Upload 10 files
Browse files- model.safetensors +2 -2
- optimizer.pt +2 -2
- rng_state.pth +3 -0
- scheduler.pt +3 -0
- trainer_state.json +513 -0
- training_args.bin +1 -1
- vocab.json +2 -1
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:75193d18bc129832e1f6711c59f27410fbbd5d6869cca5b81dd66df666fbfc1a
|
| 3 |
+
size 1264706196
|
optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e332e85556670d58ba8358746d2ab63b030cfa6f80961bed89ff75cb4dc77976
|
| 3 |
+
size 2495957174
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:17d726fa02cb8e68090d57bdc05a8e284a8cfd8f8f5eb3d82662a7ce28969111
|
| 3 |
+
size 14244
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:81180eadefc4016597f65e4de756793f7db2130f4433759e71bb53d8963d8222
|
| 3 |
+
size 1064
|
trainer_state.json
ADDED
|
@@ -0,0 +1,513 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.3443602919578552,
|
| 3 |
+
"best_model_checkpoint": "./wav2vec2-base-demo/checkpoint-12000",
|
| 4 |
+
"epoch": 5.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 12160,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.20559210526315788,
|
| 13 |
+
"grad_norm": 1.9561805725097656,
|
| 14 |
+
"learning_rate": 4.9500000000000004e-05,
|
| 15 |
+
"loss": 22.2524,
|
| 16 |
+
"step": 500
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.20559210526315788,
|
| 20 |
+
"eval_cer": 1.0,
|
| 21 |
+
"eval_loss": 5.067812442779541,
|
| 22 |
+
"eval_mer": 1.0,
|
| 23 |
+
"eval_runtime": 77.9239,
|
| 24 |
+
"eval_samples_per_second": 15.605,
|
| 25 |
+
"eval_steps_per_second": 1.951,
|
| 26 |
+
"eval_wer": 1.0,
|
| 27 |
+
"eval_wil": 1,
|
| 28 |
+
"eval_wip": 0,
|
| 29 |
+
"step": 500
|
| 30 |
+
},
|
| 31 |
+
{
|
| 32 |
+
"epoch": 0.41118421052631576,
|
| 33 |
+
"grad_norm": 3.8478355407714844,
|
| 34 |
+
"learning_rate": 9.95e-05,
|
| 35 |
+
"loss": 5.0193,
|
| 36 |
+
"step": 1000
|
| 37 |
+
},
|
| 38 |
+
{
|
| 39 |
+
"epoch": 0.41118421052631576,
|
| 40 |
+
"eval_cer": 0.8728056343950383,
|
| 41 |
+
"eval_loss": 4.214690208435059,
|
| 42 |
+
"eval_mer": 1.0,
|
| 43 |
+
"eval_runtime": 77.784,
|
| 44 |
+
"eval_samples_per_second": 15.633,
|
| 45 |
+
"eval_steps_per_second": 1.954,
|
| 46 |
+
"eval_wer": 1.0,
|
| 47 |
+
"eval_wil": 1.0,
|
| 48 |
+
"eval_wip": 0.0,
|
| 49 |
+
"step": 1000
|
| 50 |
+
},
|
| 51 |
+
{
|
| 52 |
+
"epoch": 0.6167763157894737,
|
| 53 |
+
"grad_norm": 3.14013934135437,
|
| 54 |
+
"learning_rate": 9.556451612903226e-05,
|
| 55 |
+
"loss": 3.1538,
|
| 56 |
+
"step": 1500
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.6167763157894737,
|
| 60 |
+
"eval_cer": 0.3883107326815936,
|
| 61 |
+
"eval_loss": 1.8805818557739258,
|
| 62 |
+
"eval_mer": 0.9181367144432194,
|
| 63 |
+
"eval_runtime": 77.8665,
|
| 64 |
