss": 0.1815,
|
| 59 |
+
"step": 70
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 1.6666666666666665,
|
| 63 |
+
"grad_norm": 0.38558876514434814,
|
| 64 |
+
"learning_rate": 7.9e-06,
|
| 65 |
+
"loss": 0.1614,
|
| 66 |
+
"step": 80
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 1.875,
|
| 70 |
+
"grad_norm": 0.28808021545410156,
|
| 71 |
+
"learning_rate": 8.9e-06,
|
| 72 |
+
"loss": 0.1422,
|
| 73 |
+
"step": 90
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 2.0833333333333335,
|
| 77 |
+
"grad_norm": 0.29702240228652954,
|
| 78 |
+
"learning_rate": 9.900000000000002e-06,
|
| 79 |
+
"loss": 0.1262,
|
| 80 |
+
"step": 100
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 2.2916666666666665,
|
| 84 |
+
"grad_norm": 0.2959185540676117,
|
| 85 |
+
"learning_rate": 1.09e-05,
|
| 86 |
+
"loss": 0.1081,
|
| 87 |
+
"step": 110
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 2.5,
|
| 91 |
+
"grad_norm": 0.2338050752878189,
|
| 92 |
+
"learning_rate": 1.19e-05,
|
| 93 |
+
"loss": 0.0981,
|
| 94 |
+
"step": 120
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 2.7083333333333335,
|
| 98 |
+
"grad_norm": 0.2578125,
|
| 99 |
+
"learning_rate": 1.29e-05,
|
| 100 |
+
"loss": 0.0895,
|
| 101 |
+
"step": 130
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 2.9166666666666665,
|
| 105 |
+
"grad_norm": 0.2224149852991104,
|
| 106 |
+
"learning_rate": 1.3900000000000002e-05,
|
| 107 |
+
"loss": 0.0776,
|
| 108 |
+
"step": 140
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 3.125,
|
| 112 |
+
"grad_norm": 0.19243746995925903,
|
| 113 |
+
"learning_rate": 1.49e-05,
|
| 114 |
+
"loss": 0.0712,
|
| 115 |
+
"step": 150
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 3.3333333333333335,
|
| 119 |
+
"grad_norm": 0.2733956575393677,
|
| 120 |
+
"learning_rate": 1.59e-05,
|
| 121 |
+
"loss": 0.0674,
|
| 122 |
+
"step": 160
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 3.5416666666666665,
|
| 126 |
+
"grad_norm": 0.23844897747039795,
|
| 127 |
+
"learning_rate": 1.69e-05,
|
| 128 |
+
"loss": 0.0617,
|
| 129 |
+
"step": 170
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 3.75,
|
| 133 |
+
"grad_norm": 0.20417346060276031,
|
| 134 |
+
"learning_rate": 1.79e-05,
|
| 135 |
+
"loss": 0.056,
|
| 136 |
+
"step": 180
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 3.9583333333333335,
|
| 140 |
+
"grad_norm": 0.24881872534751892,
|
| 141 |
+
"learning_rate": 1.8900000000000002e-05,
|
| 142 |
+
"loss": 0.0525,
|
| 143 |
+
"step": 190
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 4.166666666666667,
|
| 147 |
+
"grad_norm": 0.25316762924194336,
|
| 148 |
+
"learning_rate": 1.9900000000000003e-05,
|
| 149 |
+
"loss": 0.0512,
|
| 150 |
+
"step": 200
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 4.375,
|
| 154 |
+
"grad_norm": 0.19489045441150665,
|
| 155 |
+
"learning_rate": 2.09e-05,
|
| 156 |
+
"loss": 0.0454,
|
| 157 |
+
"step": 210
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 4.583333333333333,
|
| 161 |
+
"grad_norm": 0.1898735761642456,
|
| 162 |
+
"learning_rate": 2.19e-05,
|
| 163 |
+
"loss": 0.0446,
|
| 164 |
+
"step": 220
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 4.791666666666667,
|
| 168 |
+
"grad_norm": 0.20489181578159332,
|
| 169 |
+
"learning_rate": 2.29e-05,
|
| 170 |
+
"loss": 0.0422,
|
| 171 |
+
"step": 230
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 5.0,
|
| 175 |
+
"grad_norm": 0.3320506811141968,
|
| 176 |
+
"learning_rate": 2.39e-05,
|
| 177 |
+
"loss": 0.0407,
|
| 178 |
+
"step": 240
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 5.208333333333333,
|
| 182 |
+
"grad_norm": 0.24517174065113068,
|
| 183 |
+
"learning_rate": 2.4900000000000002e-05,
|
| 184 |
+
"loss": 0.0374,
|
| 185 |
+
"step": 250
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 5.416666666666667,
|
| 189 |
+
"grad_norm": 0.26409515738487244,
|
| 190 |
+
"learning_rate": 2.5900000000000003e-05,
|
| 191 |
+
"loss": 0.0364,
|
| 192 |
+
"step": 260
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 5.625,
|
| 196 |
+
"grad_norm": 0.2976372241973877,
|
| 197 |
+
"learning_rate": 2.6900000000000003e-05,
|
| 198 |
+
"loss": 0.0353,
|
| 199 |
+
"step": 270
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 5.833333333333333,
|
| 203 |
+
"grad_norm": 0.24544252455234528,
|
| 204 |
+
"learning_rate": 2.7900000000000004e-05,
|
| 205 |
+
"loss": 0.0332,
|
| 206 |
+
"step": 280
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 6.041666666666667,
|
| 210 |
+
"grad_norm": 0.3133062720298767,
|
| 211 |
+
"learning_rate": 2.8899999999999998e-05,
|
| 212 |
+
"loss": 0.0332,
|
| 213 |
+
"step": 290
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 6.25,
|
| 217 |
+
"grad_norm": 0.29001384973526,
|
| 218 |
+
"learning_rate": 2.9900000000000002e-05,
|
| 219 |
+
"loss": 0.0326,
|
| 220 |
+
"step": 300
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 6.458333333333333,
|
| 224 |
+
"grad_norm": 0.2940799295902252,
|
| 225 |
+
"learning_rate": 3.09e-05,
|
| 226 |
+
"loss": 0.0321,
|
| 227 |
+
"step": 310
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 6.666666666666667,
|
| 231 |
+
"grad_norm": 0.2921755909919739,
|
| 232 |
+
"learning_rate": 3.19e-05,
|
| 233 |
+
"loss": 0.0306,
|
| 234 |
+
"step": 320
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 6.875,
|
| 238 |
+
"grad_norm": 0.3143065869808197,
|
| 239 |
+
"learning_rate": 3.29e-05,
|
| 240 |
+
"loss": 0.0298,
|
| 241 |
+
"step": 330
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 7.083333333333333,
|
| 245 |
+
"grad_norm": 0.35790130496025085,
|
| 246 |
+
"learning_rate": 3.3900000000000004e-05,
|
| 247 |
+
"loss": 0.0295,
|
| 248 |
+
"step": 340
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 7.291666666666667,
|
| 252 |
+
"grad_norm": 0.24478909373283386,
|
| 253 |
+
"learning_rate": 3.49e-05,
|
| 254 |
+
"loss": 0.028,
|
| 255 |
+
"step": 350
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 7.5,
|
| 259 |
+
"grad_norm": 0.19725550711154938,
|
| 260 |
+
"learning_rate": 3.59e-05,
|
| 261 |
+
"loss": 0.0273,
|
| 262 |
+
"step": 360
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 7.708333333333333,
|
| 266 |
+
"grad_norm": 0.19303350150585175,
|
| 267 |
+
"learning_rate": 3.69e-05,
|
| 268 |
+
"loss": 0.0279,
|
| 269 |
+
"step": 370
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 7.916666666666667,
|
| 273 |
+
"grad_norm": 0.22829170525074005,
|
| 274 |
+
"learning_rate": 3.79e-05,
|
| 275 |
+
"loss": 0.0275,
|
| 276 |
+
"step": 380
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 8.125,
|
| 280 |
+
"grad_norm": 0.312612384557724,
|
| 281 |
+
"learning_rate": 3.8900000000000004e-05,
|
| 282 |
+
"loss": 0.0252,
|
| 283 |
+
"step": 390
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 8.333333333333334,
|
| 287 |
+
"grad_norm": 0.26859182119369507,
|
| 288 |
+
"learning_rate": 3.99e-05,
|
| 289 |
+
"loss": 0.0262,
|
| 290 |
+
"step": 400
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 8.541666666666666,
|
| 294 |
+
"grad_norm": 0.24778631329536438,
|
| 295 |
+
"learning_rate": 4.09e-05,
|
| 296 |
+
"loss": 0.024,
|
| 297 |
+
"step": 410
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 8.75,
|
| 301 |
+
"grad_norm": 0.2118053287267685,
|
| 302 |
+
"learning_rate": 4.19e-05,
|
| 303 |
+
"loss": 0.0254,
|
| 304 |
+
"step": 420
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 8.958333333333334,
|
| 308 |
+
"grad_norm": 0.2575972080230713,
|
| 309 |
+
"learning_rate": 4.29e-05,
|
| 310 |
+
"loss": 0.024,
|
| 311 |
+
"step": 430
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 9.166666666666666,
|
| 315 |
+
"grad_norm": 0.2238076776266098,
|
| 316 |
+
"learning_rate": 4.39e-05,
|
| 317 |
+
"loss": 0.0247,
|
| 318 |
+
"step": 440
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 9.375,
|
| 322 |
+
"grad_norm": 0.32065877318382263,
|
| 323 |
+
"learning_rate": 4.49e-05,
|
| 324 |
+
"loss": 0.0247,
|
| 325 |
+
"step": 450
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 9.583333333333334,
|
| 329 |
+
"grad_norm": 0.28235775232315063,
|
| 330 |
+
"learning_rate": 4.5900000000000004e-05,
|
| 331 |
+
"loss": 0.0242,
|
| 332 |
+
"step": 460
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 9.791666666666666,
|
| 336 |
+
"grad_norm": 0.3509873151779175,
|
| 337 |
+
"learning_rate": 4.69e-05,
|
| 338 |
+
"loss": 0.0235,
|
| 339 |
+
"step": 470
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 10.0,
|
| 343 |
+
"grad_norm": 0.26749303936958313,
|
| 344 |
+
"learning_rate": 4.79e-05,
|
| 345 |
+
"loss": 0.0234,
|
| 346 |
+
"step": 480
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 10.208333333333334,
|
| 350 |
+
"grad_norm": 0.2362636923789978,
|
| 351 |
+
"learning_rate": 4.89e-05,
|
| 352 |
+
"loss": 0.0227,
|
| 353 |
+
"step": 490
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 10.416666666666666,
|
| 357 |
+
"grad_norm": 0.1780959814786911,
|
| 358 |
+
"learning_rate": 4.99e-05,
|
| 359 |
+
"loss": 0.0224,
|
| 360 |
+
"step": 500
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 10.625,
|
| 364 |
+
"grad_norm": 0.2748551070690155,
|
| 365 |
+
"learning_rate": 5.0900000000000004e-05,
|
| 366 |
+
"loss": 0.0245,
|
| 367 |
+
"step": 510
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 10.833333333333334,
|
| 371 |
+
"grad_norm": 0.20780707895755768,
|
| 372 |
+
"learning_rate": 5.19e-05,
|
| 373 |
+
"loss": 0.0212,
|
| 374 |
+
"step": 520
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 11.041666666666666,
|
| 378 |
+
"grad_norm": 0.3468799591064453,
|
| 379 |
+
"learning_rate": 5.2900000000000005e-05,
|
| 380 |
+
"loss": 0.0227,
|
| 381 |
+
"step": 530
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 11.25,
|
| 385 |
+
"grad_norm": 0.2328055202960968,
|
| 386 |
+
"learning_rate": 5.390000000000001e-05,
|
| 387 |
+
"loss": 0.0212,
|
| 388 |
+
"step": 540
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 11.458333333333334,
|
| 392 |
+
"grad_norm": 0.264704167842865,
|
| 393 |
+
"learning_rate": 5.4900000000000006e-05,
|
| 394 |
+
"loss": 0.0217,
|
| 395 |
+
"step": 550
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 11.666666666666666,
|
| 399 |
+
"grad_norm": 0.2830072045326233,
|
| 400 |
+
"learning_rate": 5.590000000000001e-05,
|
| 401 |
+
"loss": 0.0223,
|
| 402 |
+
"step": 560
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 11.875,
|
| 406 |
+
"grad_norm": 0.3284772038459778,
|
| 407 |
+
"learning_rate": 5.69e-05,
|
| 408 |
+
"loss": 0.0208,
|
| 409 |
+
"step": 570
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 12.083333333333334,
|
| 413 |
+
"grad_norm": 0.24000631272792816,
|
| 414 |
+
"learning_rate": 5.79e-05,
|
| 415 |
+
"loss": 0.0206,
|
| 416 |
+
"step": 580
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 12.291666666666666,
|
| 420 |
+
"grad_norm": 0.24238787591457367,
|
| 421 |
+
"learning_rate": 5.89e-05,
|
| 422 |
+
"loss": 0.0214,
|
| 423 |
+
"step": 590
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 12.5,
|
| 427 |
+
"grad_norm": 0.25679489970207214,
|
| 428 |
+
"learning_rate": 5.99e-05,
|
| 429 |
+
"loss": 0.0206,
|
| 430 |
+
"step": 600
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 12.708333333333334,
|
| 434 |
+
"grad_norm": 0.2557210326194763,
|
| 435 |
+
"learning_rate": 6.09e-05,
|
| 436 |
+
"loss": 0.0217,
|
| 437 |
+
"step": 610
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 12.916666666666666,
|
| 441 |
+
"grad_norm": 0.24070383608341217,
|
| 442 |
+
"learning_rate": 6.19e-05,
|
| 443 |
+
"loss": 0.0209,
|
| 444 |
+
"step": 620
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 13.125,
|
| 448 |
+
"grad_norm": 0.25813865661621094,
|
| 449 |
+
"learning_rate": 6.29e-05,
|
| 450 |
+
"loss": 0.0208,
|
| 451 |
+
"step": 630
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 13.333333333333334,
|
| 455 |
+
"grad_norm": 0.2484397441148758,
|
| 456 |
+
"learning_rate": 6.390000000000001e-05,
|
| 457 |
+
"loss": 0.0206,
|
| 458 |
+
"step": 640
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 13.541666666666666,
|
| 462 |
+
"grad_norm": 0.24208255112171173,
|
| 463 |
+
"learning_rate": 6.49e-05,
|
| 464 |
+
"loss": 0.0189,
|
| 465 |
+
"step": 650
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 13.75,
|
| 469 |
+
"grad_norm": 0.19995449483394623,
|
| 470 |
+
"learning_rate": 6.59e-05,
|
| 471 |
+
"loss": 0.0196,
|
| 472 |
+
"step": 660
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 13.958333333333334,
|
| 476 |
+
"grad_norm": 0.24547183513641357,
|
| 477 |
+
"learning_rate": 6.690000000000001e-05,
|
| 478 |
+
"loss": 0.0189,
|
| 479 |
+
"step": 670
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 14.166666666666666,
|
| 483 |
+
"grad_norm": 0.2594086229801178,
|
| 484 |
+
"learning_rate": 6.790000000000001e-05,
|
| 485 |
+
"loss": 0.019,
|
| 486 |
+
"step": 680
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 14.375,
|
| 490 |
+
"grad_norm": 0.2065507173538208,
|
| 491 |
+
"learning_rate": 6.89e-05,
|
| 492 |
+
"loss": 0.0192,
|
| 493 |
+
"step": 690
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 14.583333333333334,
|
| 497 |
+
"grad_norm": 0.2905842661857605,
|
| 498 |
+
"learning_rate": 6.99e-05,
|
| 499 |
+
"loss": 0.0187,
|
| 500 |
+
"step": 700
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 14.791666666666666,
|
| 504 |
+
"grad_norm": 0.2601730227470398,
|
| 505 |
+
"learning_rate": 7.09e-05,
|
| 506 |
+
"loss": 0.0196,
|
| 507 |
+
"step": 710
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 15.0,
|
| 511 |
+
"grad_norm": 0.28263160586357117,
|
| 512 |
+
"learning_rate": 7.19e-05,
|
| 513 |
+
"loss": 0.0192,
|
| 514 |
+
"step": 720
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 15.208333333333334,