+
"eval_samples_per_second": 15.616,
|
| 65 |
+
"eval_steps_per_second": 1.952,
|
| 66 |
+
"eval_wer": 1.0084771419921283,
|
| 67 |
+
"eval_wil": 0.9917651180786788,
|
| 68 |
+
"eval_wip": 0.008234881921321211,
|
| 69 |
+
"step": 1500
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.8223684210526315,
|
| 73 |
+
"grad_norm": 3.22117280960083,
|
| 74 |
+
"learning_rate": 9.108422939068101e-05,
|
| 75 |
+
"loss": 2.0197,
|
| 76 |
+
"step": 2000
|
| 77 |
+
},
|
| 78 |
+
{
|
| 79 |
+
"epoch": 0.8223684210526315,
|
| 80 |
+
"eval_cer": 0.2805003679175865,
|
| 81 |
+
"eval_loss": 1.3426703214645386,
|
| 82 |
+
"eval_mer": 0.8550304436068425,
|
| 83 |
+
"eval_runtime": 78.2982,
|
| 84 |
+
"eval_samples_per_second": 15.53,
|
| 85 |
+
"eval_steps_per_second": 1.941,
|
| 86 |
+
"eval_wer": 0.8928247048138056,
|
| 87 |
+
"eval_wil": 0.9747704107377132,
|
| 88 |
+
"eval_wip": 0.025229589262286807,
|
| 89 |
+
"step": 2000
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 1.0279605263157894,
|
| 93 |
+
"grad_norm": 1.7833839654922485,
|
| 94 |
+
"learning_rate": 8.660394265232975e-05,
|
| 95 |
+
"loss": 1.6264,
|
| 96 |
+
"step": 2500
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 1.0279605263157894,
|
| 100 |
+
"eval_cer": 0.21826973615053086,
|
| 101 |
+
"eval_loss": 1.0422340631484985,
|
| 102 |
+
"eval_mer": 0.7764672036823935,
|
| 103 |
+
"eval_runtime": 77.8097,
|
| 104 |
+
"eval_samples_per_second": 15.628,
|
| 105 |
+
"eval_steps_per_second": 1.953,
|
| 106 |
+
"eval_wer": 0.8171359370269452,
|
| 107 |
+
"eval_wil": 0.9414535501765618,
|
| 108 |
+
"eval_wip": 0.05854644982343813,
|
| 109 |
+
"step": 2500
|
| 110 |
+
},
|
| 111 |
+
{
|
| 112 |
+
"epoch": 1.2335526315789473,
|
| 113 |
+
"grad_norm": 2.036498785018921,
|
| 114 |
+
"learning_rate": 8.21236559139785e-05,
|
| 115 |
+
"loss": 1.3881,
|
| 116 |
+
"step": 3000
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 1.2335526315789473,
|
| 120 |
+
"eval_cer": 0.18149900136655103,
|
| 121 |
+
"eval_loss": 0.8646105527877808,
|
| 122 |
+
"eval_mer": 0.7196640602374746,
|
| 123 |
+
"eval_runtime": 77.7163,
|
| 124 |
+
"eval_samples_per_second": 15.647,
|
| 125 |
+
"eval_steps_per_second": 1.956,
|
| 126 |
+
"eval_wer": 0.7523463518013926,
|
| 127 |
+
"eval_wil": 0.9099972183417321,
|
| 128 |
+
"eval_wip": 0.09000278165826789,
|
| 129 |
+
"step": 3000
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 1.4391447368421053,
|
| 133 |
+
"grad_norm": 2.3967819213867188,
|
| 134 |
+
"learning_rate": 7.764336917562724e-05,
|
| 135 |
+
"loss": 1.2429,
|
| 136 |
+
"step": 3500
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 1.4391447368421053,
|
| 140 |
+
"eval_cer": 0.15513507831388626,
|
| 141 |
+
"eval_loss": 0.7484866976737976,
|
| 142 |
+
"eval_mer": 0.6633607254179654,
|
| 143 |
+
"eval_runtime": 77.7935,
|
| 144 |
+
"eval_samples_per_second": 15.631,
|
| 145 |
+
"eval_steps_per_second": 1.954,
|
| 146 |
+
"eval_wer": 0.7087496215561611,
|
| 147 |
+
"eval_wil": 0.872450282646712,
|
| 148 |
+
"eval_wip": 0.12754971735328804,
|
| 149 |
+
"step": 3500
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 1.6447368421052633,
|
| 153 |
+
"grad_norm": 1.4661535024642944,
|
| 154 |
+
"learning_rate": 7.316308243727599e-05,
|
| 155 |
+
"loss": 1.1192,
|
| 156 |
+
"step": 4000
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 1.6447368421052633,
|
| 160 |
+
"eval_cer": 0.13915694313045307,
|
| 161 |
+
"eval_loss": 0.6634953022003174,
|
| 162 |
+
"eval_mer": 0.6329740751529275,
|
| 163 |
+
"eval_runtime": 78.4056,
|
| 164 |
+
"eval_samples_per_second": 15.509,
|
| 165 |
+
"eval_steps_per_second": 1.939,
|
| 166 |
+
"eval_wer": 0.6578867696033909,
|
| 167 |
+
"eval_wil": 0.849560578731243,
|
| 168 |
+
"eval_wip": 0.15043942126875695,
|
| 169 |
+
"step": 4000
|
| 170 |
+
},
|
| 171 |
+
{
|
| 172 |
+
"epoch": 1.850328947368421,
|
| 173 |
+
"grad_norm": 2.9152979850769043,
|
| 174 |
+
"learning_rate": 6.868279569892474e-05,
|
| 175 |
+
"loss": 1.0214,
|
| 176 |
+
"step": 4500
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 1.850328947368421,
|
| 180 |
+
"eval_cer": 0.12391464311994113,
|
| 181 |
+
"eval_loss": 0.5948176383972168,
|
| 182 |
+
"eval_mer": 0.5913916786226686,
|
| 183 |
+
"eval_runtime": 78.3397,
|
| 184 |
+
"eval_samples_per_second": 15.522,
|
| 185 |
+
"eval_steps_per_second": 1.94,
|
| 186 |
+
"eval_wer": 0.6239782016348774,
|
| 187 |
+
"eval_wil": 0.8145258407398566,
|
| 188 |
+
"eval_wip": 0.18547415926014343,
|
| 189 |
+
"step": 4500
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 2.0559210526315788,
|
| 193 |
+
"grad_norm": 2.9141180515289307,
|
| 194 |
+
"learning_rate": 6.420250896057348e-05,
|
| 195 |
+
"loss": 0.9568,
|
| 196 |
+
"step": 5000
|
| 197 |
+
},
|
| 198 |
+
{
|
| 199 |
+
"epoch": 2.0559210526315788,
|
| 200 |
+
"eval_cer": 0.11611479028697572,
|
| 201 |
+
"eval_loss": 0.5608509182929993,
|
| 202 |
+
"eval_mer": 0.572582892170521,
|
| 203 |
+
"eval_runtime": 78.5329,
|
| 204 |
+
"eval_samples_per_second": 15.484,
|
| 205 |
+
"eval_steps_per_second": 1.935,
|
| 206 |
+
"eval_wer": 0.6221616712079927,
|
| 207 |
+
"eval_wil": 0.7951610373200031,
|
| 208 |
+
"eval_wip": 0.20483896267999693,
|
| 209 |
+
"step": 5000
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 2.2615131578947367,
|
| 213 |
+
"grad_norm": 2.146526575088501,
|
| 214 |
+
"learning_rate": 5.972222222222223e-05,
|
| 215 |
+
"loss": 0.9028,
|
| 216 |
+
"step": 5500
|
| 217 |
+
},
|
| 218 |
+
{
|