|
| 518 |
+
"grad_norm": 0.3240365982055664,
|
| 519 |
+
"learning_rate": 7.29e-05,
|
| 520 |
+
"loss": 0.0195,
|
| 521 |
+
"step": 730
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 15.416666666666666,
|
| 525 |
+
"grad_norm": 0.24705904722213745,
|
| 526 |
+
"learning_rate": 7.390000000000001e-05,
|
| 527 |
+
"loss": 0.0187,
|
| 528 |
+
"step": 740
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 15.625,
|
| 532 |
+
"grad_norm": 0.2265540510416031,
|
| 533 |
+
"learning_rate": 7.49e-05,
|
| 534 |
+
"loss": 0.0179,
|
| 535 |
+
"step": 750
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 15.833333333333334,
|
| 539 |
+
"grad_norm": 0.193108931183815,
|
| 540 |
+
"learning_rate": 7.59e-05,
|
| 541 |
+
"loss": 0.0183,
|
| 542 |
+
"step": 760
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 16.041666666666668,
|
| 546 |
+
"grad_norm": 0.36317235231399536,
|
| 547 |
+
"learning_rate": 7.69e-05,
|
| 548 |
+
"loss": 0.0182,
|
| 549 |
+
"step": 770
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 16.25,
|
| 553 |
+
"grad_norm": 0.3126901686191559,
|
| 554 |
+
"learning_rate": 7.790000000000001e-05,
|
| 555 |
+
"loss": 0.0197,
|
| 556 |
+
"step": 780
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 16.458333333333332,
|
| 560 |
+
"grad_norm": 0.238056942820549,
|
| 561 |
+
"learning_rate": 7.890000000000001e-05,
|
| 562 |
+
"loss": 0.0182,
|
| 563 |
+
"step": 790
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 16.666666666666668,
|
| 567 |
+
"grad_norm": 0.25261858105659485,
|
| 568 |
+
"learning_rate": 7.99e-05,
|
| 569 |
+
"loss": 0.0176,
|
| 570 |
+
"step": 800
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 16.875,
|
| 574 |
+
"grad_norm": 0.1865801066160202,
|
| 575 |
+
"learning_rate": 8.090000000000001e-05,
|
| 576 |
+
"loss": 0.0174,
|
| 577 |
+
"step": 810
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 17.083333333333332,
|
| 581 |
+
"grad_norm": 0.18573996424674988,
|
| 582 |
+
"learning_rate": 8.19e-05,
|
| 583 |
+
"loss": 0.0176,
|
| 584 |
+
"step": 820
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 17.291666666666668,
|
| 588 |
+
"grad_norm": 0.21988113224506378,
|
| 589 |
+
"learning_rate": 8.29e-05,
|
| 590 |
+
"loss": 0.0174,
|
| 591 |
+
"step": 830
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 17.5,
|
| 595 |
+
"grad_norm": 0.196188285946846,
|
| 596 |
+
"learning_rate": 8.39e-05,
|
| 597 |
+
"loss": 0.0172,
|
| 598 |
+
"step": 840
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 17.708333333333332,
|
| 602 |
+
"grad_norm": 0.219802588224411,
|
| 603 |
+
"learning_rate": 8.49e-05,
|
| 604 |
+
"loss": 0.0181,
|
| 605 |
+
"step": 850
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 17.916666666666668,
|
| 609 |
+
"grad_norm": 0.2536110579967499,
|
| 610 |
+
"learning_rate": 8.59e-05,
|
| 611 |
+
"loss": 0.0178,
|
| 612 |
+
"step": 860
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 18.125,
|
| 616 |
+
"grad_norm": 0.3585800528526306,
|
| 617 |
+
"learning_rate": 8.69e-05,
|
| 618 |
+
"loss": 0.0176,
|
| 619 |
+
"step": 870
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 18.333333333333332,
|
| 623 |
+
"grad_norm": 0.18298107385635376,
|
| 624 |
+
"learning_rate": 8.790000000000001e-05,
|
| 625 |
+
"loss": 0.0173,
|
| 626 |
+
"step": 880
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 18.541666666666668,
|
| 630 |
+
"grad_norm": 0.22495491802692413,
|
| 631 |
+
"learning_rate": 8.89e-05,
|
| 632 |
+
"loss": 0.0166,
|
| 633 |
+
"step": 890
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 18.75,
|
| 637 |
+
"grad_norm": 0.21044448018074036,
|
| 638 |
+
"learning_rate": 8.99e-05,
|
| 639 |
+
"loss": 0.0161,
|
| 640 |
+
"step": 900
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 18.958333333333332,
|
| 644 |
+
"grad_norm": 0.19759798049926758,
|
| 645 |
+
"learning_rate": 9.090000000000001e-05,
|
| 646 |
+
"loss": 0.0153,
|
| 647 |
+
"step": 910
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 19.166666666666668,
|
| 651 |
+
"grad_norm": 0.22768504917621613,
|
| 652 |
+
"learning_rate": 9.190000000000001e-05,
|
| 653 |
+
"loss": 0.0168,
|
| 654 |
+
"step": 920
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 19.375,
|
| 658 |
+
"grad_norm": 0.1936219483613968,
|
| 659 |
+
"learning_rate": 9.290000000000001e-05,
|
| 660 |
+
"loss": 0.0162,
|
| 661 |
+
"step": 930
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 19.583333333333332,
|
| 665 |
+
"grad_norm": 0.21084393560886383,
|
| 666 |
+
"learning_rate": 9.39e-05,
|
| 667 |
+
"loss": 0.0179,
|
| 668 |
+
"step": 940
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 19.791666666666668,
|
| 672 |
+
"grad_norm": 0.20455226302146912,
|
| 673 |
+
"learning_rate": 9.49e-05,
|
| 674 |
+
"loss": 0.0154,
|
| 675 |
+
"step": 950
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 20.0,
|
| 679 |
+
"grad_norm": 0.26040470600128174,
|
| 680 |
+
"learning_rate": 9.59e-05,
|
| 681 |
+
"loss": 0.0155,
|
| 682 |
+
"step": 960
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 20.208333333333332,
|
| 686 |
+
"grad_norm": 0.2832554280757904,
|
| 687 |
+
"learning_rate": 9.69e-05,
|
| 688 |
+
"loss": 0.0164,
|
| 689 |
+
"step": 970
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 20.416666666666668,
|
| 693 |
+
"grad_norm": 0.2514568567276001,
|
| 694 |
+
"learning_rate": 9.790000000000001e-05,
|
| 695 |
+
"loss": 0.017,
|
| 696 |
+
"step": 980
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 20.625,
|
| 700 |
+
"grad_norm": 0.20968522131443024,
|
| 701 |
+
"learning_rate": 9.89e-05,
|
| 702 |
+
"loss": 0.016,
|
| 703 |
+
"step": 990
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 20.833333333333332,
|
| 707 |
+
"grad_norm": 0.20158836245536804,
|
| 708 |
+
"learning_rate": 9.99e-05,
|
| 709 |
+
"loss": 0.0153,
|
| 710 |
+
"step": 1000
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 21.041666666666668,
|
| 714 |
+
"grad_norm": 0.24486659467220306,
|
| 715 |
+
"learning_rate": 9.999994463727085e-05,
|
| 716 |
+
"loss": 0.0157,
|
| 717 |
+
"step": 1010
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 21.25,
|
| 721 |
+
"grad_norm": 0.2102823108434677,
|
| 722 |
+
"learning_rate": 9.999975326009292e-05,
|
| 723 |
+
"loss": 0.0166,
|
| 724 |
+
"step": 1020
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 21.458333333333332,
|
| 728 |
+
"grad_norm": 0.2416916787624359,
|
| 729 |
+
"learning_rate": 9.999942518549879e-05,
|
| 730 |
+
"loss": 0.0161,
|
| 731 |
+
"step": 1030
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 21.666666666666668,
|
| 735 |
+
"grad_norm": 0.14526234567165375,
|
| 736 |
+
"learning_rate": 9.999896041438544e-05,
|
| 737 |
+
"loss": 0.0156,
|
| 738 |
+
"step": 1040
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 21.875,
|
| 742 |
+
"grad_norm": 0.21768969297409058,
|
| 743 |
+
"learning_rate": 9.999835894802353e-05,
|
| 744 |
+
"loss": 0.0149,
|
| 745 |
+
"step": 1050
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 22.083333333333332,
|
| 749 |
+
"grad_norm": 0.2318434864282608,
|
| 750 |
+
"learning_rate": 9.999762078805743e-05,
|
| 751 |
+
"loss": 0.0153,
|
| 752 |
+
"step": 1060
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 22.291666666666668,
|
| 756 |
+
"grad_norm": 0.2798757255077362,
|
| 757 |
+
"learning_rate": 9.999674593650526e-05,
|
| 758 |
+
"loss": 0.0155,
|
| 759 |
+
"step": 1070
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 22.5,
|
| 763 |
+
"grad_norm": 0.24006976187229156,
|
| 764 |
+
"learning_rate": 9.99957343957588e-05,
|
| 765 |
+
"loss": 0.0153,
|
| 766 |
+
"step": 1080
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 22.708333333333332,
|
| 770 |
+
"grad_norm": 0.21017949283123016,
|
| 771 |
+
"learning_rate": 9.99945861685836e-05,
|
| 772 |
+
"loss": 0.0162,
|
| 773 |
+
"step": 1090
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 22.916666666666668,
|
| 777 |
+
"grad_norm": 0.19796672463417053,
|
| 778 |
+
"learning_rate": 9.999330125811884e-05,
|
| 779 |
+
"loss": 0.0149,
|
| 780 |
+
"step": 1100
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 23.125,
|
| 784 |
+
"grad_norm": 0.2603440582752228,
|
| 785 |
+
"learning_rate": 9.999187966787744e-05,
|
| 786 |
+
"loss": 0.015,
|
| 787 |
+
"step": 1110
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 23.333333333333332,
|
| 791 |
+
"grad_norm": 0.23967282474040985,
|
| 792 |
+
"learning_rate": 9.999032140174595e-05,
|
| 793 |
+
"loss": 0.0151,
|
| 794 |
+
"step": 1120
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 23.541666666666668,
|
| 798 |
+
"grad_norm": 0.18136964738368988,
|
| 799 |
+
"learning_rate": 9.998862646398464e-05,
|
| 800 |
+
"loss": 0.0152,
|
| 801 |
+
"step": 1130
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 23.75,
|
| 805 |
+
"grad_norm": 0.252003014087677,
|
| 806 |
+
"learning_rate": 9.998679485922739e-05,
|
| 807 |
+
"loss": 0.0153,
|
| 808 |
+
"step": 1140
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 23.958333333333332,
|
| 812 |
+
"grad_norm": 0.2043199986219406,
|
| 813 |
+
"learning_rate": 9.998482659248174e-05,
|
| 814 |
+
"loss": 0.0157,
|
| 815 |
+
"step": 1150
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 24.166666666666668,
|
| 819 |
+
"grad_norm": 0.18087252974510193,
|
| 820 |
+
"learning_rate": 9.998272166912883e-05,
|
| 821 |
+
"loss": 0.0154,
|
| 822 |
+
"step": 1160
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 24.375,
|
| 826 |
+
"grad_norm": 0.21990996599197388,
|
| 827 |
+
"learning_rate": 9.998048009492347e-05,
|
| 828 |
+
"loss": 0.015,
|
| 829 |
+
"step": 1170
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 24.583333333333332,
|
| 833 |
+
"grad_norm": 0.2346096634864807,
|
| 834 |
+
"learning_rate": 9.997810187599403e-05,
|
| 835 |
+
"loss": 0.0154,
|
| 836 |
+
"step": 1180
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 24.791666666666668,
|
| 840 |
+
"grad_norm": 0.18085162341594696,
|
| 841 |
+
"learning_rate": 9.997558701884249e-05,
|
| 842 |
+
"loss": 0.015,
|
| 843 |
+
"step": 1190
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 25.0,
|
| 847 |
+
"grad_norm": 0.3023941218852997,
|
| 848 |
+
"learning_rate": 9.997293553034433e-05,
|
| 849 |
+
"loss": 0.0133,
|
| 850 |
+
"step": 1200
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 25.208333333333332,
|
| 854 |
+
"grad_norm": 0.20299163460731506,
|
| 855 |
+
"learning_rate": 9.997014741774866e-05,
|
| 856 |
+
"loss": 0.0133,
|
| 857 |
+
"step": 1210
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 25.416666666666668,
|
| 861 |
+
"grad_norm": 0.26265865564346313,
|
| 862 |
+
"learning_rate": 9.996722268867803e-05,
|
| 863 |
+
"loss": 0.0137,
|
| 864 |
+
"step": 1220
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 25.625,
|
| 868 |
+
"grad_norm": 0.24567152559757233,
|
| 869 |
+
"learning_rate": 9.996416135112858e-05,
|
| 870 |
+
"loss": 0.0132,
|
| 871 |
+
"step": 1230
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 25.833333333333332,
|
| 875 |
+
"grad_norm": 0.17860938608646393,
|
| 876 |
+
"learning_rate": 9.996096341346988e-05,
|
| 877 |
+
"loss": 0.0135,
|
| 878 |
+
"step": 1240
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 26.041666666666668,
|
| 882 |
+
"grad_norm": 0.27000877261161804,
|
| 883 |
+
"learning_rate": 9.995762888444495e-05,
|
| 884 |
+
"loss": 0.0127,
|
| 885 |
+
"step": 1250
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 26.25,
|
| 889 |
+
"grad_norm": 0.21311180293560028,
|
| 890 |
+
"learning_rate": 9.995415777317027e-05,
|
| 891 |
+
"loss": 0.0146,
|
| 892 |
+
"step": 1260
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 26.458333333333332,
|
| 896 |
+
"grad_norm": 0.19608379900455475,
|
| 897 |
+
"learning_rate": 9.995055008913574e-05,
|
| 898 |
+
"loss": 0.0141,
|
| 899 |
+
"step": 1270
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 26.666666666666668,
|
| 903 |
+
"grad_norm": 0.14746077358722687,
|
| 904 |
+
"learning_rate": 9.994680584220463e-05,
|
| 905 |
+
"loss": 0.0127,
|
| 906 |
+
"step": 1280
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 26.875,
|
| 910 |
+
"grad_norm": 0.2702135443687439,
|
| 911 |
+
"learning_rate": 9.994292504261355e-05,
|
| 912 |
+
"loss": 0.0139,
|
| 913 |
+
"step": 1290
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 27.083333333333332,
|
| 917 |
+
"grad_norm": 0.16915710270404816,
|
| 918 |
+
"learning_rate": 9.993890770097247e-05,
|
| 919 |
+
"loss": 0.0125,
|
| 920 |
+
"step": 1300
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 27.291666666666668,
|
| 924 |
+
"grad_norm": 0.18446531891822815,
|
| 925 |
+
"learning_rate": 9.993475382826467e-05,
|
| 926 |
+
"loss": 0.0133,
|
| 927 |
+
"step": 1310
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 27.5,
|
| 931 |
+
"grad_norm": 0.1798723340034485,
|
| 932 |
+
"learning_rate": 9.993046343584664e-05,
|
| 933 |
+
"loss": 0.0127,
|
| 934 |
+
"step": 1320
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 27.708333333333332,
|
| 938 |
+
"grad_norm": 0.1655735820531845,
|
| 939 |
+
"learning_rate": 9.992603653544816e-05,
|
| 940 |
+
"loss": 0.0123,
|
| 941 |
+
"step": 1330
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 27.916666666666668,
|
| 945 |
+
"grad_norm": 0.21007224917411804,
|
| 946 |
+
"learning_rate": 9.992147313917222e-05,
|
| 947 |
+
"loss": 0.0125,
|
| 948 |
+
"step": 1340
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 28.125,
|
| 952 |
+
"grad_norm": 0.2112513929605484,
|
| 953 |
+
"learning_rate": 9.991677325949497e-05,
|
| 954 |
+
"loss": 0.0125,
|
| 955 |
+
"step": 1350
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 28.333333333333332,
|
| 959 |
+
"grad_norm": 0.20448802411556244,
|
| 960 |
+
"learning_rate": 9.991193690926568e-05,
|
| 961 |
+
"loss": 0.0133,
|
| 962 |
+
"step": 1360
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 28.541666666666668,
|
| 966 |
+
"grad_norm": 0.1936739832162857,
|
| 967 |
+
"learning_rate": 9.990696410170678e-05,
|
| 968 |
+
"loss": 0.0125,
|
| 969 |
+
"step": 1370
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 28.75,
|
| 973 |
+
"grad_norm": 0.20615226030349731,
|
| 974 |
+
"learning_rate": 9.990185485041371e-05,
|
| 975 |
+
"loss": 0.0121,
|
| 976 |
+
"step": 1380
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 28.958333333333332,
|
| 980 |
+
"grad_norm": 0.18444038927555084,
|
| 981 |
+
"learning_rate": 9.989660916935498e-05,
|
| 982 |
+
"loss": 0.0123,
|
| 983 |
+
"step": 1390
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 29.166666666666668,
|
| 987 |
+
"grad_norm": 0.21376805007457733,
|
| 988 |
+
"learning_rate": 9.989122707287208e-05,
|
| 989 |
+
"loss": 0.013,
|
| 990 |
+
"step": 1400
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 29.375,
|
| 994 |
+
"grad_norm": 0.25530359148979187,
|
| 995 |
+
"learning_rate": 9.988570857567945e-05,
|
| 996 |
+
"loss": 0.0125,
|
| 997 |
+
"step": 1410
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 29.583333333333332,
|
| 1001 |
+
"grad_norm": 0.21851158142089844,
|
| 1002 |
+
"learning_rate": 9.988005369286446e-05,
|
| 1003 |
+
"loss": 0.0123,
|
| 1004 |
+
"step": 1420
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 29.791666666666668,
|
| 1008 |
+
"grad_norm": 0.19980329275131226,
|
| 1009 |
+
"learning_rate": 9.987426243988734e-05,
|
| 1010 |
+
"loss": 0.0134,
|
| 1011 |
+
"step": 1430
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 30.0,
|
| 1015 |
+
"grad_norm": 0.3672507107257843,
|
| 1016 |
+
"learning_rate": 9.986833483258114e-05,
|
| 1017 |
+
"loss": 0.0108,
|
| 1018 |
+
"step": 1440
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 30.208333333333332,
|
| 1022 |
+
"grad_norm": 0.26760515570640564,
|
| 1023 |
+
"learning_rate": 9.986227088715173e-05,
|
| 1024 |
+
"loss": 0.0144,
|
| 1025 |
+
"step": 1450
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 30.416666666666668,
|
| 1029 |
+
"grad_norm": 0.1989632248878479,
|
| 1030 |
+
"learning_rate": 9.98560706201777e-05,
|
| 1031 |
+
"loss": 0.0135,
|
| 1032 |
+
"step": 1460
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 30.625,
|
| 1036 |
+
"grad_norm": 0.20122896134853363,
|
| 1037 |
+
"learning_rate": 9.984973404861036e-05,
|
| 1038 |
+
"loss": 0.0137,
|
| 1039 |
+
"step": 1470
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 30.833333333333332,
|
| 1043 |
+
"grad_norm": 0.22380387783050537,
|
| 1044 |
+
"learning_rate": 9.984326118977361e-05,
|
| 1045 |
+
"loss": 0.0138,
|
| 1046 |
+
"step": 1480
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 31.041666666666668,
|
| 1050 |
+
"grad_norm": 0.17185676097869873,
|
| 1051 |
+
"learning_rate": 9.983665206136406e-05,
|
| 1052 |
+
"loss": 0.0119,
|
| 1053 |
+
"step": 1490
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 31.25,
|
| 1057 |
+
"grad_norm": 0.20098410546779633,
|
| 1058 |
+
"learning_rate": 9.982990668145075e-05,
|
| 1059 |
+
"loss": 0.0119,
|
| 1060 |
+
"step": 1500
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 31.458333333333332,
|
| 1064 |
+
"grad_norm": 0.16579347848892212,
|
| 1065 |
+
"learning_rate": 9.982302506847534e-05,
|
| 1066 |
+
"loss": 0.013,
|
| 1067 |
+
"step": 1510
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 31.666666666666668,
|
| 1071 |
+
"grad_norm": 0.15271247923374176,
|
| 1072 |
+
"learning_rate": 9.981600724125189e-05,
|
| 1073 |
+
"loss": 0.0131,
|
| 1074 |
+
"step": 1520
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 31.875,
|
| 1078 |
+
"grad_norm": 0.16309182345867157,
|
| 1079 |
+
"learning_rate": 9.980885321896685e-05,
|
| 1080 |
+
"loss": 0.0121,
|
| 1081 |
+
"step": 1530
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 32.083333333333336,
|
| 1085 |
+
"grad_norm": 0.17842234671115875,
|
| 1086 |
+
"learning_rate": 9.980156302117905e-05,
|
| 1087 |
+
"loss": 0.0112,
|
| 1088 |
+
"step": 1540
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 32.291666666666664,
|
| 1092 |
+
"grad_norm": 0.18424782156944275,
|
| 1093 |
+
"learning_rate": 9.979413666781963e-05,
|
| 1094 |
+
"loss": 0.0117,
|
| 1095 |
+
"step": 1550
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 32.5,
|
| 1099 |
+
"grad_norm": 0.17072798311710358,
|
| 1100 |
+
"learning_rate": 9.978657417919193e-05,
|
| 1101 |
+
"loss": 0.0113,
|
| 1102 |
+
"step": 1560
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 32.708333333333336,
|
| 1106 |
+
"grad_norm": 0.16320225596427917,
|
| 1107 |
+
"learning_rate": 9.977887557597153e-05,
|
| 1108 |
+
"loss": 0.0113,
|
| 1109 |
+
"step": 1570
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 32.916666666666664,
|
| 1113 |
+
"grad_norm": 0.17314594984054565,
|
| 1114 |
+
"learning_rate": 9.97710408792061e-05,
|
| 1115 |
+
"loss": 0.0108,
|
| 1116 |
+
"step": 1580
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 33.125,
|
| 1120 |
+
"grad_norm": 0.21347780525684357,
|
| 1121 |
+
"learning_rate": 9.976307011031542e-05,
|
| 1122 |
+
"loss": 0.0111,
|
| 1123 |
+
"step": 1590
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 33.333333333333336,
|
| 1127 |
+
"grad_norm": 0.1748444139957428,
|
| 1128 |
+
"learning_rate": 9.975496329109126e-05,
|
| 1129 |
+
"loss": 0.0118,
|
| 1130 |
+
"step": 1600
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 33.541666666666664,
|
| 1134 |
+
"grad_norm": 0.22180508077144623,
|
| 1135 |
+
"learning_rate": 9.974672044369732e-05,
|
| 1136 |
+
"loss": 0.0111,
|
| 1137 |
+
"step": 1610
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 33.75,
|
| 1141 |
+
"grad_norm": 0.1914110630750656,
|
| 1142 |
+
"learning_rate": 9.97383415906693e-05,
|
| 1143 |
+
"loss": 0.0124,
|
| 1144 |
+
"step": 1620
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 33.958333333333336,
|
| 1148 |
+
"grad_norm": 0.1542980968952179,
|
| 1149 |
+
"learning_rate": 9.97298267549146e-05,
|
| 1150 |
+
"loss": 0.0119,
|
| 1151 |
+
"step": 1630
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 34.166666666666664,
|
| 1155 |
+
"grad_norm": 0.2009705901145935,
|
| 1156 |
+
"learning_rate": 9.972117595971249e-05,
|
| 1157 |
+
"loss": 0.0124,
|
| 1158 |
+
"step": 1640
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 34.375,
|
| 1162 |
+
"grad_norm": 0.1812678873538971,
|
| 1163 |
+
"learning_rate": 9.971238922871391e-05,
|
| 1164 |
+
"loss": 0.0116,
|
| 1165 |
+
"step": 1650
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 34.583333333333336,
|
| 1169 |
+
"grad_norm": 0.13899721205234528,
|
| 1170 |
+
"learning_rate": 9.970346658594142e-05,
|
| 1171 |
+
"loss": 0.0113,
|
| 1172 |
+
"step": 1660
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 34.791666666666664,
|
| 1176 |
+
"grad_norm": 0.12181176245212555,
|
| 1177 |
+
"learning_rate": 9.969440805578923e-05,
|
| 1178 |
+
"loss": 0.0115,
|
| 1179 |
+
"step": 1670
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 35.0,
|
| 1183 |
+
"grad_norm": 0.20665058493614197,
|
| 1184 |
+
"learning_rate": 9.968521366302298e-05,
|
| 1185 |
+
"loss": 0.0104,
|
| 1186 |
+
"step": 1680
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 35.208333333333336,
|
| 1190 |
+
"grad_norm": 0.14869576692581177,
|
| 1191 |
+
"learning_rate": 9.967588343277981e-05,
|
| 1192 |
+
"loss": 0.0111,
|
| 1193 |
+
"step": 1690
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 35.416666666666664,
|
| 1197 |
+
"grad_norm": 0.21121446788311005,
|
| 1198 |
+
"learning_rate": 9.966641739056818e-05,
|
| 1199 |
+
"loss": 0.0121,
|
| 1200 |
+
"step": 1700
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 35.625,
|
| 1204 |
+
"grad_norm": 0.2114773988723755,
|
| 1205 |
+
"learning_rate": 9.965681556226793e-05,
|
| 1206 |
+
"loss": 0.0117,
|
| 1207 |
+
"step": 1710
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 35.833333333333336,
|
| 1211 |
+
"grad_norm": 0.2102501392364502,
|
| 1212 |
+
"learning_rate": 9.964707797413006e-05,
|
| 1213 |
+
"loss": 0.0111,
|
| 1214 |
+
"step": 1720
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 36.041666666666664,
|
| 1218 |
+
"grad_norm": 0.197468101978302,
|
| 1219 |
+
"learning_rate": 9.963720465277679e-05,
|
| 1220 |
+
"loss": 0.0113,
|
| 1221 |
+
"step": 1730
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 36.25,
|
| 1225 |
+
"grad_norm": 0.21283449232578278,
|
| 1226 |
+
"learning_rate": 9.96271956252014e-05,
|
| 1227 |
+
"loss": 0.0111,
|
| 1228 |
+
"step": 1740
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 36.458333333333336,
|
| 1232 |
+
"grad_norm": 0.20028117299079895,
|
| 1233 |
+
"learning_rate": 9.961705091876816e-05,
|
| 1234 |
+
"loss": 0.012,
|
| 1235 |
+
"step": 1750
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 36.666666666666664,
|
| 1239 |
+
"grad_norm": 0.15327388048171997,
|
| 1240 |
+
"learning_rate": 9.960677056121235e-05,
|
| 1241 |
+
"loss": 0.0104,
|
| 1242 |
+
"step": 1760
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 36.875,
|
| 1246 |
+
"grad_norm": 0.16232164204120636,
|
| 1247 |
+
"learning_rate": 9.959635458064005e-05,
|
| 1248 |
+
"loss": 0.0107,
|
| 1249 |
+
"step": 1770
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 37.083333333333336,
|
| 1253 |
+
"grad_norm": 0.18801745772361755,
|
| 1254 |
+
"learning_rate": 9.958580300552815e-05,
|
| 1255 |
+
"loss": 0.0106,
|
| 1256 |
+
"step": 1780
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 37.291666666666664,
|
| 1260 |
+
"grad_norm": 0.15016761422157288,
|
| 1261 |
+
"learning_rate": 9.957511586472426e-05,
|
| 1262 |
+
"loss": 0.0104,
|
| 1263 |
+
"step": 1790
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 37.5,
|
| 1267 |
+
"grad_norm": 0.15682189166545868,
|
| 1268 |
+
"learning_rate": 9.956429318744662e-05,
|
| 1269 |
+
"loss": 0.0101,
|
| 1270 |
+
"step": 1800
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 37.708333333333336,
|
| 1274 |
+
"grad_norm": 0.2034134864807129,
|
| 1275 |
+
"learning_rate": 9.955333500328404e-05,
|
| 1276 |
+
"loss": 0.0106,
|
| 1277 |
+
"step": 1810
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 37.916666666666664,
|
| 1281 |
+
"grad_norm": 0.16577675938606262,
|
| 1282 |
+
"learning_rate": 9.95422413421957e-05,
|
| 1283 |
+
"loss": 0.0104,
|
| 1284 |
+
"step": 1820
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 38.125,
|
| 1288 |
+
"grad_norm": 0.16364583373069763,
|
| 1289 |
+
"learning_rate": 9.953101223451133e-05,
|
| 1290 |
+
"loss": 0.0101,
|
| 1291 |
+
"step": 1830
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 38.333333333333336,
|
| 1295 |
+
"grad_norm": 0.14578108489513397,
|
| 1296 |
+
"learning_rate": 9.951964771093085e-05,
|
| 1297 |
+
"loss": 0.0105,
|
| 1298 |
+
"step": 1840
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 38.541666666666664,
|
| 1302 |
+
"grad_norm": 0.14456351101398468,
|
| 1303 |
+
"learning_rate": 9.950814780252442e-05,
|
| 1304 |
+
"loss": 0.0101,
|
| 1305 |
+
"step": 1850
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 38.75,
|
| 1309 |
+
"grad_norm": 0.20234094560146332,
|
| 1310 |
+
"learning_rate": 9.949651254073236e-05,
|
| 1311 |
+
"loss": 0.0109,
|
| 1312 |
+
"step": 1860
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 38.958333333333336,
|
| 1316 |
+
"grad_norm": 0.16703951358795166,
|
| 1317 |
+
"learning_rate": 9.948474195736504e-05,
|
| 1318 |
+
"loss": 0.0105,
|
| 1319 |
+
"step": 1870
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 39.166666666666664,
|
| 1323 |
+
"grad_norm": 0.17165902256965637,
|
| 1324 |
+
"learning_rate": 9.947283608460277e-05,
|
| 1325 |
+
"loss": 0.0103,
|
| 1326 |
+
"step": 1880
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 39.375,
|
| 1330 |
+
"grad_norm": 0.14556387066841125,
|
| 1331 |
+
"learning_rate": 9.946079495499577e-05,
|
| 1332 |
+
"loss": 0.0101,
|
| 1333 |
+
"step": 1890
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 39.583333333333336,
|
| 1337 |
+
"grad_norm": 0.19160117208957672,
|
| 1338 |
+
"learning_rate": 9.944861860146401e-05,
|
| 1339 |
+
"loss": 0.0113,
|
| 1340 |
+
"step": 1900
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 39.791666666666664,
|
| 1344 |
+
"grad_norm": 0.1653631329536438,
|
| 1345 |
+
"learning_rate": 9.943630705729719e-05,
|
| 1346 |
+
"loss": 0.01,
|
| 1347 |
+
"step": 1910
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 40.0,
|
| 1351 |
+
"grad_norm": 0.20655889809131622,
|
| 1352 |
+
"learning_rate": 9.942386035615459e-05,
|
| 1353 |
+
"loss": 0.0106,
|
| 1354 |
+
"step": 1920
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 40.208333333333336,
|
| 1358 |
+
"grad_norm": 0.1884182095527649,
|
| 1359 |
+
"learning_rate": 9.941127853206503e-05,
|
| 1360 |
+
"loss": 0.0108,
|
| 1361 |
+
"step": 1930
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 40.416666666666664,
|
| 1365 |
+
"grad_norm": 0.15719261765480042,
|
| 1366 |
+
"learning_rate": 9.939856161942673e-05,
|
| 1367 |
+
"loss": 0.0105,
|
| 1368 |
+
"step": 1940
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 40.625,
|
| 1372 |
+
"grad_norm": 0.15232212841510773,
|
| 1373 |
+
"learning_rate": 9.938570965300724e-05,
|
| 1374 |
+
"loss": 0.0102,
|
| 1375 |
+
"step": 1950
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 40.833333333333336,
|
| 1379 |
+
"grad_norm": 0.1885528415441513,
|
| 1380 |
+
"learning_rate": 9.937272266794335e-05,
|
| 1381 |
+
"loss": 0.0109,
|
| 1382 |
+
"step": 1960
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 41.041666666666664,
|
| 1386 |
+
"grad_norm": 0.15623626112937927,
|
| 1387 |
+
"learning_rate": 9.935960069974096e-05,