| 219 |
+
"epoch": 2.2615131578947367,
|
| 220 |
+
"eval_cer": 0.10854619993692842,
|
| 221 |
+
"eval_loss": 0.513049840927124,
|
| 222 |
+
"eval_mer": 0.5546070061819253,
|
| 223 |
+
"eval_runtime": 78.5555,
|
| 224 |
+
"eval_samples_per_second": 15.479,
|
| 225 |
+
"eval_steps_per_second": 1.935,
|
| 226 |
+
"eval_wer": 0.5703905540417802,
|
| 227 |
+
"eval_wil": 0.780121344078833,
|
| 228 |
+
"eval_wip": 0.2198786559211669,
|
| 229 |
+
"step": 5500
|
| 230 |
+
},
|
| 231 |
+
{
|
| 232 |
+
"epoch": 2.4671052631578947,
|
| 233 |
+
"grad_norm": 3.1563515663146973,
|
| 234 |
+
"learning_rate": 5.5241935483870966e-05,
|
| 235 |
+
"loss": 0.8706,
|
| 236 |
+
"step": 6000
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 2.4671052631578947,
|
| 240 |
+
"eval_cer": 0.10186061179438663,
|
| 241 |
+
"eval_loss": 0.48784276843070984,
|
| 242 |
+
"eval_mer": 0.5253894979803808,
|
| 243 |
+
"eval_runtime": 78.4958,
|
| 244 |
+
"eval_samples_per_second": 15.491,
|
| 245 |
+
"eval_steps_per_second": 1.936,
|
| 246 |
+
"eval_wer": 0.5513169845594914,
|
| 247 |
+
"eval_wil": 0.7534568789040759,
|
| 248 |
+
"eval_wip": 0.24654312109592416,
|
| 249 |
+
"step": 6000
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 2.6726973684210527,
|
| 253 |
+
"grad_norm": 1.8134748935699463,
|
| 254 |
+
"learning_rate": 5.076164874551972e-05,
|
| 255 |
+
"loss": 0.8143,
|
| 256 |
+
"step": 6500
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"epoch": 2.6726973684210527,
|
| 260 |
+
"eval_cer": 0.09692000420477241,
|
| 261 |
+
"eval_loss": 0.4577435553073883,
|
| 262 |
+
"eval_mer": 0.5118413262285376,
|
| 263 |
+
"eval_runtime": 78.5987,
|
| 264 |
+
"eval_samples_per_second": 15.471,
|
| 265 |
+
"eval_steps_per_second": 1.934,
|
| 266 |
+
"eval_wer": 0.5234635180139268,
|
| 267 |
+
"eval_wil": 0.7379847574095497,
|
| 268 |
+
"eval_wip": 0.2620152425904502,
|
| 269 |
+
"step": 6500
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 2.8782894736842106,
|
| 273 |
+
"grad_norm": 1.6383355855941772,
|
| 274 |
+
"learning_rate": 4.628136200716846e-05,
|
| 275 |
+
"loss": 0.804,
|
| 276 |
+
"step": 7000
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 2.8782894736842106,
|
| 280 |
+
"eval_cer": 0.09395563965100388,
|
| 281 |
+
"eval_loss": 0.43608564138412476,
|
| 282 |
+
"eval_mer": 0.4900351699882767,
|
| 283 |
+
"eval_runtime": 78.5125,
|
| 284 |
+
"eval_samples_per_second": 15.488,
|
| 285 |
+
"eval_steps_per_second": 1.936,
|
| 286 |
+
"eval_wer": 0.5062064789585226,
|
| 287 |
+
"eval_wil": 0.7185688506582761,
|
| 288 |
+
"eval_wip": 0.28143114934172386,
|
| 289 |
+
"step": 7000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 3.0838815789473686,
|
| 293 |
+
"grad_norm": 2.0129234790802,
|
| 294 |
+
"learning_rate": 4.180107526881721e-05,
|
| 295 |
+
"loss": 0.7903,
|
| 296 |
+
"step": 7500
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 3.0838815789473686,
|
| 300 |
+
"eval_cer": 0.08865762640597077,
|
| 301 |
+
"eval_loss": 0.4281825125217438,
|
| 302 |
+
"eval_mer": 0.4752102058567701,
|
| 303 |
+
"eval_runtime": 78.181,
|
| 304 |
+
"eval_samples_per_second": 15.554,
|
| 305 |
+
"eval_steps_per_second": 1.944,
|
| 306 |
+
"eval_wer": 0.496215561610657,
|
| 307 |
+
"eval_wil": 0.7013381847075779,
|
| 308 |
+
"eval_wip": 0.2986618152924221,
|
| 309 |
+
"step": 7500
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 3.2894736842105265,
|
| 313 |
+
"grad_norm": 1.691244125366211,
|
| 314 |
+
"learning_rate": 3.732078853046595e-05,
|
| 315 |
+
"loss": 0.7453,
|
| 316 |
+
"step": 8000
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 3.2894736842105265,
|
| 320 |
+
"eval_cer": 0.08705981288762746,
|
| 321 |
+
"eval_loss": 0.4145726263523102,
|
| 322 |
+
"eval_mer": 0.4670487106017192,
|
| 323 |
+
"eval_runtime": 78.5359,
|
| 324 |
+
"eval_samples_per_second": 15.483,
|
| 325 |
+
"eval_steps_per_second": 1.935,
|
| 326 |
+
"eval_wer": 0.49349076597033,
|
| 327 |
+
"eval_wil": 0.6922093904961242,
|
| 328 |
+
"eval_wip": 0.3077906095038758,
|
| 329 |
+
"step": 8000
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 3.495065789473684,
|
| 333 |
+
"grad_norm": 1.3035862445831299,
|
| 334 |
+
"learning_rate": 3.28494623655914e-05,
|
| 335 |
+
"loss": 0.729,
|
| 336 |
+
"step": 8500
|
| 337 |
+
},
|
| 338 |
+
{
|
| 339 |
+
"epoch": 3.495065789473684,
|
| 340 |
+
"eval_cer": 0.08359087564385578,
|
| 341 |
+
"eval_loss": 0.3914414346218109,
|
| 342 |
+
"eval_mer": 0.44933920704845814,
|
| 343 |
+
"eval_runtime": 78.6019,
|
| 344 |
+
"eval_samples_per_second": 15.47,
|
| 345 |
+
"eval_steps_per_second": 1.934,
|
| 346 |
+
"eval_wer": 0.46321525885558584,
|
| 347 |
+
"eval_wil": 0.6742046841130319,
|
| 348 |
+
"eval_wip": 0.3257953158869681,
|
| 349 |
+
"step": 8500
|
| 350 |
+
},
|
| 351 |
+
{
|
| 352 |
+
"epoch": 3.700657894736842,
|
| 353 |
+
"grad_norm": 1.044214129447937,
|
| 354 |
+
"learning_rate": 2.8369175627240147e-05,
|
| 355 |
+
"loss": 0.726,
|
| 356 |
+
"step": 9000
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 3.700657894736842,
|
| 360 |
+
"eval_cer": 0.08087879743508883,
|
| 361 |
+
"eval_loss": 0.38009122014045715,
|
| 362 |
+
"eval_mer": 0.44460556844547566,
|
| 363 |
+
"eval_runtime": 78.8304,
|
| 364 |
+
"eval_samples_per_second": 15.426,
|
| 365 |
+
"eval_steps_per_second": 1.928,
|
| 366 |
+
"eval_wer": 0.4641235240690282,
|
| 367 |
+
"eval_wil": 0.66887236331981,
|
| 368 |
+
"eval_wip": 0.33112763668018996,
|
| 369 |
+
"step": 9000