|
| 1388 |
+
"loss": 0.0099,
|
| 1389 |
+
"step": 1970
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 41.25,
|
| 1393 |
+
"grad_norm": 0.15828962624073029,
|
| 1394 |
+
"learning_rate": 9.934634378427506e-05,
|
| 1395 |
+
"loss": 0.0111,
|
| 1396 |
+
"step": 1980
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 41.458333333333336,
|
| 1400 |
+
"grad_norm": 0.17898498475551605,
|
| 1401 |
+
"learning_rate": 9.933295195778954e-05,
|
| 1402 |
+
"loss": 0.0096,
|
| 1403 |
+
"step": 1990
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 41.666666666666664,
|
| 1407 |
+
"grad_norm": 0.156189426779747,
|
| 1408 |
+
"learning_rate": 9.931942525689715e-05,
|
| 1409 |
+
"loss": 0.0097,
|
| 1410 |
+
"step": 2000
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 41.875,
|
| 1414 |
+
"grad_norm": 0.18208195269107819,
|
| 1415 |
+
"learning_rate": 9.930576371857936e-05,
|
| 1416 |
+
"loss": 0.0104,
|
| 1417 |
+
"step": 2010
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 42.083333333333336,
|
| 1421 |
+
"grad_norm": 0.22289738059043884,
|
| 1422 |
+
"learning_rate": 9.929196738018629e-05,
|
| 1423 |
+
"loss": 0.0101,
|
| 1424 |
+
"step": 2020
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 42.291666666666664,
|
| 1428 |
+
"grad_norm": 0.1856096237897873,
|
| 1429 |
+
"learning_rate": 9.927803627943662e-05,
|
| 1430 |
+
"loss": 0.0102,
|
| 1431 |
+
"step": 2030
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 42.5,
|
| 1435 |
+
"grad_norm": 0.18785765767097473,
|
| 1436 |
+
"learning_rate": 9.926397045441744e-05,
|
| 1437 |
+
"loss": 0.01,
|
| 1438 |
+
"step": 2040
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 42.708333333333336,
|
| 1442 |
+
"grad_norm": 0.17422276735305786,
|
| 1443 |
+
"learning_rate": 9.924976994358417e-05,
|
| 1444 |
+
"loss": 0.01,
|
| 1445 |
+
"step": 2050
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 42.916666666666664,
|
| 1449 |
+
"grad_norm": 0.14945165812969208,
|
| 1450 |
+
"learning_rate": 9.923543478576048e-05,
|
| 1451 |
+
"loss": 0.0099,
|
| 1452 |
+
"step": 2060
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 43.125,
|
| 1456 |
+
"grad_norm": 0.1581973135471344,
|
| 1457 |
+
"learning_rate": 9.922096502013813e-05,
|
| 1458 |
+
"loss": 0.0114,
|
| 1459 |
+
"step": 2070
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 43.333333333333336,
|
| 1463 |
+
"grad_norm": 0.1979743242263794,
|
| 1464 |
+
"learning_rate": 9.92063606862769e-05,
|
| 1465 |
+
"loss": 0.0115,
|
| 1466 |
+
"step": 2080
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 43.541666666666664,
|
| 1470 |
+
"grad_norm": 0.1677751988172531,
|
| 1471 |
+
"learning_rate": 9.919162182410453e-05,
|
| 1472 |
+
"loss": 0.0103,
|
| 1473 |
+
"step": 2090
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 43.75,
|
| 1477 |
+
"grad_norm": 0.12610012292861938,
|
| 1478 |
+
"learning_rate": 9.917674847391645e-05,
|
| 1479 |
+
"loss": 0.0098,
|
| 1480 |
+
"step": 2100
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 43.958333333333336,
|
| 1484 |
+
"grad_norm": 0.1920396089553833,
|
| 1485 |
+
"learning_rate": 9.916174067637584e-05,
|
| 1486 |
+
"loss": 0.0094,
|
| 1487 |
+
"step": 2110
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 44.166666666666664,
|
| 1491 |
+
"grad_norm": 0.159335657954216,
|
| 1492 |
+
"learning_rate": 9.914659847251348e-05,
|
| 1493 |
+
"loss": 0.0105,
|
| 1494 |
+
"step": 2120
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 44.375,
|
| 1498 |
+
"grad_norm": 0.18806587159633636,
|
| 1499 |
+
"learning_rate": 9.913132190372753e-05,
|
| 1500 |
+
"loss": 0.0101,
|
| 1501 |
+
"step": 2130
|
| 1502 |
+
},
|
| 1503 |
+
{
|
| 1504 |
+
"epoch": 44.583333333333336,
|
| 1505 |
+
"grad_norm": 0.1827884465456009,
|
| 1506 |
+
"learning_rate": 9.911591101178359e-05,
|
| 1507 |
+
"loss": 0.0103,
|
| 1508 |
+
"step": 2140
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 44.791666666666664,
|
| 1512 |
+
"grad_norm": 0.21888400614261627,
|
| 1513 |
+
"learning_rate": 9.910036583881443e-05,
|
| 1514 |
+
"loss": 0.01,
|
| 1515 |
+
"step": 2150
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 45.0,
|
| 1519 |
+
"grad_norm": 0.20317022502422333,
|
| 1520 |
+
"learning_rate": 9.908468642731995e-05,
|
| 1521 |
+
"loss": 0.0089,
|
| 1522 |
+
"step": 2160
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 45.208333333333336,
|
| 1526 |
+
"grad_norm": 0.23497819900512695,
|
| 1527 |
+
"learning_rate": 9.906887282016707e-05,
|
| 1528 |
+
"loss": 0.0105,
|
| 1529 |
+
"step": 2170
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 45.416666666666664,
|
| 1533 |
+
"grad_norm": 0.18091560900211334,
|
| 1534 |
+
"learning_rate": 9.90529250605896e-05,
|
| 1535 |
+
"loss": 0.0099,
|
| 1536 |
+
"step": 2180
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 45.625,
|
| 1540 |
+
"grad_norm": 0.14061926305294037,
|
| 1541 |
+
"learning_rate": 9.903684319218809e-05,
|
| 1542 |
+
"loss": 0.01,
|
| 1543 |
+
"step": 2190
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 45.833333333333336,
|
| 1547 |
+
"grad_norm": 0.19375453889369965,
|
| 1548 |
+
"learning_rate": 9.902062725892976e-05,
|
| 1549 |
+
"loss": 0.0096,
|
| 1550 |
+
"step": 2200
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 46.041666666666664,
|
| 1554 |
+
"grad_norm": 0.21020196378231049,
|
| 1555 |
+
"learning_rate": 9.900427730514834e-05,
|
| 1556 |
+
"loss": 0.0096,
|
| 1557 |
+
"step": 2210
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 46.25,
|
| 1561 |
+
"grad_norm": 0.15842080116271973,
|
| 1562 |
+
"learning_rate": 9.8987793375544e-05,
|
| 1563 |
+
"loss": 0.0097,
|
| 1564 |
+
"step": 2220
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 46.458333333333336,
|
| 1568 |
+
"grad_norm": 0.15409104526042938,
|
| 1569 |
+
"learning_rate": 9.897117551518318e-05,
|
| 1570 |
+
"loss": 0.0094,
|
| 1571 |
+
"step": 2230
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 46.666666666666664,
|
| 1575 |
+
"grad_norm": 0.1582673192024231,
|
| 1576 |
+
"learning_rate": 9.895442376949844e-05,
|
| 1577 |
+
"loss": 0.0102,
|
| 1578 |
+
"step": 2240
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 46.875,
|
| 1582 |
+
"grad_norm": 0.14312148094177246,
|
| 1583 |
+
"learning_rate": 9.893753818428845e-05,
|
| 1584 |
+
"loss": 0.0094,
|
| 1585 |
+
"step": 2250
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 47.083333333333336,
|
| 1589 |
+
"grad_norm": 0.17130939662456512,
|
| 1590 |
+
"learning_rate": 9.892051880571773e-05,
|
| 1591 |
+
"loss": 0.0092,
|
| 1592 |
+
"step": 2260
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 47.291666666666664,
|
| 1596 |
+
"grad_norm": 0.18089613318443298,
|
| 1597 |
+
"learning_rate": 9.890336568031663e-05,
|
| 1598 |
+
"loss": 0.0098,
|
| 1599 |
+
"step": 2270
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 47.5,
|
| 1603 |
+
"grad_norm": 0.14833058416843414,
|
| 1604 |
+
"learning_rate": 9.888607885498113e-05,
|
| 1605 |
+
"loss": 0.0098,
|
| 1606 |
+
"step": 2280
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 47.708333333333336,
|
| 1610 |
+
"grad_norm": 0.1335451900959015,
|
| 1611 |
+
"learning_rate": 9.886865837697275e-05,
|
| 1612 |
+
"loss": 0.0088,
|
| 1613 |
+
"step": 2290
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 47.916666666666664,
|
| 1617 |
+
"grad_norm": 0.1326667219400406,
|
| 1618 |
+
"learning_rate": 9.88511042939184e-05,
|
| 1619 |
+
"loss": 0.0094,
|
| 1620 |
+
"step": 2300
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 48.125,
|
| 1624 |
+
"grad_norm": 0.16099606454372406,
|
| 1625 |
+
"learning_rate": 9.883341665381028e-05,
|
| 1626 |
+
"loss": 0.0092,
|
| 1627 |
+
"step": 2310
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 48.333333333333336,
|
| 1631 |
+
"grad_norm": 0.16287587583065033,
|
| 1632 |
+
"learning_rate": 9.881559550500575e-05,
|
| 1633 |
+
"loss": 0.0099,
|
| 1634 |
+
"step": 2320
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 48.541666666666664,
|
| 1638 |
+
"grad_norm": 0.20915569365024567,
|
| 1639 |
+
"learning_rate": 9.879764089622712e-05,
|
| 1640 |
+
"loss": 0.0097,
|
| 1641 |
+
"step": 2330
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 48.75,
|
| 1645 |
+
"grad_norm": 0.1475881040096283,
|
| 1646 |
+
"learning_rate": 9.87795528765616e-05,
|
| 1647 |
+
"loss": 0.0097,
|
| 1648 |
+
"step": 2340
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 48.958333333333336,
|
| 1652 |
+
"grad_norm": 0.15001028776168823,
|
| 1653 |
+
"learning_rate": 9.876133149546118e-05,
|
| 1654 |
+
"loss": 0.0096,
|
| 1655 |
+
"step": 2350
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 49.166666666666664,
|
| 1659 |
+
"grad_norm": 0.1850421130657196,
|
| 1660 |
+
"learning_rate": 9.874297680274238e-05,
|
| 1661 |
+
"loss": 0.0095,
|
| 1662 |
+
"step": 2360
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 49.375,
|
| 1666 |
+
"grad_norm": 0.1945352703332901,
|
| 1667 |
+
"learning_rate": 9.872448884858624e-05,
|
| 1668 |
+
"loss": 0.0096,
|
| 1669 |
+
"step": 2370
|
| 1670 |
+
},
|
| 1671 |
+
{
|
| 1672 |
+
"epoch": 49.583333333333336,
|
| 1673 |
+
"grad_norm": 0.13718049228191376,
|
| 1674 |
+
"learning_rate": 9.870586768353815e-05,
|
| 1675 |
+
"loss": 0.0087,
|
| 1676 |
+
"step": 2380
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 49.791666666666664,
|
| 1680 |
+
"grad_norm": 0.1327553540468216,
|
| 1681 |
+
"learning_rate": 9.868711335850764e-05,
|
| 1682 |
+
"loss": 0.0092,
|
| 1683 |
+
"step": 2390
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 50.0,
|
| 1687 |
+
"grad_norm": 0.260092556476593,
|
| 1688 |
+
"learning_rate": 9.866822592476833e-05,
|
| 1689 |
+
"loss": 0.0096,
|
| 1690 |
+
"step": 2400
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"epoch": 50.208333333333336,
|
| 1694 |
+
"grad_norm": 0.175159752368927,
|
| 1695 |
+
"learning_rate": 9.86492054339577e-05,
|
| 1696 |
+
"loss": 0.0091,
|
| 1697 |
+
"step": 2410
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 50.416666666666664,
|
| 1701 |
+
"grad_norm": 0.17031779885292053,
|
| 1702 |
+
"learning_rate": 9.863005193807711e-05,
|
| 1703 |
+
"loss": 0.0095,
|
| 1704 |
+
"step": 2420
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 50.625,
|
| 1708 |
+
"grad_norm": 0.1681128591299057,
|
| 1709 |
+
"learning_rate": 9.861076548949143e-05,
|
| 1710 |
+
"loss": 0.0107,
|
| 1711 |
+
"step": 2430
|
| 1712 |
+
},
|
| 1713 |
+
{
|
| 1714 |
+
"epoch": 50.833333333333336,
|
| 1715 |
+
"grad_norm": 0.17762134969234467,
|
| 1716 |
+
"learning_rate": 9.859134614092912e-05,
|
| 1717 |
+
"loss": 0.0102,
|
| 1718 |
+
"step": 2440
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 51.041666666666664,
|
| 1722 |
+
"grad_norm": 0.14008237421512604,
|
| 1723 |
+
"learning_rate": 9.857179394548191e-05,
|
| 1724 |
+
"loss": 0.0093,
|
| 1725 |
+
"step": 2450
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 51.25,
|
| 1729 |
+
"grad_norm": 0.1564144790172577,
|
| 1730 |
+
"learning_rate": 9.855210895660477e-05,
|
| 1731 |
+
"loss": 0.0096,
|
| 1732 |
+
"step": 2460
|
| 1733 |
+
},
|
| 1734 |
+
{
|
| 1735 |
+
"epoch": 51.458333333333336,
|
| 1736 |
+
"grad_norm": 0.13487133383750916,
|
| 1737 |
+
"learning_rate": 9.853229122811568e-05,
|
| 1738 |
+
"loss": 0.0099,
|
| 1739 |
+
"step": 2470
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 51.666666666666664,
|
| 1743 |
+
"grad_norm": 0.1658105105161667,
|
| 1744 |
+
"learning_rate": 9.851234081419559e-05,
|
| 1745 |
+
"loss": 0.0085,
|
| 1746 |
+
"step": 2480
|
| 1747 |
+
},
|
| 1748 |
+
{
|
| 1749 |
+
"epoch": 51.875,
|
| 1750 |
+
"grad_norm": 0.15481163561344147,
|
| 1751 |
+
"learning_rate": 9.849225776938814e-05,
|
| 1752 |
+
"loss": 0.0091,
|
| 1753 |
+
"step": 2490
|
| 1754 |
+
},
|
| 1755 |
+
{
|
| 1756 |
+
"epoch": 52.083333333333336,
|
| 1757 |
+
"grad_norm": 0.19152818620204926,
|
| 1758 |
+
"learning_rate": 9.847204214859964e-05,
|
| 1759 |
+
"loss": 0.0081,
|
| 1760 |
+
"step": 2500
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 52.291666666666664,
|
| 1764 |
+
"grad_norm": 0.14624345302581787,
|
| 1765 |
+
"learning_rate": 9.845169400709879e-05,
|
| 1766 |
+
"loss": 0.0087,
|
| 1767 |
+
"step": 2510
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 52.5,
|
| 1771 |
+
"grad_norm": 0.14260073006153107,
|
| 1772 |
+
"learning_rate": 9.843121340051664e-05,
|
| 1773 |
+
"loss": 0.0096,
|
| 1774 |
+
"step": 2520
|
| 1775 |
+
},
|
| 1776 |
+
{
|
| 1777 |
+
"epoch": 52.708333333333336,
|
| 1778 |
+
"grad_norm": 0.18242524564266205,
|
| 1779 |
+
"learning_rate": 9.841060038484641e-05,
|
| 1780 |
+
"loss": 0.0089,
|
| 1781 |
+
"step": 2530
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 52.916666666666664,
|
| 1785 |
+
"grad_norm": 0.135667085647583,
|
| 1786 |
+
"learning_rate": 9.838985501644328e-05,
|
| 1787 |
+
"loss": 0.0095,
|
| 1788 |
+
"step": 2540
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 53.125,
|
| 1792 |
+
"grad_norm": 0.11698679625988007,
|
| 1793 |
+
"learning_rate": 9.83689773520243e-05,
|
| 1794 |
+
"loss": 0.008,
|
| 1795 |
+
"step": 2550
|
| 1796 |
+
},
|
| 1797 |
+
{
|
| 1798 |
+
"epoch": 53.333333333333336,
|
| 1799 |
+
"grad_norm": 0.13887479901313782,
|
| 1800 |
+
"learning_rate": 9.834796744866819e-05,
|
| 1801 |
+
"loss": 0.0086,
|
| 1802 |
+
"step": 2560
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 53.541666666666664,
|
| 1806 |
+
"grad_norm": 0.12642112374305725,
|
| 1807 |
+
"learning_rate": 9.832682536381525e-05,
|
| 1808 |
+
"loss": 0.0092,
|
| 1809 |
+
"step": 2570
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 53.75,
|
| 1813 |
+
"grad_norm": 0.11770359426736832,
|
| 1814 |
+
"learning_rate": 9.830555115526711e-05,
|
| 1815 |
+
"loss": 0.009,
|
| 1816 |
+
"step": 2580
|
| 1817 |
+
},
|
| 1818 |
+
{
|
| 1819 |
+
"epoch": 53.958333333333336,
|
| 1820 |
+
"grad_norm": 0.10433600842952728,
|
| 1821 |
+
"learning_rate": 9.828414488118667e-05,
|
| 1822 |
+
"loss": 0.0084,
|
| 1823 |
+
"step": 2590
|
| 1824 |
+
},
|
| 1825 |
+
{
|
| 1826 |
+
"epoch": 54.166666666666664,
|
| 1827 |
+
"grad_norm": 0.12209464609622955,
|
| 1828 |
+
"learning_rate": 9.826260660009785e-05,
|
| 1829 |
+
"loss": 0.0094,
|
| 1830 |
+
"step": 2600
|
| 1831 |
+
},
|
| 1832 |
+
{
|
| 1833 |
+
"epoch": 54.375,
|
| 1834 |
+
"grad_norm": 0.1385062336921692,
|
| 1835 |
+
"learning_rate": 9.824093637088547e-05,
|
| 1836 |
+
"loss": 0.009,
|
| 1837 |
+
"step": 2610
|
| 1838 |
+
},
|
| 1839 |
+
{
|
| 1840 |
+
"epoch": 54.583333333333336,
|
| 1841 |
+
"grad_norm": 0.12731477618217468,
|
| 1842 |
+
"learning_rate": 9.821913425279514e-05,
|
| 1843 |
+
"loss": 0.0089,
|
| 1844 |
+
"step": 2620
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 54.791666666666664,
|
| 1848 |
+
"grad_norm": 0.15336956083774567,
|
| 1849 |
+
"learning_rate": 9.8197200305433e-05,
|
| 1850 |
+
"loss": 0.0085,
|
| 1851 |
+
"step": 2630
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 55.0,
|
| 1855 |
+
"grad_norm": 0.1945595145225525,
|
| 1856 |
+
"learning_rate": 9.817513458876564e-05,
|
| 1857 |
+
"loss": 0.0096,
|
| 1858 |
+
"step": 2640
|
| 1859 |
+
},
|
| 1860 |
+
{
|
| 1861 |
+
"epoch": 55.208333333333336,
|
| 1862 |
+
"grad_norm": 0.11767341941595078,
|
| 1863 |
+
"learning_rate": 9.815293716311987e-05,
|
| 1864 |
+
"loss": 0.0088,
|
| 1865 |
+
"step": 2650
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 55.416666666666664,
|
| 1869 |
+
"grad_norm": 0.14541785418987274,
|
| 1870 |
+
"learning_rate": 9.813060808918262e-05,
|
| 1871 |
+
"loss": 0.0092,
|
| 1872 |
+
"step": 2660
|
| 1873 |
+
},
|
| 1874 |
+
{
|
| 1875 |
+
"epoch": 55.625,
|
| 1876 |
+
"grad_norm": 0.16650767624378204,
|
| 1877 |
+
"learning_rate": 9.810814742800069e-05,
|
| 1878 |
+
"loss": 0.0097,
|
| 1879 |
+
"step": 2670
|
| 1880 |
+
},
|
| 1881 |
+
{
|
| 1882 |
+
"epoch": 55.833333333333336,
|
| 1883 |
+
"grad_norm": 0.1150713860988617,
|
| 1884 |
+
"learning_rate": 9.808555524098074e-05,
|
| 1885 |
+
"loss": 0.0086,
|
| 1886 |
+
"step": 2680
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 56.041666666666664,
|
| 1890 |
+
"grad_norm": 0.1088116317987442,
|
| 1891 |
+
"learning_rate": 9.806283158988887e-05,
|
| 1892 |
+
"loss": 0.0078,
|
| 1893 |
+
"step": 2690
|
| 1894 |
+
},
|
| 1895 |
+
{
|
| 1896 |
+
"epoch": 56.25,
|
| 1897 |
+
"grad_norm": 0.14631342887878418,
|
| 1898 |
+
"learning_rate": 9.803997653685072e-05,
|
| 1899 |
+
"loss": 0.0082,
|
| 1900 |
+
"step": 2700
|
| 1901 |
+
},
|
| 1902 |
+
{
|
| 1903 |
+
"epoch": 56.458333333333336,
|
| 1904 |
+
"grad_norm": 0.1343754678964615,
|
| 1905 |
+
"learning_rate": 9.801699014435112e-05,
|
| 1906 |
+
"loss": 0.008,
|
| 1907 |
+
"step": 2710
|
| 1908 |
+
},
|
| 1909 |
+
{
|
| 1910 |
+
"epoch": 56.666666666666664,
|
| 1911 |
+
"grad_norm": 0.13529370725154877,
|
| 1912 |
+
"learning_rate": 9.799387247523398e-05,
|
| 1913 |
+
"loss": 0.0085,
|
| 1914 |
+
"step": 2720
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 56.875,
|
| 1918 |
+
"grad_norm": 0.11147897690534592,
|
| 1919 |
+
"learning_rate": 9.797062359270215e-05,
|
| 1920 |
+
"loss": 0.0079,
|
| 1921 |
+
"step": 2730
|
| 1922 |
+
},
|
| 1923 |
+
{
|
| 1924 |
+
"epoch": 57.083333333333336,
|
| 1925 |
+
"grad_norm": 0.19471466541290283,
|
| 1926 |
+
"learning_rate": 9.794724356031715e-05,
|
| 1927 |
+
"loss": 0.0083,
|
| 1928 |
+
"step": 2740
|
| 1929 |
+
},
|
| 1930 |
+
{
|
| 1931 |
+
"epoch": 57.291666666666664,
|
| 1932 |
+
"grad_norm": 0.1863505244255066,
|
| 1933 |
+
"learning_rate": 9.792373244199913e-05,
|
| 1934 |
+
"loss": 0.0085,
|
| 1935 |
+
"step": 2750
|
| 1936 |
+
},
|
| 1937 |
+
{
|
| 1938 |
+
"epoch": 57.5,
|
| 1939 |
+
"grad_norm": 0.13585540652275085,
|
| 1940 |
+
"learning_rate": 9.790009030202658e-05,
|
| 1941 |
+
"loss": 0.0081,
|
| 1942 |
+
"step": 2760
|
| 1943 |
+
},
|
| 1944 |
+
{
|
| 1945 |
+
"epoch": 57.708333333333336,
|
| 1946 |
+
"grad_norm": 0.13892769813537598,
|
| 1947 |
+
"learning_rate": 9.78763172050362e-05,
|
| 1948 |
+
"loss": 0.0086,
|
| 1949 |
+
"step": 2770
|
| 1950 |
+
},
|
| 1951 |
+
{
|
| 1952 |
+
"epoch": 57.916666666666664,
|
| 1953 |
+
"grad_norm": 0.15619535744190216,
|
| 1954 |
+
"learning_rate": 9.785241321602274e-05,
|
| 1955 |
+
"loss": 0.0078,
|
| 1956 |
+
"step": 2780
|
| 1957 |
+
},
|
| 1958 |
+
{
|
| 1959 |
+
"epoch": 58.125,
|
| 1960 |
+
"grad_norm": 0.17480793595314026,
|
| 1961 |
+
"learning_rate": 9.782837840033879e-05,
|
| 1962 |
+
"loss": 0.0088,
|
| 1963 |
+
"step": 2790
|
| 1964 |
+
},
|
| 1965 |
+
{
|
| 1966 |
+
"epoch": 58.333333333333336,
|
| 1967 |
+
"grad_norm": 0.19116300344467163,
|
| 1968 |
+
"learning_rate": 9.780421282369461e-05,
|
| 1969 |
+
"loss": 0.0091,
|
| 1970 |
+
"step": 2800
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 58.541666666666664,
|
| 1974 |
+
"grad_norm": 0.2017354965209961,
|
| 1975 |
+
"learning_rate": 9.777991655215797e-05,
|
| 1976 |
+
"loss": 0.0082,
|
| 1977 |
+
"step": 2810
|
| 1978 |
+
},
|
| 1979 |
+
{
|
| 1980 |
+
"epoch": 58.75,
|
| 1981 |
+
"grad_norm": 0.17326346039772034,
|
| 1982 |
+
"learning_rate": 9.775548965215394e-05,
|
| 1983 |
+
"loss": 0.0088,
|
| 1984 |
+
"step": 2820
|
| 1985 |
+
},
|
| 1986 |
+
{
|
| 1987 |
+
"epoch": 58.958333333333336,
|
| 1988 |
+
"grad_norm": 0.17388294637203217,
|
| 1989 |
+
"learning_rate": 9.773093219046474e-05,
|
| 1990 |
+
"loss": 0.0084,
|
| 1991 |
+
"step": 2830
|
| 1992 |
+
},
|
| 1993 |
+
{
|
| 1994 |
+
"epoch": 59.166666666666664,
|
| 1995 |
+
"grad_norm": 0.1788823902606964,
|
| 1996 |
+
"learning_rate": 9.770624423422954e-05,
|
| 1997 |
+
"loss": 0.0084,
|
| 1998 |
+
"step": 2840
|
| 1999 |
+
},
|
| 2000 |
+
{
|
| 2001 |
+
"epoch": 59.375,
|
| 2002 |
+
"grad_norm": 0.11936558783054352,
|
| 2003 |
+
"learning_rate": 9.768142585094426e-05,
|
| 2004 |
+
"loss": 0.0091,
|
| 2005 |
+
"step": 2850
|
| 2006 |
+
},
|
| 2007 |
+
{
|
| 2008 |
+
"epoch": 59.583333333333336,
|
| 2009 |
+
"grad_norm": 0.14019201695919037,
|
| 2010 |
+
"learning_rate": 9.765647710846142e-05,
|
| 2011 |
+
"loss": 0.0092,
|
| 2012 |
+
"step": 2860
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 59.791666666666664,
|
| 2016 |
+
"grad_norm": 0.12400297075510025,
|
| 2017 |
+
"learning_rate": 9.763139807498991e-05,
|
| 2018 |
+
"loss": 0.0084,
|
| 2019 |
+
"step": 2870
|
| 2020 |
+
},
|
| 2021 |
+
{
|
| 2022 |
+
"epoch": 60.0,
|
| 2023 |
+
"grad_norm": 0.1594945192337036,
|
| 2024 |
+
"learning_rate": 9.760618881909487e-05,
|
| 2025 |
+
"loss": 0.0083,
|
| 2026 |
+
"step": 2880
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"epoch": 60.208333333333336,
|
| 2030 |
+
"grad_norm": 0.17475134134292603,
|
| 2031 |
+
"learning_rate": 9.758084940969744e-05,
|
| 2032 |
+
"loss": 0.0093,
|
| 2033 |
+
"step": 2890
|
| 2034 |
+
},
|
| 2035 |
+
{
|
| 2036 |
+
"epoch": 60.416666666666664,
|
| 2037 |
+
"grad_norm": 0.125895693898201,
|
| 2038 |
+
"learning_rate": 9.755537991607459e-05,
|
| 2039 |
+
"loss": 0.0087,
|
| 2040 |
+
"step": 2900
|
| 2041 |
+
},
|
| 2042 |
+
{
|
| 2043 |
+
"epoch": 60.625,
|
| 2044 |
+
"grad_norm": 0.14806149899959564,
|
| 2045 |
+
"learning_rate": 9.752978040785895e-05,
|
| 2046 |
+
"loss": 0.0072,
|
| 2047 |
+
"step": 2910
|
| 2048 |
+
},
|
| 2049 |
+
{
|
| 2050 |
+
"epoch": 60.833333333333336,
|
| 2051 |
+
"grad_norm": 0.14935727417469025,
|
| 2052 |
+
"learning_rate": 9.750405095503859e-05,
|
| 2053 |
+
"loss": 0.0084,
|
| 2054 |
+
"step": 2920
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"epoch": 61.041666666666664,
|
| 2058 |
+
"grad_norm": 0.12621425092220306,
|
| 2059 |
+
"learning_rate": 9.747819162795686e-05,
|
| 2060 |
+
"loss": 0.0076,
|
| 2061 |
+
"step": 2930
|
| 2062 |
+
},
|
| 2063 |
+
{
|
| 2064 |
+
"epoch": 61.25,
|
| 2065 |
+
"grad_norm": 0.15602892637252808,
|
| 2066 |
+
"learning_rate": 9.745220249731217e-05,
|
| 2067 |
+
"loss": 0.0084,
|
| 2068 |
+
"step": 2940
|
| 2069 |
+
},
|
| 2070 |
+
{
|
| 2071 |
+
"epoch": 61.458333333333336,
|
| 2072 |
+
"grad_norm": 0.12207546830177307,
|
| 2073 |
+
"learning_rate": 9.742608363415781e-05,
|
| 2074 |
+
"loss": 0.0084,
|
| 2075 |
+
"step": 2950
|
| 2076 |
+
},
|
| 2077 |
+
{
|
| 2078 |
+
"epoch": 61.666666666666664,
|
| 2079 |
+
"grad_norm": 0.1394585818052292,
|
| 2080 |
+
"learning_rate": 9.739983510990176e-05,
|
| 2081 |
+
"loss": 0.0077,
|
| 2082 |
+
"step": 2960
|
| 2083 |
+
},
|
| 2084 |
+
{
|
| 2085 |
+
"epoch": 61.875,
|
| 2086 |
+
"grad_norm": 0.13204297423362732,
|
| 2087 |
+
"learning_rate": 9.737345699630647e-05,
|
| 2088 |
+
"loss": 0.0079,
|
| 2089 |
+
"step": 2970
|
| 2090 |
+
},
|
| 2091 |
+
{
|
| 2092 |
+
"epoch": 62.083333333333336,
|
| 2093 |
+
"grad_norm": 0.13679321110248566,
|
| 2094 |
+
"learning_rate": 9.734694936548869e-05,
|
| 2095 |
+
"loss": 0.0074,
|
| 2096 |
+
"step": 2980
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"epoch": 62.291666666666664,
|
| 2100 |
+
"grad_norm": 0.11424277722835541,
|
| 2101 |
+
"learning_rate": 9.732031228991932e-05,
|
| 2102 |
+
"loss": 0.0086,
|
| 2103 |
+
"step": 2990
|
| 2104 |
+
},
|
| 2105 |
+
{
|
| 2106 |
+
"epoch": 62.5,
|
| 2107 |
+
"grad_norm": 0.112228162586689,
|
| 2108 |
+
"learning_rate": 9.729354584242302e-05,
|
| 2109 |
+
"loss": 0.0087,
|
| 2110 |
+
"step": 3000
|
| 2111 |
+
},
|
| 2112 |
+
{
|
| 2113 |
+
"epoch": 62.708333333333336,
|
| 2114 |
+
"grad_norm": 0.15181872248649597,
|
| 2115 |
+
"learning_rate": 9.726665009617832e-05,
|
| 2116 |
+
"loss": 0.0081,
|
| 2117 |
+
"step": 3010
|
| 2118 |
+
},
|
| 2119 |
+
{
|
| 2120 |
+
"epoch": 62.916666666666664,
|
| 2121 |
+
"grad_norm": 0.1584879606962204,
|
| 2122 |
+
"learning_rate": 9.723962512471714e-05,
|
| 2123 |
+
"loss": 0.0083,
|
| 2124 |
+
"step": 3020
|
| 2125 |
+
},
|
| 2126 |
+
{
|
| 2127 |
+
"epoch": 63.125,
|
| 2128 |
+
"grad_norm": 0.15587449073791504,
|
| 2129 |
+
"learning_rate": 9.72124710019247e-05,
|
| 2130 |
+
"loss": 0.0084,
|
| 2131 |
+
"step": 3030
|
| 2132 |
+
},
|
| 2133 |
+
{
|
| 2134 |
+
"epoch": 63.333333333333336,
|
| 2135 |
+
"grad_norm": 0.13749879598617554,
|
| 2136 |
+
"learning_rate": 9.718518780203934e-05,
|
| 2137 |
+
"loss": 0.0087,
|
| 2138 |
+
"step": 3040
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 63.541666666666664,
|
| 2142 |
+
"grad_norm": 0.15750397741794586,
|
| 2143 |
+
"learning_rate": 9.715777559965228e-05,
|
| 2144 |
+
"loss": 0.0085,
|
| 2145 |
+
"step": 3050
|
| 2146 |
+
},
|
| 2147 |
+
{
|
| 2148 |
+
"epoch": 63.75,
|
| 2149 |
+
"grad_norm": 0.16457325220108032,
|
| 2150 |
+
"learning_rate": 9.713023446970746e-05,
|
| 2151 |
+
"loss": 0.008,
|
| 2152 |
+
"step": 3060
|
| 2153 |
+
},
|
| 2154 |
+
{
|
| 2155 |
+
"epoch": 63.958333333333336,
|
| 2156 |
+
"grad_norm": 0.14457859098911285,
|
| 2157 |
+
"learning_rate": 9.710256448750126e-05,
|
| 2158 |
+
"loss": 0.008,
|
| 2159 |
+
"step": 3070
|
| 2160 |
+
},
|
| 2161 |
+
{
|
| 2162 |
+
"epoch": 64.16666666666667,
|
| 2163 |
+
"grad_norm": 0.21400192379951477,
|
| 2164 |
+
"learning_rate": 9.707476572868235e-05,
|
| 2165 |
+
"loss": 0.0087,
|
| 2166 |
+
"step": 3080
|
| 2167 |
+
},
|
| 2168 |
+
{
|
| 2169 |
+
"epoch": 64.375,
|
| 2170 |
+
"grad_norm": 0.176253542304039,
|
| 2171 |
+
"learning_rate": 9.704683826925149e-05,
|
| 2172 |
+
"loss": 0.0082,
|
| 2173 |
+
"step": 3090
|
| 2174 |
+
},
|
| 2175 |
+
{
|
| 2176 |
+
"epoch": 64.58333333333333,
|
| 2177 |
+
"grad_norm": 0.14613905549049377,
|
| 2178 |
+
"learning_rate": 9.701878218556129e-05,
|
| 2179 |
+
"loss": 0.0091,
|
| 2180 |
+
"step": 3100
|
| 2181 |
+
},
|
| 2182 |
+
{
|
| 2183 |
+
"epoch": 64.79166666666667,
|
| 2184 |
+
"grad_norm": 0.12556883692741394,
|
| 2185 |
+
"learning_rate": 9.699059755431598e-05,
|
| 2186 |
+
"loss": 0.0086,
|
| 2187 |
+
"step": 3110
|
| 2188 |
+
},
|
| 2189 |
+
{
|
| 2190 |
+
"epoch": 65.0,
|
| 2191 |
+
"grad_norm": 0.12506556510925293,
|
| 2192 |
+
"learning_rate": 9.696228445257132e-05,
|
| 2193 |
+
"loss": 0.0078,
|
| 2194 |
+
"step": 3120
|
| 2195 |
+
},
|
| 2196 |
+
{
|
| 2197 |
+
"epoch": 65.20833333333333,
|
| 2198 |
+
"grad_norm": 0.1324303150177002,
|
| 2199 |
+
"learning_rate": 9.693384295773419e-05,
|
| 2200 |
+
"loss": 0.0076,
|
| 2201 |
+
"step": 3130
|
| 2202 |
+
},
|
| 2203 |
+
{
|
| 2204 |
+
"epoch": 65.41666666666667,
|
| 2205 |
+
"grad_norm": 0.14260755479335785,
|
| 2206 |
+
"learning_rate": 9.690527314756259e-05,
|
| 2207 |
+
"loss": 0.0075,
|
| 2208 |
+
"step": 3140
|
| 2209 |
+
},
|
| 2210 |
+
{
|
| 2211 |
+
"epoch": 65.625,
|
| 2212 |
+
"grad_norm": 0.17024897038936615,
|
| 2213 |
+
"learning_rate": 9.687657510016527e-05,
|
| 2214 |
+
"loss": 0.0073,
|
| 2215 |
+
"step": 3150
|
| 2216 |
+
},
|
| 2217 |
+
{
|
| 2218 |
+
"epoch": 65.83333333333333,
|
| 2219 |
+
"grad_norm": 0.12163723260164261,
|
| 2220 |
+
"learning_rate": 9.684774889400161e-05,
|
| 2221 |
+
"loss": 0.0084,
|
| 2222 |
+
"step": 3160
|
| 2223 |
+
},
|
| 2224 |
+
{
|
| 2225 |
+
"epoch": 66.04166666666667,
|
| 2226 |
+
"grad_norm": 0.17265737056732178,
|
| 2227 |
+
"learning_rate": 9.681879460788135e-05,
|
| 2228 |
+
"loss": 0.0085,
|
| 2229 |
+
"step": 3170
|
| 2230 |
+
},
|
| 2231 |
+
{
|
| 2232 |
+
"epoch": 66.25,
|
| 2233 |
+
"grad_norm": 0.14137807488441467,
|
| 2234 |
+
"learning_rate": 9.67897123209644e-05,
|
| 2235 |
+
"loss": 0.0087,
|
| 2236 |
+
"step": 3180
|
| 2237 |
+
},
|
| 2238 |
+
{
|
| 2239 |
+
"epoch": 66.45833333333333,
|
| 2240 |
+
"grad_norm": 0.11415492743253708,
|
| 2241 |
+
"learning_rate": 9.676050211276062e-05,
|
| 2242 |
+
"loss": 0.0084,
|
| 2243 |
+