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 3.90625,
|
| 373 |
+
"grad_norm": 1.4717159271240234,
|
| 374 |
+
"learning_rate": 2.3888888888888892e-05,
|
| 375 |
+
"loss": 0.6771,
|
| 376 |
+
"step": 9500
|
| 377 |
+
},
|
| 378 |
+
{
|
| 379 |
+
"epoch": 3.90625,
|
| 380 |
+
"eval_cer": 0.07873436350257543,
|
| 381 |
+
"eval_loss": 0.36759230494499207,
|
| 382 |
+
"eval_mer": 0.4377556984219754,
|
| 383 |
+
"eval_runtime": 78.6773,
|
| 384 |
+
"eval_samples_per_second": 15.456,
|
| 385 |
+
"eval_steps_per_second": 1.932,
|
| 386 |
+
"eval_wer": 0.4535270965788677,
|
| 387 |
+
"eval_wil": 0.6607955640884103,
|
| 388 |
+
"eval_wip": 0.33920443591158966,
|
| 389 |
+
"step": 9500
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 4.1118421052631575,
|
| 393 |
+
"grad_norm": 1.0172888040542603,
|
| 394 |
+
"learning_rate": 1.941756272401434e-05,
|
| 395 |
+
"loss": 0.6777,
|
| 396 |
+
"step": 10000
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 4.1118421052631575,
|
| 400 |
+
"eval_cer": 0.07698938294964786,
|
| 401 |
+
"eval_loss": 0.3616805672645569,
|
| 402 |
+
"eval_mer": 0.42764323685764916,
|
| 403 |
+
"eval_runtime": 78.6213,
|
| 404 |
+
"eval_samples_per_second": 15.467,
|
| 405 |
+
"eval_steps_per_second": 1.933,
|
| 406 |
+
"eval_wer": 0.4383893430214956,
|
| 407 |
+
"eval_wil": 0.6509819467652879,
|
| 408 |
+
"eval_wip": 0.34901805323471213,
|
| 409 |
+
"step": 10000
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 4.317434210526316,
|
| 413 |
+
"grad_norm": 1.7192729711532593,
|
| 414 |
+
"learning_rate": 1.4937275985663083e-05,
|
| 415 |
+
"loss": 0.6671,
|
| 416 |
+
"step": 10500
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 4.317434210526316,
|
| 420 |
+
"eval_cer": 0.07656890570797856,
|
| 421 |
+
"eval_loss": 0.35535532236099243,
|
| 422 |
+
"eval_mer": 0.42558208075449455,
|
| 423 |
+
"eval_runtime": 78.649,
|
| 424 |
+
"eval_samples_per_second": 15.461,
|
| 425 |
+
"eval_steps_per_second": 1.933,
|
| 426 |
+
"eval_wer": 0.4371783227369058,
|
| 427 |
+
"eval_wil": 0.6475495813620156,
|
| 428 |
+
"eval_wip": 0.35245041863798443,
|
| 429 |
+
"step": 10500
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 4.5230263157894735,
|
| 433 |
+
"grad_norm": 1.8435664176940918,
|
| 434 |
+
"learning_rate": 1.0456989247311828e-05,
|
| 435 |
+
"loss": 0.6609,
|
| 436 |
+
"step": 11000
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 4.5230263157894735,
|
| 440 |
+
"eval_cer": 0.07555976032797225,
|
| 441 |
+
"eval_loss": 0.3524816632270813,
|
| 442 |
+
"eval_mer": 0.41701000588581516,
|
| 443 |
+
"eval_runtime": 78.6898,
|
| 444 |
+
"eval_samples_per_second": 15.453,
|
| 445 |
+
"eval_steps_per_second": 1.932,
|
| 446 |
+
"eval_wer": 0.42900393581592494,