"step": 3190
|
| 2244 |
+
},
|
| 2245 |
+
{
|
| 2246 |
+
"epoch": 66.66666666666667,
|
| 2247 |
+
"grad_norm": 0.10539334267377853,
|
| 2248 |
+
"learning_rate": 9.673116406312962e-05,
|
| 2249 |
+
"loss": 0.0085,
|
| 2250 |
+
"step": 3200
|
| 2251 |
+
},
|
| 2252 |
+
{
|
| 2253 |
+
"epoch": 66.875,
|
| 2254 |
+
"grad_norm": 0.1746184527873993,
|
| 2255 |
+
"learning_rate": 9.67016982522805e-05,
|
| 2256 |
+
"loss": 0.0079,
|
| 2257 |
+
"step": 3210
|
| 2258 |
+
},
|
| 2259 |
+
{
|
| 2260 |
+
"epoch": 67.08333333333333,
|
| 2261 |
+
"grad_norm": 0.15164965391159058,
|
| 2262 |
+
"learning_rate": 9.667210476077164e-05,
|
| 2263 |
+
"loss": 0.0078,
|
| 2264 |
+
"step": 3220
|
| 2265 |
+
},
|
| 2266 |
+
{
|
| 2267 |
+
"epoch": 67.29166666666667,
|
| 2268 |
+
"grad_norm": 0.1245863139629364,
|
| 2269 |
+
"learning_rate": 9.664238366951055e-05,
|
| 2270 |
+
"loss": 0.0083,
|
| 2271 |
+
"step": 3230
|
| 2272 |
+
},
|
| 2273 |
+
{
|
| 2274 |
+
"epoch": 67.5,
|
| 2275 |
+
"grad_norm": 0.13058176636695862,
|
| 2276 |
+
"learning_rate": 9.661253505975355e-05,
|
| 2277 |
+
"loss": 0.008,
|
| 2278 |
+
"step": 3240
|
| 2279 |
+
},
|
| 2280 |
+
{
|
| 2281 |
+
"epoch": 67.70833333333333,
|
| 2282 |
+
"grad_norm": 0.1412813365459442,
|
| 2283 |
+
"learning_rate": 9.658255901310557e-05,
|
| 2284 |
+
"loss": 0.0084,
|
| 2285 |
+
"step": 3250
|
| 2286 |
+
},
|
| 2287 |
+
{
|
| 2288 |
+
"epoch": 67.91666666666667,
|
| 2289 |
+
"grad_norm": 0.13035815954208374,
|
| 2290 |
+
"learning_rate": 9.655245561152e-05,
|
| 2291 |
+
"loss": 0.0082,
|
| 2292 |
+
"step": 3260
|
| 2293 |
+
},
|
| 2294 |
+
{
|
| 2295 |
+
"epoch": 68.125,
|
| 2296 |
+
"grad_norm": 0.14681102335453033,
|
| 2297 |
+
"learning_rate": 9.65222249372984e-05,
|
| 2298 |
+
"loss": 0.0078,
|
| 2299 |
+
"step": 3270
|
| 2300 |
+
},
|
| 2301 |
+
{
|
| 2302 |
+
"epoch": 68.33333333333333,
|
| 2303 |
+
"grad_norm": 0.12792235612869263,
|
| 2304 |
+
"learning_rate": 9.649186707309026e-05,
|
| 2305 |
+
"loss": 0.0082,
|
| 2306 |
+
"step": 3280
|
| 2307 |
+
},
|
| 2308 |
+
{
|
| 2309 |
+
"epoch": 68.54166666666667,
|
| 2310 |
+
"grad_norm": 0.15287558734416962,
|
| 2311 |
+
"learning_rate": 9.646138210189283e-05,
|
| 2312 |
+
"loss": 0.0079,
|
| 2313 |
+
"step": 3290
|
| 2314 |
+
},
|
| 2315 |
+
{
|
| 2316 |
+
"epoch": 68.75,
|
| 2317 |
+
"grad_norm": 0.1277853548526764,
|
| 2318 |
+
"learning_rate": 9.643077010705087e-05,
|
| 2319 |
+
"loss": 0.008,
|
| 2320 |
+
"step": 3300
|
| 2321 |
+
},
|
| 2322 |
+
{
|
| 2323 |
+
"epoch": 68.95833333333333,
|
| 2324 |
+
"grad_norm": 0.10948234051465988,
|
| 2325 |
+
"learning_rate": 9.640003117225637e-05,
|
| 2326 |
+
"loss": 0.0081,
|
| 2327 |
+
"step": 3310
|
| 2328 |
+
},
|
| 2329 |
+
{
|
| 2330 |
+
"epoch": 69.16666666666667,
|
| 2331 |
+
"grad_norm": 0.16888520121574402,
|
| 2332 |
+
"learning_rate": 9.636916538154846e-05,
|
| 2333 |
+
"loss": 0.008,
|
| 2334 |
+
"step": 3320
|
| 2335 |
+
},
|
| 2336 |
+
{
|
| 2337 |
+
"epoch": 69.375,
|
| 2338 |
+
"grad_norm": 0.1860867738723755,
|
| 2339 |
+
"learning_rate": 9.633817281931296e-05,
|
| 2340 |
+
"loss": 0.0081,
|
| 2341 |
+
"step": 3330
|
| 2342 |
+
},
|
| 2343 |
+
{
|
| 2344 |
+
"epoch": 69.58333333333333,
|
| 2345 |
+
"grad_norm": 0.12217903882265091,
|
| 2346 |
+
"learning_rate": 9.630705357028242e-05,
|
| 2347 |
+
"loss": 0.008,
|
| 2348 |
+
"step": 3340
|
| 2349 |
+
},
|
| 2350 |
+
{
|
| 2351 |
+
"epoch": 69.79166666666667,
|
| 2352 |
+
"grad_norm": 0.14392343163490295,
|
| 2353 |
+
"learning_rate": 9.627580771953563e-05,
|
| 2354 |
+
"loss": 0.0084,
|
| 2355 |
+
"step": 3350
|
| 2356 |
+
},
|
| 2357 |
+
{
|
| 2358 |
+
"epoch": 70.0,
|
| 2359 |
+
"grad_norm": 0.2517831027507782,
|
| 2360 |
+
"learning_rate": 9.624443535249759e-05,
|
| 2361 |
+
"loss": 0.0091,
|
| 2362 |
+
"step": 3360
|
| 2363 |
+
},
|
| 2364 |
+
{
|
| 2365 |
+
"epoch": 70.20833333333333,
|
| 2366 |
+
"grad_norm": 0.1237884908914566,
|
| 2367 |
+
"learning_rate": 9.621293655493913e-05,
|
| 2368 |
+
"loss": 0.0086,
|
| 2369 |
+
"step": 3370
|
| 2370 |
+
},
|
| 2371 |
+
{
|
| 2372 |
+
"epoch": 70.41666666666667,
|
| 2373 |
+
"grad_norm": 0.1395934820175171,
|
| 2374 |
+
"learning_rate": 9.618131141297675e-05,
|
| 2375 |
+
"loss": 0.0091,
|
| 2376 |
+
"step": 3380
|
| 2377 |
+
},
|
| 2378 |
+
{
|
| 2379 |
+
"epoch": 70.625,
|
| 2380 |
+
"grad_norm": 0.15799179673194885,
|
| 2381 |
+
"learning_rate": 9.614956001307242e-05,
|
| 2382 |
+
"loss": 0.0098,
|
| 2383 |
+
"step": 3390
|
| 2384 |
+
},
|
| 2385 |
+
{
|
| 2386 |
+
"epoch": 70.83333333333333,
|
| 2387 |
+
"grad_norm": 0.14630959928035736,
|
| 2388 |
+
"learning_rate": 9.611768244203321e-05,
|
| 2389 |
+
"loss": 0.0082,
|
| 2390 |
+
"step": 3400
|
| 2391 |
+
},
|
| 2392 |
+
{
|
| 2393 |
+
"epoch": 71.04166666666667,
|
| 2394 |
+
"grad_norm": 0.19569124281406403,
|
| 2395 |
+
"learning_rate": 9.60856787870112e-05,
|
| 2396 |
+
"loss": 0.0089,
|
| 2397 |
+
"step": 3410
|
| 2398 |
+
},
|
| 2399 |
+
{
|
| 2400 |
+
"epoch": 71.25,
|
| 2401 |
+
"grad_norm": 0.15080992877483368,
|
| 2402 |
+
"learning_rate": 9.605354913550318e-05,
|
| 2403 |
+
"loss": 0.0081,
|
| 2404 |
+
"step": 3420
|
| 2405 |
+
},
|
| 2406 |
+
{
|
| 2407 |
+
"epoch": 71.45833333333333,
|
| 2408 |
+
"grad_norm": 0.12431894987821579,
|
| 2409 |
+
"learning_rate": 9.602129357535037e-05,
|
| 2410 |
+
"loss": 0.0077,
|
| 2411 |
+
"step": 3430
|
| 2412 |
+
},
|
| 2413 |
+
{
|
| 2414 |
+
"epoch": 71.66666666666667,
|
| 2415 |
+
"grad_norm": 0.1317983716726303,
|
| 2416 |
+
"learning_rate": 9.598891219473825e-05,
|
| 2417 |
+
"loss": 0.0081,
|
| 2418 |
+
"step": 3440
|
| 2419 |
+
},
|
| 2420 |
+
{
|
| 2421 |
+
"epoch": 71.875,
|
| 2422 |
+
"grad_norm": 0.1271415799856186,
|
| 2423 |
+
"learning_rate": 9.595640508219625e-05,
|
| 2424 |
+
"loss": 0.0079,
|
| 2425 |
+
"step": 3450
|
| 2426 |
+
},
|
| 2427 |
+
{
|
| 2428 |
+
"epoch": 72.08333333333333,
|
| 2429 |
+
"grad_norm": 0.12340828031301498,
|
| 2430 |
+
"learning_rate": 9.592377232659761e-05,
|
| 2431 |
+
"loss": 0.0073,
|
| 2432 |
+
"step": 3460
|
| 2433 |
+
},
|
| 2434 |
+
{
|
| 2435 |
+
"epoch": 72.29166666666667,
|
| 2436 |
+
"grad_norm": 0.16125109791755676,
|
| 2437 |
+
"learning_rate": 9.589101401715904e-05,
|
| 2438 |
+
"loss": 0.0082,
|
| 2439 |
+
"step": 3470
|
| 2440 |
+
},
|
| 2441 |
+
{
|
| 2442 |
+
"epoch": 72.5,
|
| 2443 |
+
"grad_norm": 0.13908740878105164,
|
| 2444 |
+
"learning_rate": 9.585813024344045e-05,
|
| 2445 |
+
"loss": 0.0081,
|
| 2446 |
+
"step": 3480
|
| 2447 |
+
},
|
| 2448 |
+
{
|
| 2449 |
+
"epoch": 72.70833333333333,
|
| 2450 |
+
"grad_norm": 0.12069405615329742,
|
| 2451 |
+
"learning_rate": 9.58251210953449e-05,
|
| 2452 |
+
"loss": 0.0072,
|
| 2453 |
+
"step": 3490
|
| 2454 |
+
},
|
| 2455 |
+
{
|
| 2456 |
+
"epoch": 72.91666666666667,
|
| 2457 |
+
"grad_norm": 0.15209130942821503,
|
| 2458 |
+
"learning_rate": 9.579198666311809e-05,
|
| 2459 |
+
"loss": 0.0082,
|
| 2460 |
+
"step": 3500
|
| 2461 |
+
},
|
| 2462 |
+
{
|
| 2463 |
+
"epoch": 73.125,
|
| 2464 |
+
"grad_norm": 0.12193018198013306,
|
| 2465 |
+
"learning_rate": 9.575872703734832e-05,
|
| 2466 |
+
"loss": 0.0078,
|
| 2467 |
+
"step": 3510
|
| 2468 |
+
},
|
| 2469 |
+
{
|
| 2470 |
+
"epoch": 73.33333333333333,
|
| 2471 |
+
"grad_norm": 0.17105849087238312,
|
| 2472 |
+
"learning_rate": 9.572534230896611e-05,
|
| 2473 |
+
"loss": 0.0085,
|
| 2474 |
+
"step": 3520
|
| 2475 |
+
},
|
| 2476 |
+
{
|
| 2477 |
+
"epoch": 73.54166666666667,
|
| 2478 |
+
"grad_norm": 0.1370059996843338,
|
| 2479 |
+
"learning_rate": 9.569183256924403e-05,
|
| 2480 |
+
"loss": 0.0085,
|
| 2481 |
+
"step": 3530
|
| 2482 |
+
},
|
| 2483 |
+
{
|
| 2484 |
+
"epoch": 73.75,
|
| 2485 |
+
"grad_norm": 0.13424254953861237,
|
| 2486 |
+
"learning_rate": 9.565819790979646e-05,
|
| 2487 |
+
"loss": 0.0085,
|
| 2488 |
+
"step": 3540
|
| 2489 |
+
},
|
| 2490 |
+
{
|
| 2491 |
+
"epoch": 73.95833333333333,
|
| 2492 |
+
"grad_norm": 0.12105348706245422,
|
| 2493 |
+
"learning_rate": 9.562443842257925e-05,
|
| 2494 |
+
"loss": 0.0077,
|
| 2495 |
+
"step": 3550
|
| 2496 |
+
},
|
| 2497 |
+
{
|
| 2498 |
+
"epoch": 74.16666666666667,
|
| 2499 |
+
"grad_norm": 0.14536826312541962,
|
| 2500 |
+
"learning_rate": 9.559055419988956e-05,
|
| 2501 |
+
"loss": 0.0087,
|
| 2502 |
+
"step": 3560
|
| 2503 |
+
},
|
| 2504 |
+
{
|
| 2505 |
+
"epoch": 74.375,
|
| 2506 |
+
"grad_norm": 0.1609853357076645,
|
| 2507 |
+
"learning_rate": 9.555654533436557e-05,
|
| 2508 |
+
"loss": 0.0092,
|
| 2509 |
+
"step": 3570
|
| 2510 |
+
},
|
| 2511 |
+
{
|
| 2512 |
+
"epoch": 74.58333333333333,
|
| 2513 |
+
"grad_norm": 0.15583093464374542,
|
| 2514 |
+
"learning_rate": 9.552241191898621e-05,
|
| 2515 |
+
"loss": 0.0084,
|
| 2516 |
+
"step": 3580
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 74.79166666666667,
|
| 2520 |
+
"grad_norm": 0.13674955070018768,
|
| 2521 |
+
"learning_rate": 9.548815404707092e-05,
|
| 2522 |
+
"loss": 0.0078,
|
| 2523 |
+
"step": 3590
|
| 2524 |
+
},
|
| 2525 |
+
{
|
| 2526 |
+
"epoch": 75.0,
|
| 2527 |
+
"grad_norm": 0.1659046858549118,
|
| 2528 |
+
"learning_rate": 9.545377181227942e-05,
|
| 2529 |
+
"loss": 0.008,
|
| 2530 |
+
"step": 3600
|
| 2531 |
+
},
|
| 2532 |
+
{
|
| 2533 |
+
"epoch": 75.20833333333333,
|
| 2534 |
+
"grad_norm": 0.09946411848068237,
|
| 2535 |
+
"learning_rate": 9.541926530861145e-05,
|
| 2536 |
+
"loss": 0.0076,
|
| 2537 |
+
"step": 3610
|
| 2538 |
+
},
|
| 2539 |
+
{
|
| 2540 |
+
"epoch": 75.41666666666667,
|
| 2541 |
+
"grad_norm": 0.13440348207950592,
|
| 2542 |
+
"learning_rate": 9.538463463040645e-05,
|
| 2543 |
+
"loss": 0.0071,
|
| 2544 |
+
"step": 3620
|
| 2545 |
+
},
|
| 2546 |
+
{
|
| 2547 |
+
"epoch": 75.625,
|
| 2548 |
+
"grad_norm": 0.13562451303005219,
|
| 2549 |
+
"learning_rate": 9.534987987234337e-05,
|
| 2550 |
+
"loss": 0.0072,
|
| 2551 |
+
"step": 3630
|
| 2552 |
+
},
|
| 2553 |
+
{
|
| 2554 |
+
"epoch": 75.83333333333333,
|
| 2555 |
+
"grad_norm": 0.12133369594812393,
|
| 2556 |
+
"learning_rate": 9.53150011294404e-05,
|
| 2557 |
+
"loss": 0.0078,
|
| 2558 |
+
"step": 3640
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 76.04166666666667,
|
| 2562 |
+
"grad_norm": 0.11865601688623428,
|
| 2563 |
+
"learning_rate": 9.527999849705471e-05,
|
| 2564 |
+
"loss": 0.0073,
|
| 2565 |
+
"step": 3650
|
| 2566 |
+
},
|
| 2567 |
+
{
|
| 2568 |
+
"epoch": 76.25,
|
| 2569 |
+
"grad_norm": 0.13767482340335846,
|
| 2570 |
+
"learning_rate": 9.524487207088213e-05,
|
| 2571 |
+
"loss": 0.008,
|
| 2572 |
+
"step": 3660
|
| 2573 |
+
},
|
| 2574 |
+
{
|
| 2575 |
+
"epoch": 76.45833333333333,
|
| 2576 |
+
"grad_norm": 0.12942279875278473,
|
| 2577 |
+
"learning_rate": 9.520962194695698e-05,
|
| 2578 |
+
"loss": 0.0082,
|
| 2579 |
+
"step": 3670
|
| 2580 |
+
},
|
| 2581 |
+
{
|
| 2582 |
+
"epoch": 76.66666666666667,
|
| 2583 |
+
"grad_norm": 0.14708350598812103,
|
| 2584 |
+
"learning_rate": 9.517424822165175e-05,
|
| 2585 |
+
"loss": 0.0089,
|
| 2586 |
+
"step": 3680
|
| 2587 |
+
},
|
| 2588 |
+
{
|
| 2589 |
+
"epoch": 76.875,
|
| 2590 |
+
"grad_norm": 0.12420503050088882,
|
| 2591 |
+
"learning_rate": 9.513875099167685e-05,
|
| 2592 |
+
"loss": 0.0079,
|
| 2593 |
+
"step": 3690
|
| 2594 |
+
},
|
| 2595 |
+
{
|
| 2596 |
+
"epoch": 77.08333333333333,
|
| 2597 |
+
"grad_norm": 0.14663736522197723,
|
| 2598 |
+
"learning_rate": 9.510313035408035e-05,
|
| 2599 |
+
"loss": 0.0084,
|
| 2600 |
+
"step": 3700
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"epoch": 77.29166666666667,
|
| 2604 |
+
"grad_norm": 0.1613157093524933,
|
| 2605 |
+
"learning_rate": 9.506738640624775e-05,
|
| 2606 |
+
"loss": 0.0085,
|
| 2607 |
+
"step": 3710
|
| 2608 |
+
},
|
| 2609 |
+
{
|
| 2610 |
+
"epoch": 77.5,
|
| 2611 |
+
"grad_norm": 0.15745188295841217,
|
| 2612 |
+
"learning_rate": 9.50315192459016e-05,
|
| 2613 |
+
"loss": 0.0084,
|
| 2614 |
+
"step": 3720
|
| 2615 |
+
},
|
| 2616 |
+
{
|
| 2617 |
+
"epoch": 77.70833333333333,
|
| 2618 |
+
"grad_norm": 0.1454235464334488,
|
| 2619 |
+
"learning_rate": 9.499552897110136e-05,
|
| 2620 |
+
"loss": 0.0081,
|
| 2621 |
+
"step": 3730
|
| 2622 |
+
},
|
| 2623 |
+
{
|
| 2624 |
+
"epoch": 77.91666666666667,
|
| 2625 |
+
"grad_norm": 0.13637305796146393,
|
| 2626 |
+
"learning_rate": 9.495941568024304e-05,
|
| 2627 |
+
"loss": 0.0078,
|
| 2628 |
+
"step": 3740
|
| 2629 |
+
},
|
| 2630 |
+
{
|
| 2631 |
+
"epoch": 78.125,
|
| 2632 |
+
"grad_norm": 0.16548554599285126,
|
| 2633 |
+
"learning_rate": 9.492317947205904e-05,
|
| 2634 |
+
"loss": 0.0078,
|
| 2635 |
+
"step": 3750
|
| 2636 |
+
},
|
| 2637 |
+
{
|
| 2638 |
+
"epoch": 78.33333333333333,
|
| 2639 |
+
"grad_norm": 0.11904530972242355,
|
| 2640 |
+
"learning_rate": 9.488682044561775e-05,
|
| 2641 |
+
"loss": 0.008,
|
| 2642 |
+
"step": 3760
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"epoch": 78.54166666666667,
|
| 2646 |
+
"grad_norm": 0.13436388969421387,
|
| 2647 |
+
"learning_rate": 9.485033870032335e-05,
|
| 2648 |
+
"loss": 0.0078,
|
| 2649 |
+
"step": 3770
|
| 2650 |
+
},
|
| 2651 |
+
{
|
| 2652 |
+
"epoch": 78.75,
|
| 2653 |
+
"grad_norm": 0.13463658094406128,
|
| 2654 |
+
"learning_rate": 9.481373433591556e-05,
|
| 2655 |
+
"loss": 0.008,
|
| 2656 |
+
"step": 3780
|
| 2657 |
+
},
|
| 2658 |
+
{
|
| 2659 |
+
"epoch": 78.95833333333333,
|
| 2660 |
+
"grad_norm": 0.12881360948085785,
|
| 2661 |
+
"learning_rate": 9.47770074524693e-05,
|
| 2662 |
+
"loss": 0.0074,
|
| 2663 |
+
"step": 3790
|
| 2664 |
+
},
|
| 2665 |
+
{
|
| 2666 |
+
"epoch": 79.16666666666667,
|
| 2667 |
+
"grad_norm": 0.1699158102273941,
|
| 2668 |
+
"learning_rate": 9.474015815039446e-05,
|
| 2669 |
+
"loss": 0.0091,
|
| 2670 |
+
"step": 3800
|
| 2671 |
+
},
|
| 2672 |
+
{
|
| 2673 |
+
"epoch": 79.375,
|
| 2674 |
+
"grad_norm": 0.15745197236537933,
|
| 2675 |
+
"learning_rate": 9.470318653043565e-05,
|
| 2676 |
+
"loss": 0.009,
|
| 2677 |
+
"step": 3810
|
| 2678 |
+
},
|
| 2679 |
+
{
|
| 2680 |
+
"epoch": 79.58333333333333,
|
| 2681 |
+
"grad_norm": 0.1425720900297165,
|
| 2682 |
+
"learning_rate": 9.466609269367185e-05,
|
| 2683 |
+
"loss": 0.0081,
|
| 2684 |
+
"step": 3820
|
| 2685 |
+
},
|
| 2686 |
+
{
|
| 2687 |
+
"epoch": 79.79166666666667,
|
| 2688 |
+
"grad_norm": 0.13255997002124786,
|
| 2689 |
+
"learning_rate": 9.46288767415162e-05,
|
| 2690 |
+
"loss": 0.0085,
|
| 2691 |
+
"step": 3830
|
| 2692 |
+
},
|
| 2693 |
+
{
|
| 2694 |
+
"epoch": 80.0,
|
| 2695 |
+
"grad_norm": 0.18882817029953003,
|
| 2696 |
+
"learning_rate": 9.459153877571567e-05,
|
| 2697 |
+
"loss": 0.0077,
|
| 2698 |
+
"step": 3840
|
| 2699 |
+
},
|
| 2700 |
+
{
|
| 2701 |
+
"epoch": 80.20833333333333,
|
| 2702 |
+
"grad_norm": 0.12347506731748581,
|
| 2703 |
+
"learning_rate": 9.455407889835087e-05,
|
| 2704 |
+
"loss": 0.0076,
|
| 2705 |
+
"step": 3850
|
| 2706 |
+
},
|
| 2707 |
+
{
|
| 2708 |
+
"epoch": 80.41666666666667,
|
| 2709 |
+
"grad_norm": 0.134022518992424,
|
| 2710 |
+
"learning_rate": 9.451649721183564e-05,
|
| 2711 |
+
"loss": 0.008,
|
| 2712 |
+
"step": 3860
|
| 2713 |
+
},
|
| 2714 |
+
{
|
| 2715 |
+
"epoch": 80.625,
|
| 2716 |
+
"grad_norm": 0.10200574994087219,
|
| 2717 |
+
"learning_rate": 9.447879381891692e-05,
|
| 2718 |
+
"loss": 0.0081,
|
| 2719 |
+
"step": 3870
|
| 2720 |
+
},
|
| 2721 |
+
{
|
| 2722 |
+
"epoch": 80.83333333333333,
|
| 2723 |
+
"grad_norm": 0.11393435299396515,
|
| 2724 |
+
"learning_rate": 9.444096882267428e-05,
|
| 2725 |
+
"loss": 0.0067,
|
| 2726 |
+
"step": 3880
|
| 2727 |
+
},
|
| 2728 |
+
{
|
| 2729 |
+
"epoch": 81.04166666666667,
|
| 2730 |
+
"grad_norm": 0.17549581825733185,
|
| 2731 |
+
"learning_rate": 9.440302232651988e-05,
|
| 2732 |
+
"loss": 0.0068,
|
| 2733 |
+
"step": 3890
|
| 2734 |
+
},
|
| 2735 |
+
{
|
| 2736 |
+
"epoch": 81.25,
|
| 2737 |
+
"grad_norm": 0.129764124751091,
|
| 2738 |
+
"learning_rate": 9.436495443419795e-05,
|
| 2739 |
+
"loss": 0.0072,
|
| 2740 |
+
"step": 3900
|
| 2741 |
+
},
|
| 2742 |
+
{
|
| 2743 |
+
"epoch": 81.45833333333333,
|
| 2744 |
+
"grad_norm": 0.13367408514022827,
|
| 2745 |
+
"learning_rate": 9.432676524978466e-05,
|
| 2746 |
+
"loss": 0.0075,
|
| 2747 |
+
"step": 3910
|
| 2748 |
+
},
|
| 2749 |
+
{
|
| 2750 |
+
"epoch": 81.66666666666667,
|
| 2751 |
+
"grad_norm": 0.12855835258960724,
|
| 2752 |
+
"learning_rate": 9.42884548776878e-05,
|
| 2753 |
+
"loss": 0.007,
|
| 2754 |
+
"step": 3920
|
| 2755 |
+
},
|
| 2756 |
+
{
|
| 2757 |
+
"epoch": 81.875,
|
| 2758 |
+
"grad_norm": 0.12377104163169861,
|
| 2759 |
+
"learning_rate": 9.425002342264646e-05,
|
| 2760 |
+
"loss": 0.0066,
|
| 2761 |
+
"step": 3930
|
| 2762 |
+
},
|
| 2763 |
+
{
|
| 2764 |
+
"epoch": 82.08333333333333,
|
| 2765 |
+
"grad_norm": 0.14512717723846436,
|
| 2766 |
+
"learning_rate": 9.421147098973077e-05,
|
| 2767 |
+
"loss": 0.0078,
|
| 2768 |
+
"step": 3940
|
| 2769 |
+
},
|
| 2770 |
+
{
|
| 2771 |
+
"epoch": 82.29166666666667,
|
| 2772 |
+
"grad_norm": 0.14285895228385925,
|
| 2773 |
+
"learning_rate": 9.41727976843416e-05,
|
| 2774 |
+
"loss": 0.007,
|
| 2775 |
+
"step": 3950
|
| 2776 |
+
},
|
| 2777 |
+
{
|
| 2778 |
+
"epoch": 82.5,
|
| 2779 |
+
"grad_norm": 0.1557202786207199,
|
| 2780 |
+
"learning_rate": 9.413400361221029e-05,
|
| 2781 |
+
"loss": 0.0077,
|
| 2782 |
+
"step": 3960
|
| 2783 |
+
},
|
| 2784 |
+
{
|
| 2785 |
+
"epoch": 82.70833333333333,
|
| 2786 |
+
"grad_norm": 0.12391534447669983,
|
| 2787 |
+
"learning_rate": 9.409508887939835e-05,
|
| 2788 |
+
"loss": 0.0075,
|
| 2789 |
+
"step": 3970
|
| 2790 |
+
},
|
| 2791 |
+
{
|
| 2792 |
+
"epoch": 82.91666666666667,
|
| 2793 |
+
"grad_norm": 0.1519940346479416,
|
| 2794 |
+
"learning_rate": 9.40560535922972e-05,
|
| 2795 |
+
"loss": 0.0069,
|
| 2796 |
+
"step": 3980
|
| 2797 |
+
},
|
| 2798 |
+
{
|
| 2799 |
+
"epoch": 83.125,
|
| 2800 |
+
"grad_norm": 0.1247396171092987,
|
| 2801 |
+
"learning_rate": 9.40168978576278e-05,
|
| 2802 |
+
"loss": 0.0074,
|
| 2803 |
+
"step": 3990
|
| 2804 |
+
},
|
| 2805 |
+
{
|
| 2806 |
+
"epoch": 83.33333333333333,
|
| 2807 |
+
"grad_norm": 0.12143047899007797,
|
| 2808 |
+
"learning_rate": 9.397762178244043e-05,
|
| 2809 |
+
"loss": 0.0077,
|
| 2810 |
+
"step": 4000
|
| 2811 |
+
},
|
| 2812 |
+
{
|
| 2813 |
+
"epoch": 83.54166666666667,
|
| 2814 |
+
"grad_norm": 0.1239767074584961,
|
| 2815 |
+
"learning_rate": 9.393822547411439e-05,
|
| 2816 |
+
"loss": 0.0073,
|
| 2817 |
+
"step": 4010
|
| 2818 |
+
},
|
| 2819 |
+
{
|
| 2820 |
+
"epoch": 83.75,
|
| 2821 |
+
"grad_norm": 0.14913630485534668,
|
| 2822 |
+
"learning_rate": 9.389870904035769e-05,
|
| 2823 |
+
"loss": 0.0072,
|
| 2824 |
+
"step": 4020
|
| 2825 |
+
},
|
| 2826 |
+
{
|
| 2827 |
+
"epoch": 83.95833333333333,
|
| 2828 |
+
"grad_norm": 0.12743103504180908,
|
| 2829 |
+
"learning_rate": 9.385907258920672e-05,
|
| 2830 |
+
"loss": 0.0073,
|
| 2831 |
+
"step": 4030
|
| 2832 |
+
},
|
| 2833 |
+
{
|
| 2834 |
+
"epoch": 84.16666666666667,
|
| 2835 |
+
"grad_norm": 0.16922424733638763,
|
| 2836 |
+
"learning_rate": 9.381931622902607e-05,
|
| 2837 |
+
"loss": 0.0073,
|
| 2838 |
+
"step": 4040
|
| 2839 |
+
},
|
| 2840 |
+
{
|
| 2841 |
+
"epoch": 84.375,
|
| 2842 |
+
"grad_norm": 0.12121975421905518,
|
| 2843 |
+
"learning_rate": 9.377944006850807e-05,
|
| 2844 |
+
"loss": 0.0084,
|
| 2845 |
+
"step": 4050
|
| 2846 |
+
},
|
| 2847 |
+
{
|
| 2848 |
+
"epoch": 84.58333333333333,
|
| 2849 |
+
"grad_norm": 0.12847809493541718,
|
| 2850 |
+
"learning_rate": 9.373944421667265e-05,
|
| 2851 |
+
"loss": 0.0078,
|
| 2852 |
+
"step": 4060
|
| 2853 |
+
},
|
| 2854 |
+
{
|
| 2855 |
+
"epoch": 84.79166666666667,
|
| 2856 |
+
"grad_norm": 0.10404125601053238,
|
| 2857 |
+
"learning_rate": 9.369932878286691e-05,
|
| 2858 |
+
"loss": 0.0073,
|
| 2859 |
+
"step": 4070
|
| 2860 |
+
},
|
| 2861 |
+
{
|
| 2862 |
+
"epoch": 85.0,
|
| 2863 |
+
"grad_norm": 0.20326335728168488,
|
| 2864 |
+
"learning_rate": 9.365909387676494e-05,
|
| 2865 |
+
"loss": 0.0075,
|
| 2866 |
+
"step": 4080
|
| 2867 |
+
},
|
| 2868 |
+
{
|
| 2869 |
+
"epoch": 85.20833333333333,
|
| 2870 |
+
"grad_norm": 0.1208629459142685,
|
| 2871 |
+
"learning_rate": 9.361873960836744e-05,
|
| 2872 |
+
"loss": 0.0075,
|
| 2873 |
+
"step": 4090
|
| 2874 |
+
},
|
| 2875 |
+
{
|
| 2876 |
+
"epoch": 85.41666666666667,
|
| 2877 |
+
"grad_norm": 0.15183036029338837,
|
| 2878 |
+
"learning_rate": 9.357826608800142e-05,
|
| 2879 |
+
"loss": 0.0074,
|
| 2880 |
+
"step": 4100
|
| 2881 |
+
},
|
| 2882 |
+
{
|
| 2883 |
+
"epoch": 85.625,
|
| 2884 |
+
"grad_norm": 0.11224193871021271,
|
| 2885 |
+
"learning_rate": 9.353767342631994e-05,
|
| 2886 |
+
"loss": 0.0072,
|
| 2887 |
+
"step": 4110
|
| 2888 |
+
},
|
| 2889 |
+
{
|
| 2890 |
+
"epoch": 85.83333333333333,
|
| 2891 |
+
"grad_norm": 0.14832796156406403,
|
| 2892 |
+
"learning_rate": 9.34969617343018e-05,
|
| 2893 |
+
"loss": 0.0076,
|
| 2894 |
+
"step": 4120
|
| 2895 |
+
},
|
| 2896 |
+
{
|
| 2897 |
+
"epoch": 86.04166666666667,
|
| 2898 |
+
"grad_norm": 0.13874362409114838,
|
| 2899 |
+
"learning_rate": 9.345613112325122e-05,
|
| 2900 |
+
"loss": 0.008,
|
| 2901 |
+
"step": 4130
|
| 2902 |
+
},
|
| 2903 |
+
{
|
| 2904 |
+
"epoch": 86.25,
|
| 2905 |
+
"grad_norm": 0.12126526981592178,
|
| 2906 |
+
"learning_rate": 9.34151817047975e-05,
|
| 2907 |
+
"loss": 0.0077,
|
| 2908 |
+
"step": 4140
|
| 2909 |
+
},
|
| 2910 |
+
{
|
| 2911 |
+
"epoch": 86.45833333333333,
|
| 2912 |
+
"grad_norm": 0.11635075509548187,
|
| 2913 |
+
"learning_rate": 9.33741135908948e-05,
|
| 2914 |
+
"loss": 0.008,
|
| 2915 |
+
"step": 4150
|
| 2916 |
+
},
|
| 2917 |
+
{
|
| 2918 |
+
"epoch": 86.66666666666667,
|
| 2919 |
+
"grad_norm": 0.09447265416383743,
|
| 2920 |
+
"learning_rate": 9.33329268938218e-05,
|
| 2921 |
+
"loss": 0.0069,
|
| 2922 |
+
"step": 4160
|
| 2923 |
+
},
|
| 2924 |
+
{
|
| 2925 |
+
"epoch": 86.875,
|
| 2926 |
+
"grad_norm": 0.11846235394477844,
|
| 2927 |
+
"learning_rate": 9.329162172618132e-05,
|
| 2928 |
+
"loss": 0.0074,
|
| 2929 |
+
"step": 4170
|
| 2930 |
+
},
|
| 2931 |
+
{
|
| 2932 |
+
"epoch": 87.08333333333333,
|
| 2933 |
+
"grad_norm": 0.13078013062477112,
|
| 2934 |
+
"learning_rate": 9.325019820090013e-05,
|
| 2935 |
+
"loss": 0.0072,
|
| 2936 |
+
"step": 4180
|
| 2937 |
+
},
|
| 2938 |
+
{
|
| 2939 |
+
"epoch": 87.29166666666667,
|
| 2940 |
+
"grad_norm": 0.12578906118869781,
|
| 2941 |
+
"learning_rate": 9.320865643122855e-05,
|
| 2942 |
+
"loss": 0.0078,
|
| 2943 |
+
"step": 4190
|
| 2944 |
+
},
|
| 2945 |
+
{
|
| 2946 |
+
"epoch": 87.5,
|
| 2947 |
+
"grad_norm": 0.1526239514350891,
|
| 2948 |
+
"learning_rate": 9.316699653074023e-05,
|
| 2949 |
+
"loss": 0.0073,
|
| 2950 |
+
"step": 4200
|
| 2951 |
+
},
|
| 2952 |
+
{
|
| 2953 |
+
"epoch": 87.70833333333333,
|
| 2954 |
+
"grad_norm": 0.11341243982315063,
|
| 2955 |
+
"learning_rate": 9.312521861333172e-05,
|
| 2956 |
+
"loss": 0.0067,
|
| 2957 |
+
"step": 4210
|
| 2958 |
+
},
|
| 2959 |
+
{
|
| 2960 |
+
"epoch": 87.91666666666667,
|
| 2961 |
+
"grad_norm": 0.09736169129610062,
|
| 2962 |
+
"learning_rate": 9.308332279322224e-05,
|
| 2963 |
+
"loss": 0.0066,
|
| 2964 |
+
"step": 4220
|
| 2965 |
+
},
|
| 2966 |
+
{
|
| 2967 |
+
"epoch": 88.125,
|
| 2968 |
+
"grad_norm": 0.13081710040569305,
|
| 2969 |
+
"learning_rate": 9.304130918495338e-05,
|
| 2970 |
+
"loss": 0.0074,
|
| 2971 |
+
"step": 4230
|
| 2972 |
+
},
|
| 2973 |
+
{
|
| 2974 |
+
"epoch": 88.33333333333333,
|
| 2975 |
+
"grad_norm": 0.12707583606243134,
|
| 2976 |
+
"learning_rate": 9.299917790338874e-05,
|
| 2977 |
+
"loss": 0.0076,
|
| 2978 |
+
"step": 4240
|
| 2979 |
+
},
|
| 2980 |
+
{
|
| 2981 |
+
"epoch": 88.54166666666667,
|
| 2982 |
+
"grad_norm": 0.11190862953662872,
|
| 2983 |
+
"learning_rate": 9.295692906371363e-05,
|
| 2984 |
+
"loss": 0.0068,
|
| 2985 |
+
"step": 4250
|
| 2986 |
+
},
|
| 2987 |
+
{
|
| 2988 |
+
"epoch": 88.75,
|
| 2989 |
+
"grad_norm": 0.11754106730222702,
|
| 2990 |
+
"learning_rate": 9.291456278143476e-05,
|
| 2991 |
+
"loss": 0.0075,
|
| 2992 |
+
"step": 4260
|
| 2993 |
+
},
|
| 2994 |
+
{
|
| 2995 |
+
"epoch": 88.95833333333333,
|
| 2996 |
+
"grad_norm": 0.10456420481204987,
|
| 2997 |
+
"learning_rate": 9.287207917237994e-05,
|
| 2998 |
+
"loss": 0.0072,
|
| 2999 |
+
"step": 4270
|
| 3000 |
+
},
|
| 3001 |
+
{
|
| 3002 |
+
"epoch": 89.16666666666667,
|
| 3003 |
+
"grad_norm": 0.13785317540168762,
|
| 3004 |
+
"learning_rate": 9.282947835269773e-05,
|
| 3005 |
+
"loss": 0.0072,
|
| 3006 |
+
"step": 4280
|
| 3007 |
+
},
|
| 3008 |
+
{
|
| 3009 |
+
"epoch": 89.375,
|
| 3010 |
+
"grad_norm": 0.10497057437896729,
|
| 3011 |
+
"learning_rate": 9.278676043885715e-05,
|
| 3012 |
+
"loss": 0.0078,
|
| 3013 |
+
"step": 4290
|
| 3014 |
+
},
|
| 3015 |
+
{
|
| 3016 |
+
"epoch": 89.58333333333333,
|
| 3017 |
+
"grad_norm": 0.10352105647325516,
|
| 3018 |
+
"learning_rate": 9.274392554764733e-05,
|
| 3019 |
+
"loss": 0.0073,
|
| 3020 |
+
"step": 4300
|
| 3021 |
+
},
|
| 3022 |
+
{
|
| 3023 |
+
"epoch": 89.79166666666667,
|
| 3024 |
+
"grad_norm": 0.1163288876414299,
|
| 3025 |
+
"learning_rate": 9.270097379617723e-05,
|
| 3026 |
+
"loss": 0.0073,
|
| 3027 |
+
"step": 4310
|
| 3028 |
+
},
|
| 3029 |
+
{
|
| 3030 |
+
"epoch": 90.0,
|
| 3031 |
+
"grad_norm": 0.17180398106575012,
|
| 3032 |
+
"learning_rate": 9.26579053018753e-05,
|
| 3033 |
+
"loss": 0.0071,
|
| 3034 |
+
"step": 4320
|
| 3035 |
+
},
|
| 3036 |
+
{
|
| 3037 |
+
"epoch": 90.20833333333333,
|
| 3038 |
+
"grad_norm": 0.15238836407661438,
|
| 3039 |
+
"learning_rate": 9.261472018248918e-05,
|
| 3040 |
+
"loss": 0.0072,
|
| 3041 |
+
"step": 4330
|
| 3042 |
+
},
|
| 3043 |
+
{
|
| 3044 |
+
"epoch": 90.41666666666667,
|
| 3045 |
+
"grad_norm": 0.12145853042602539,
|
| 3046 |
+
"learning_rate": 9.25714185560853e-05,
|
| 3047 |
+
"loss": 0.0076,
|
| 3048 |
+
"step": 4340
|
| 3049 |
+
},
|
| 3050 |
+
{
|
| 3051 |
+
"epoch": 90.625,
|
| 3052 |
+
"grad_norm": 0.1380954384803772,
|
| 3053 |
+
"learning_rate": 9.252800054104868e-05,
|
| 3054 |
+
"loss": 0.0073,
|
| 3055 |
+
"step": 4350
|
| 3056 |
+
},
|
| 3057 |
+
{
|
| 3058 |
+
"epoch": 90.83333333333333,
|
| 3059 |
+
"grad_norm": 0.14991872012615204,
|
| 3060 |
+
"learning_rate": 9.248446625608252e-05,
|
| 3061 |
+
"loss": 0.0072,
|
| 3062 |
+
"step": 4360
|
| 3063 |
+
},
|
| 3064 |
+
{
|
| 3065 |
+
"epoch": 91.04166666666667,
|
| 3066 |
+
"grad_norm": 0.15483133494853973,
|
| 3067 |
+
"learning_rate": 9.244081582020789e-05,
|
| 3068 |
+
"loss": 0.0083,
|
| 3069 |
+
"step": 4370
|
| 3070 |
+
},
|
| 3071 |
+
{
|
| 3072 |
+
"epoch": 91.25,
|
| 3073 |
+
"grad_norm": 0.12043429166078568,
|
| 3074 |
+
"learning_rate": 9.239704935276339e-05,
|
| 3075 |
+
"loss": 0.0079,
|
| 3076 |
+
"step": 4380
|
| 3077 |
+
},
|
| 3078 |
+
{
|
| 3079 |
+
"epoch": 91.45833333333333,
|
| 3080 |
+
"grad_norm": 0.13251228630542755,
|
| 3081 |
+
"learning_rate": 9.235316697340489e-05,
|
| 3082 |
+
"loss": 0.0075,
|
| 3083 |
+
"step": 4390
|
| 3084 |
+
},
|
| 3085 |
+
{
|
| 3086 |
+
"epoch": 91.66666666666667,
|
| 3087 |
+
"grad_norm": 0.1269141584634781,
|
| 3088 |
+
"learning_rate": 9.230916880210512e-05,
|
| 3089 |
+
"loss": 0.0079,
|
| 3090 |
+
"step": 4400
|
| 3091 |
+
},
|
| 3092 |
+
{
|
| 3093 |
+
"epoch": 91.875,
|
| 3094 |
+
"grad_norm": 0.11620093137025833,
|
| 3095 |
+
"learning_rate": 9.226505495915342e-05,
|
| 3096 |
+
"loss": 0.0073,
|
| 3097 |
+
"step": 4410
|
| 3098 |
+
},
|
| 3099 |
+
{
|
| 3100 |
+
"epoch": 92.08333333333333,
|
| 3101 |
+
"grad_norm": 0.1434793919324875,
|
| 3102 |
+
"learning_rate": 9.222082556515536e-05,
|
| 3103 |
+
"loss": 0.0063,