|
| 447 |
+
"eval_wil": 0.6368825813682012,
|
| 448 |
+
"eval_wip": 0.3631174186317988,
|
| 449 |
+
"step": 11000
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 4.728618421052632,
|
| 453 |
+
"grad_norm": 2.3937132358551025,
|
| 454 |
+
"learning_rate": 5.9767025089605736e-06,
|
| 455 |
+
"loss": 0.6676,
|
| 456 |
+
"step": 11500
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 4.728618421052632,
|
| 460 |
+
"eval_cer": 0.07459266267213287,
|
| 461 |
+
"eval_loss": 0.34498506784439087,
|
| 462 |
+
"eval_mer": 0.4086496028243601,
|
| 463 |
+
"eval_runtime": 78.5242,
|
| 464 |
+
"eval_samples_per_second": 15.486,
|
| 465 |
+
"eval_steps_per_second": 1.936,
|
| 466 |
+
"eval_wer": 0.42052679382379654,
|
| 467 |
+
"eval_wil": 0.6279924686913689,
|
| 468 |
+
"eval_wip": 0.3720075313086311,
|
| 469 |
+
"step": 11500
|
| 470 |
+
},
|
| 471 |
+
{
|
| 472 |
+
"epoch": 4.934210526315789,
|
| 473 |
+
"grad_norm": 3.0196588039398193,
|
| 474 |
+
"learning_rate": 1.496415770609319e-06,
|
| 475 |
+
"loss": 0.6569,
|
| 476 |
+
"step": 12000
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 4.934210526315789,
|
| 480 |
+
"eval_cer": 0.07417218543046358,
|
| 481 |
+
"eval_loss": 0.3443602919578552,
|
| 482 |
+
"eval_mer": 0.4107090320682554,
|
| 483 |
+
"eval_runtime": 78.8185,
|
| 484 |
+
"eval_samples_per_second": 15.428,
|
| 485 |
+
"eval_steps_per_second": 1.928,
|
| 486 |
+
"eval_wer": 0.4226460793218286,
|
| 487 |
+
"eval_wil": 0.6302416833366282,
|
| 488 |
+
"eval_wip": 0.36975831666337183,
|
| 489 |
+
"step": 12000
|
| 490 |
+
}
|
| 491 |
+
],
|
| 492 |
+
"logging_steps": 500,
|
| 493 |
+
"max_steps": 12160,
|
| 494 |
+
"num_input_tokens_seen": 0,
|
| 495 |
+
"num_train_epochs": 5,
|
| 496 |
+
"save_steps": 500,
|
| 497 |
+
"stateful_callbacks": {
|
| 498 |
+
"TrainerControl": {
|
| 499 |
+
"args": {
|
| 500 |
+
"should_epoch_stop": false,
|
| 501 |
+
"should_evaluate": false,
|
| 502 |
+
"should_log": false,
|
| 503 |
+
"should_save": true,
|
| 504 |
+
"should_training_stop": true
|
| 505 |
+
},
|
| 506 |
+
"attributes": {}
|
| 507 |
+
}
|
| 508 |
+
},
|
| 509 |
+
"total_flos": 1.6656507256579078e+19,
|
| 510 |
+
"train_batch_size": 10,
|
| 511 |
+
"trial_name": null,
|
| 512 |
+
"trial_params": null
|
| 513 |
+
}
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5176
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9018b68a510168951a4e826c5a3788d54ef8e5ddff6ab30e3ba92dd6431f9a20
|
| 3 |
size 5176
|
vocab.json
CHANGED
|
@@ -704,5 +704,6 @@
|
|
| 704 |
"7": 702,
|
| 705 |
"8": 703,
|
| 706 |
"9": 704,
|
| 707 |
-
"0": 705
|
|
|
|
| 708 |
}
|
|
|
|
| 704 |
"7": 702,
|
| 705 |
"8": 703,
|
| 706 |
"9": 704,
|
| 707 |
+
"0": 705,
|
| 708 |
+
" ": 706
|
| 709 |
}
|