|
| 3104 |
+
"step": 4420
|
| 3105 |
+
},
|
| 3106 |
+
{
|
| 3107 |
+
"epoch": 92.29166666666667,
|
| 3108 |
+
"grad_norm": 0.1004461795091629,
|
| 3109 |
+
"learning_rate": 9.217648074103242e-05,
|
| 3110 |
+
"loss": 0.008,
|
| 3111 |
+
"step": 4430
|
| 3112 |
+
},
|
| 3113 |
+
{
|
| 3114 |
+
"epoch": 92.5,
|
| 3115 |
+
"grad_norm": 0.0979839563369751,
|
| 3116 |
+
"learning_rate": 9.213202060802161e-05,
|
| 3117 |
+
"loss": 0.0068,
|
| 3118 |
+
"step": 4440
|
| 3119 |
+
},
|
| 3120 |
+
{
|
| 3121 |
+
"epoch": 92.70833333333333,
|
| 3122 |
+
"grad_norm": 0.11008862406015396,
|
| 3123 |
+
"learning_rate": 9.208744528767528e-05,
|
| 3124 |
+
"loss": 0.0069,
|
| 3125 |
+
"step": 4450
|
| 3126 |
+
},
|
| 3127 |
+
{
|
| 3128 |
+
"epoch": 92.91666666666667,
|
| 3129 |
+
"grad_norm": 0.11195098608732224,
|
| 3130 |
+
"learning_rate": 9.204275490186064e-05,
|
| 3131 |
+
"loss": 0.007,
|
| 3132 |
+
"step": 4460
|
| 3133 |
+
},
|
| 3134 |
+
{
|
| 3135 |
+
"epoch": 93.125,
|
| 3136 |
+
"grad_norm": 0.13116782903671265,
|
| 3137 |
+
"learning_rate": 9.199794957275949e-05,
|
| 3138 |
+
"loss": 0.0068,
|
| 3139 |
+
"step": 4470
|
| 3140 |
+
},
|
| 3141 |
+
{
|
| 3142 |
+
"epoch": 93.33333333333333,
|
| 3143 |
+
"grad_norm": 0.10709001123905182,
|
| 3144 |
+
"learning_rate": 9.19530294228679e-05,
|
| 3145 |
+
"loss": 0.0062,
|
| 3146 |
+
"step": 4480
|
| 3147 |
+
},
|
| 3148 |
+
{
|
| 3149 |
+
"epoch": 93.54166666666667,
|
| 3150 |
+
"grad_norm": 0.11429505795240402,
|
| 3151 |
+
"learning_rate": 9.190799457499583e-05,
|
| 3152 |
+
"loss": 0.0072,
|
| 3153 |
+
"step": 4490
|
| 3154 |
+
},
|
| 3155 |
+
{
|
| 3156 |
+
"epoch": 93.75,
|
| 3157 |
+
"grad_norm": 0.12184348702430725,
|
| 3158 |
+
"learning_rate": 9.186284515226686e-05,
|
| 3159 |
+
"loss": 0.0071,
|
| 3160 |
+
"step": 4500
|
| 3161 |
+
},
|
| 3162 |
+
{
|
| 3163 |
+
"epoch": 93.95833333333333,
|
| 3164 |
+
"grad_norm": 0.12104353308677673,
|
| 3165 |
+
"learning_rate": 9.181758127811777e-05,
|
| 3166 |
+
"loss": 0.0074,
|
| 3167 |
+
"step": 4510
|
| 3168 |
+
},
|
| 3169 |
+
{
|
| 3170 |
+
"epoch": 94.16666666666667,
|
| 3171 |
+
"grad_norm": 0.09736120700836182,
|
| 3172 |
+
"learning_rate": 9.177220307629825e-05,
|
| 3173 |
+
"loss": 0.0076,
|
| 3174 |
+
"step": 4520
|
| 3175 |
+
},
|
| 3176 |
+
{
|
| 3177 |
+
"epoch": 94.375,
|
| 3178 |
+
"grad_norm": 0.09775745123624802,
|
| 3179 |
+
"learning_rate": 9.172671067087059e-05,
|
| 3180 |
+
"loss": 0.0076,
|
| 3181 |
+
"step": 4530
|
| 3182 |
+
},
|
| 3183 |
+
{
|
| 3184 |
+
"epoch": 94.58333333333333,
|
| 3185 |
+
"grad_norm": 0.10957901924848557,
|
| 3186 |
+
"learning_rate": 9.16811041862093e-05,
|
| 3187 |
+
"loss": 0.0069,
|
| 3188 |
+
"step": 4540
|
| 3189 |
+
},
|
| 3190 |
+
{
|
| 3191 |
+
"epoch": 94.79166666666667,
|
| 3192 |
+
"grad_norm": 0.11660820990800858,
|
| 3193 |
+
"learning_rate": 9.163538374700076e-05,
|
| 3194 |
+
"loss": 0.0067,
|
| 3195 |
+
"step": 4550
|
| 3196 |
+
},
|
| 3197 |
+
{
|
| 3198 |
+
"epoch": 95.0,
|
| 3199 |
+
"grad_norm": 0.1356566995382309,
|
| 3200 |
+
"learning_rate": 9.158954947824287e-05,
|
| 3201 |
+
"loss": 0.0065,
|
| 3202 |
+
"step": 4560
|
| 3203 |
+
},
|
| 3204 |
+
{
|
| 3205 |
+
"epoch": 95.20833333333333,
|
| 3206 |
+
"grad_norm": 0.1425929218530655,
|
| 3207 |
+
"learning_rate": 9.154360150524482e-05,
|
| 3208 |
+
"loss": 0.0069,
|
| 3209 |
+
"step": 4570
|
| 3210 |
+
},
|
| 3211 |
+
{
|
| 3212 |
+
"epoch": 95.41666666666667,
|
| 3213 |
+
"grad_norm": 0.13953809440135956,
|
| 3214 |
+
"learning_rate": 9.14975399536266e-05,
|
| 3215 |
+
"loss": 0.0078,
|
| 3216 |
+
"step": 4580
|
| 3217 |
+
},
|
| 3218 |
+
{
|
| 3219 |
+
"epoch": 95.625,
|
| 3220 |
+
"grad_norm": 0.08912316709756851,
|
| 3221 |
+
"learning_rate": 9.14513649493187e-05,
|
| 3222 |
+
"loss": 0.0067,
|
| 3223 |
+
"step": 4590
|
| 3224 |
+
},
|
| 3225 |
+
{
|
| 3226 |
+
"epoch": 95.83333333333333,
|
| 3227 |
+
"grad_norm": 0.10334280133247375,
|
| 3228 |
+
"learning_rate": 9.140507661856187e-05,
|
| 3229 |
+
"loss": 0.0066,
|
| 3230 |
+
"step": 4600
|
| 3231 |
+
},
|
| 3232 |
+
{
|
| 3233 |
+
"epoch": 96.04166666666667,
|
| 3234 |
+
"grad_norm": 0.11234814673662186,
|
| 3235 |
+
"learning_rate": 9.135867508790661e-05,
|
| 3236 |
+
"loss": 0.0068,
|
| 3237 |
+
"step": 4610
|
| 3238 |
+
},
|
| 3239 |
+
{
|
| 3240 |
+
"epoch": 96.25,
|
| 3241 |
+
"grad_norm": 0.1121845468878746,
|
| 3242 |
+
"learning_rate": 9.131216048421291e-05,
|
| 3243 |
+
"loss": 0.0078,
|
| 3244 |
+
"step": 4620
|
| 3245 |
+
},
|
| 3246 |
+
{
|
| 3247 |
+
"epoch": 96.45833333333333,
|
| 3248 |
+
"grad_norm": 0.1481083333492279,
|
| 3249 |
+
"learning_rate": 9.126553293464998e-05,
|
| 3250 |
+
"loss": 0.0073,
|
| 3251 |
+
"step": 4630
|
| 3252 |
+
},
|
| 3253 |
+
{
|
| 3254 |
+
"epoch": 96.66666666666667,
|
| 3255 |
+
"grad_norm": 0.1397561877965927,
|
| 3256 |
+
"learning_rate": 9.121879256669572e-05,
|
| 3257 |
+
"loss": 0.0079,
|
| 3258 |
+
"step": 4640
|
| 3259 |
+
},
|
| 3260 |
+
{
|
| 3261 |
+
"epoch": 96.875,
|
| 3262 |
+
"grad_norm": 0.1100471243262291,
|
| 3263 |
+
"learning_rate": 9.117193950813652e-05,
|
| 3264 |
+
"loss": 0.0072,
|
| 3265 |
+
"step": 4650
|
| 3266 |
+
},
|
| 3267 |
+
{
|
| 3268 |
+
"epoch": 97.08333333333333,
|
| 3269 |
+
"grad_norm": 0.18109332025051117,
|
| 3270 |
+
"learning_rate": 9.112497388706685e-05,
|
| 3271 |
+
"loss": 0.0071,
|
| 3272 |
+
"step": 4660
|
| 3273 |
+
},
|
| 3274 |
+
{
|
| 3275 |
+
"epoch": 97.29166666666667,
|
| 3276 |
+
"grad_norm": 0.16449885070323944,
|
| 3277 |
+
"learning_rate": 9.10778958318889e-05,
|
| 3278 |
+
"loss": 0.0081,
|
| 3279 |
+
"step": 4670
|
| 3280 |
+
},
|
| 3281 |
+
{
|
| 3282 |
+
"epoch": 97.5,
|
| 3283 |
+
"grad_norm": 0.18169593811035156,
|
| 3284 |
+
"learning_rate": 9.103070547131232e-05,
|
| 3285 |
+
"loss": 0.0076,
|
| 3286 |
+
"step": 4680
|
| 3287 |
+
},
|
| 3288 |
+
{
|
| 3289 |
+
"epoch": 97.70833333333333,
|
| 3290 |
+
"grad_norm": 0.17230816185474396,
|
| 3291 |
+
"learning_rate": 9.098340293435375e-05,
|
| 3292 |
+
"loss": 0.0075,
|
| 3293 |
+
"step": 4690
|
| 3294 |
+
},
|
| 3295 |
+
{
|
| 3296 |
+
"epoch": 97.91666666666667,
|
| 3297 |
+
"grad_norm": 0.11778809130191803,
|
| 3298 |
+
"learning_rate": 9.093598835033649e-05,
|
| 3299 |
+
"loss": 0.0072,
|
| 3300 |
+
"step": 4700
|
| 3301 |
+
},
|
| 3302 |
+
{
|
| 3303 |
+
"epoch": 98.125,
|
| 3304 |
+
"grad_norm": 0.15728609263896942,
|
| 3305 |
+
"learning_rate": 9.088846184889021e-05,
|
| 3306 |
+
"loss": 0.0072,
|
| 3307 |
+
"step": 4710
|
| 3308 |
+
},
|
| 3309 |
+
{
|
| 3310 |
+
"epoch": 98.33333333333333,
|
| 3311 |
+
"grad_norm": 0.11238937824964523,
|
| 3312 |
+
"learning_rate": 9.084082355995057e-05,
|
| 3313 |
+
"loss": 0.0073,
|
| 3314 |
+
"step": 4720
|
| 3315 |
+
},
|
| 3316 |
+
{
|
| 3317 |
+
"epoch": 98.54166666666667,
|
| 3318 |
+
"grad_norm": 0.11571408063173294,
|
| 3319 |
+
"learning_rate": 9.079307361375882e-05,
|
| 3320 |
+
"loss": 0.0067,
|
| 3321 |
+
"step": 4730
|
| 3322 |
+
},
|
| 3323 |
+
{
|
| 3324 |
+
"epoch": 98.75,
|
| 3325 |
+
"grad_norm": 0.12732599675655365,
|
| 3326 |
+
"learning_rate": 9.074521214086149e-05,
|
| 3327 |
+
"loss": 0.0072,
|
| 3328 |
+
"step": 4740
|
| 3329 |
+
},
|
| 3330 |
+
{
|
| 3331 |
+
"epoch": 98.95833333333333,
|
| 3332 |
+
"grad_norm": 0.11861906200647354,
|
| 3333 |
+
"learning_rate": 9.069723927211001e-05,
|
| 3334 |
+
"loss": 0.0073,
|
| 3335 |
+
"step": 4750
|
| 3336 |
+
},
|
| 3337 |
+
{
|
| 3338 |
+
"epoch": 99.16666666666667,
|
| 3339 |
+
"grad_norm": 0.12790387868881226,
|
| 3340 |
+
"learning_rate": 9.064915513866037e-05,
|
| 3341 |
+
"loss": 0.007,
|
| 3342 |
+
"step": 4760
|
| 3343 |
+
},
|
| 3344 |
+
{
|
| 3345 |
+
"epoch": 99.375,
|
| 3346 |
+
"grad_norm": 0.14003939926624298,
|
| 3347 |
+
"learning_rate": 9.060095987197279e-05,
|
| 3348 |
+
"loss": 0.0075,
|
| 3349 |
+
"step": 4770
|
| 3350 |
+
},
|
| 3351 |
+
{
|
| 3352 |
+
"epoch": 99.58333333333333,
|
| 3353 |
+
"grad_norm": 0.18221960961818695,
|
| 3354 |
+
"learning_rate": 9.055265360381126e-05,
|
| 3355 |
+
"loss": 0.0079,
|
| 3356 |
+
"step": 4780
|
| 3357 |
+
},
|
| 3358 |
+
{
|
| 3359 |
+
"epoch": 99.79166666666667,
|
| 3360 |
+
"grad_norm": 0.18938899040222168,
|
| 3361 |
+
"learning_rate": 9.050423646624326e-05,
|
| 3362 |
+
"loss": 0.0078,
|
| 3363 |
+
"step": 4790
|
| 3364 |
+
},
|
| 3365 |
+
{
|
| 3366 |
+
"epoch": 100.0,
|
| 3367 |
+
"grad_norm": 0.1797226518392563,
|
| 3368 |
+
"learning_rate": 9.045570859163943e-05,
|
| 3369 |
+
"loss": 0.007,
|
| 3370 |
+
"step": 4800
|
| 3371 |
+
},
|
| 3372 |
+
{
|
| 3373 |
+
"epoch": 100.20833333333333,
|
| 3374 |
+
"grad_norm": 0.11891371011734009,
|
| 3375 |
+
"learning_rate": 9.04070701126731e-05,
|
| 3376 |
+
"loss": 0.0074,
|
| 3377 |
+
"step": 4810
|
| 3378 |
+
},
|
| 3379 |
+
{
|
| 3380 |
+
"epoch": 100.41666666666667,
|
| 3381 |
+
"grad_norm": 0.11718389391899109,
|
| 3382 |
+
"learning_rate": 9.035832116232001e-05,
|
| 3383 |
+
"loss": 0.0078,
|
| 3384 |
+
"step": 4820
|
| 3385 |
+
},
|
| 3386 |
+
{
|
| 3387 |
+
"epoch": 100.625,
|
| 3388 |
+
"grad_norm": 0.1415216028690338,
|
| 3389 |
+
"learning_rate": 9.030946187385796e-05,
|
| 3390 |
+
"loss": 0.0075,
|
| 3391 |
+
"step": 4830
|
| 3392 |
+
},
|
| 3393 |
+
{
|
| 3394 |
+
"epoch": 100.83333333333333,
|
| 3395 |
+
"grad_norm": 0.14168231189250946,
|
| 3396 |
+
"learning_rate": 9.026049238086635e-05,
|
| 3397 |
+
"loss": 0.0065,
|
| 3398 |
+
"step": 4840
|
| 3399 |
+
},
|
| 3400 |
+
{
|
| 3401 |
+
"epoch": 101.04166666666667,
|
| 3402 |
+
"grad_norm": 0.17353275418281555,
|
| 3403 |
+
"learning_rate": 9.021141281722591e-05,
|
| 3404 |
+
"loss": 0.0067,
|
| 3405 |
+
"step": 4850
|
| 3406 |
+
},
|
| 3407 |
+
{
|
| 3408 |
+
"epoch": 101.25,
|
| 3409 |
+
"grad_norm": 0.13674576580524445,
|
| 3410 |
+
"learning_rate": 9.01622233171183e-05,
|
| 3411 |
+
"loss": 0.0074,
|
| 3412 |
+
"step": 4860
|
| 3413 |
+
},
|
| 3414 |
+
{
|
| 3415 |
+
"epoch": 101.45833333333333,
|
| 3416 |
+
"grad_norm": 0.16381537914276123,
|
| 3417 |
+
"learning_rate": 9.011292401502574e-05,
|
| 3418 |
+
"loss": 0.0067,
|
| 3419 |
+
"step": 4870
|
| 3420 |
+
},
|
| 3421 |
+
{
|
| 3422 |
+
"epoch": 101.66666666666667,
|
| 3423 |
+
"grad_norm": 0.12143448740243912,
|
| 3424 |
+
"learning_rate": 9.006351504573063e-05,
|
| 3425 |
+
"loss": 0.0073,
|
| 3426 |
+
"step": 4880
|
| 3427 |
+
},
|
| 3428 |
+
{
|
| 3429 |
+
"epoch": 101.875,
|
| 3430 |
+
"grad_norm": 0.14107128977775574,
|
| 3431 |
+
"learning_rate": 9.001399654431519e-05,
|
| 3432 |
+
"loss": 0.0065,
|
| 3433 |
+
"step": 4890
|
| 3434 |
+
},
|
| 3435 |
+
{
|
| 3436 |
+
"epoch": 102.08333333333333,
|
| 3437 |
+
"grad_norm": 0.12365766614675522,
|
| 3438 |
+
"learning_rate": 8.996436864616116e-05,
|
| 3439 |
+
"loss": 0.0071,
|
| 3440 |
+
"step": 4900
|
| 3441 |
+
},
|
| 3442 |
+
{
|
| 3443 |
+
"epoch": 102.29166666666667,
|
| 3444 |
+
"grad_norm": 0.11717549711465836,
|
| 3445 |
+
"learning_rate": 8.991463148694925e-05,
|
| 3446 |
+
"loss": 0.0071,
|
| 3447 |
+
"step": 4910
|
| 3448 |
+
},
|
| 3449 |
+
{
|
| 3450 |
+
"epoch": 102.5,
|
| 3451 |
+
"grad_norm": 0.12707993388175964,
|
| 3452 |
+
"learning_rate": 8.986478520265902e-05,
|
| 3453 |
+
"loss": 0.007,
|
| 3454 |
+
"step": 4920
|
| 3455 |
+
},
|
| 3456 |
+
{
|
| 3457 |
+
"epoch": 102.70833333333333,
|
| 3458 |
+
"grad_norm": 0.1060505360364914,
|
| 3459 |
+
"learning_rate": 8.981482992956827e-05,
|
| 3460 |
+
"loss": 0.007,
|
| 3461 |
+
"step": 4930
|
| 3462 |
+
},
|
| 3463 |
+
{
|
| 3464 |
+
"epoch": 102.91666666666667,
|
| 3465 |
+
"grad_norm": 0.0786166563630104,
|
| 3466 |
+
"learning_rate": 8.976476580425282e-05,
|
| 3467 |
+
"loss": 0.0068,
|
| 3468 |
+
"step": 4940
|
| 3469 |
+
},
|
| 3470 |
+
{
|
| 3471 |
+
"epoch": 103.125,
|
| 3472 |
+
"grad_norm": 0.10416026413440704,
|
| 3473 |
+
"learning_rate": 8.971459296358606e-05,
|
| 3474 |
+
"loss": 0.0075,
|
| 3475 |
+
"step": 4950
|
| 3476 |
+
},
|
| 3477 |
+
{
|
| 3478 |
+
"epoch": 103.33333333333333,
|
| 3479 |
+
"grad_norm": 0.12413540482521057,
|
| 3480 |
+
"learning_rate": 8.966431154473864e-05,
|
| 3481 |
+
"loss": 0.0066,
|
| 3482 |
+
"step": 4960
|
| 3483 |
+
},
|
| 3484 |
+
{
|
| 3485 |
+
"epoch": 103.54166666666667,
|
| 3486 |
+
"grad_norm": 0.09850534051656723,
|
| 3487 |
+
"learning_rate": 8.961392168517803e-05,
|
| 3488 |
+
"loss": 0.0067,
|
| 3489 |
+
"step": 4970
|
| 3490 |
+
},
|
| 3491 |
+
{
|
| 3492 |
+
"epoch": 103.75,
|
| 3493 |
+
"grad_norm": 0.11118543148040771,
|
| 3494 |
+
"learning_rate": 8.956342352266821e-05,
|
| 3495 |
+
"loss": 0.0058,
|
| 3496 |
+
"step": 4980
|
| 3497 |
+
},
|
| 3498 |
+
{
|
| 3499 |
+
"epoch": 103.95833333333333,
|
| 3500 |
+
"grad_norm": 0.11706725507974625,
|
| 3501 |
+
"learning_rate": 8.95128171952692e-05,
|
| 3502 |
+
"loss": 0.0068,
|
| 3503 |
+
"step": 4990
|
| 3504 |
+
},
|
| 3505 |
+
{
|
| 3506 |
+
"epoch": 104.16666666666667,
|
| 3507 |
+
"grad_norm": 0.10897029936313629,
|
| 3508 |
+
"learning_rate": 8.946210284133676e-05,
|
| 3509 |
+
"loss": 0.0064,
|
| 3510 |
+
"step": 5000
|
| 3511 |
+
}
|
| 3512 |
+
],
|
| 3513 |
+
"logging_steps": 10,
|
| 3514 |
+
"max_steps": 20000,
|
| 3515 |
+
"num_input_tokens_seen": 0,
|
| 3516 |
+
"num_train_epochs": 417,
|
| 3517 |
+
"save_steps": 1000,
|
| 3518 |
+
"stateful_callbacks": {
|
| 3519 |
+
"TrainerControl": {
|
| 3520 |
+
"args": {
|
| 3521 |
+
"should_epoch_stop": false,
|
| 3522 |
+
"should_evaluate": false,
|
| 3523 |
+
"should_log": false,
|
| 3524 |
+
"should_save": true,
|
| 3525 |
+
"should_training_stop": false
|
| 3526 |
+
},
|
| 3527 |
+
"attributes": {}
|
| 3528 |
+
}
|
| 3529 |
+
},
|
| 3530 |
+
"total_flos": 0.0,
|
| 3531 |
+
"train_batch_size": 512,
|
| 3532 |
+
"trial_name": null,
|
| 3533 |
+
"trial_params": null
|
| 3534 |
+
}
|