End of training
Browse files- all_results.json +27 -27
- eval_results.json +21 -21
- train_results.json +6 -6
- trainer_state.json +318 -94
all_results.json
CHANGED
|
@@ -1,30 +1,30 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 4.0,
|
| 3 |
-
"eval_avg_cer": 0.
|
| 4 |
-
"eval_avg_wer":
|
| 5 |
-
"eval_gilaki_cer": 0.
|
| 6 |
-
"eval_gilaki_wer":
|
| 7 |
-
"eval_hawrami_cer": 0.
|
| 8 |
-
"eval_hawrami_wer":
|
| 9 |
-
"eval_laki_kurdish_cer": 0.
|
| 10 |
-
"eval_laki_kurdish_wer":
|
| 11 |
-
"eval_loss":
|
| 12 |
-
"eval_mazanderani_cer": 0.
|
| 13 |
-
"eval_mazanderani_wer":
|
| 14 |
-
"eval_runtime":
|
| 15 |
-
"eval_samples":
|
| 16 |
-
"eval_samples_per_second": 7.
|
| 17 |
-
"eval_southern_kurdish_cer":
|
| 18 |
-
"eval_southern_kurdish_wer":
|
| 19 |
-
"eval_steps_per_second": 0.
|
| 20 |
-
"eval_talysh_cer": 0.
|
| 21 |
-
"eval_talysh_wer":
|
| 22 |
-
"eval_zazaki_cer": 0.
|
| 23 |
-
"eval_zazaki_wer":
|
| 24 |
-
"total_flos":
|
| 25 |
-
"train_loss":
|
| 26 |
-
"train_runtime":
|
| 27 |
-
"train_samples":
|
| 28 |
-
"train_samples_per_second":
|
| 29 |
-
"train_steps_per_second": 0.
|
| 30 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 4.0,
|
| 3 |
+
"eval_avg_cer": 0.2678389158869433,
|
| 4 |
+
"eval_avg_wer": 0.7003291969579674,
|
| 5 |
+
"eval_gilaki_cer": 0.38182211642501623,
|
| 6 |
+
"eval_gilaki_wer": 0.9606072252113759,
|
| 7 |
+
"eval_hawrami_cer": 0.08648390486197345,
|
| 8 |
+
"eval_hawrami_wer": 0.4144927536231884,
|
| 9 |
+
"eval_laki_kurdish_cer": 0.16528387831462502,
|
| 10 |
+
"eval_laki_kurdish_wer": 0.603680981595092,
|
| 11 |
+
"eval_loss": 0.5420735478401184,
|
| 12 |
+
"eval_mazanderani_cer": 0.2510331798323297,
|
| 13 |
+
"eval_mazanderani_wer": 0.6836474783494652,
|
| 14 |
+
"eval_runtime": 458.9768,
|
| 15 |
+
"eval_samples": 3258,
|
| 16 |
+
"eval_samples_per_second": 7.098,
|
| 17 |
+
"eval_southern_kurdish_cer": 0.17209076399972242,
|
| 18 |
+
"eval_southern_kurdish_wer": 0.5609543712991989,
|
| 19 |
+
"eval_steps_per_second": 0.057,
|
| 20 |
+
"eval_talysh_cer": 0.5,
|
| 21 |
+
"eval_talysh_wer": 0.9166666666666666,
|
| 22 |
+
"eval_zazaki_cer": 0.31815856777493606,
|
| 23 |
+
"eval_zazaki_wer": 0.7622549019607843,
|
| 24 |
+
"total_flos": 5.41529079939072e+18,
|
| 25 |
+
"train_loss": 0.6912094768954486,
|
| 26 |
+
"train_runtime": 12274.8821,
|
| 27 |
+
"train_samples": 20873,
|
| 28 |
+
"train_samples_per_second": 6.802,
|
| 29 |
+
"train_steps_per_second": 0.053
|
| 30 |
}
|
eval_results.json
CHANGED
|
@@ -1,24 +1,24 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 4.0,
|
| 3 |
-
"eval_avg_cer": 0.
|
| 4 |
-
"eval_avg_wer":
|
| 5 |
-
"eval_gilaki_cer": 0.
|
| 6 |
-
"eval_gilaki_wer":
|
| 7 |
-
"eval_hawrami_cer": 0.
|
| 8 |
-
"eval_hawrami_wer":
|
| 9 |
-
"eval_laki_kurdish_cer": 0.
|
| 10 |
-
"eval_laki_kurdish_wer":
|
| 11 |
-
"eval_loss":
|
| 12 |
-
"eval_mazanderani_cer": 0.
|
| 13 |
-
"eval_mazanderani_wer":
|
| 14 |
-
"eval_runtime":
|
| 15 |
-
"eval_samples":
|
| 16 |
-
"eval_samples_per_second": 7.
|
| 17 |
-
"eval_southern_kurdish_cer":
|
| 18 |
-
"eval_southern_kurdish_wer":
|
| 19 |
-
"eval_steps_per_second": 0.
|
| 20 |
-
"eval_talysh_cer": 0.
|
| 21 |
-
"eval_talysh_wer":
|
| 22 |
-
"eval_zazaki_cer": 0.
|
| 23 |
-
"eval_zazaki_wer":
|
| 24 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 4.0,
|
| 3 |
+
"eval_avg_cer": 0.2678389158869433,
|
| 4 |
+
"eval_avg_wer": 0.7003291969579674,
|
| 5 |
+
"eval_gilaki_cer": 0.38182211642501623,
|
| 6 |
+
"eval_gilaki_wer": 0.9606072252113759,
|
| 7 |
+
"eval_hawrami_cer": 0.08648390486197345,
|
| 8 |
+
"eval_hawrami_wer": 0.4144927536231884,
|
| 9 |
+
"eval_laki_kurdish_cer": 0.16528387831462502,
|
| 10 |
+
"eval_laki_kurdish_wer": 0.603680981595092,
|
| 11 |
+
"eval_loss": 0.5420735478401184,
|
| 12 |
+
"eval_mazanderani_cer": 0.2510331798323297,
|
| 13 |
+
"eval_mazanderani_wer": 0.6836474783494652,
|
| 14 |
+
"eval_runtime": 458.9768,
|
| 15 |
+
"eval_samples": 3258,
|
| 16 |
+
"eval_samples_per_second": 7.098,
|
| 17 |
+
"eval_southern_kurdish_cer": 0.17209076399972242,
|
| 18 |
+
"eval_southern_kurdish_wer": 0.5609543712991989,
|
| 19 |
+
"eval_steps_per_second": 0.057,
|
| 20 |
+
"eval_talysh_cer": 0.5,
|
| 21 |
+
"eval_talysh_wer": 0.9166666666666666,
|
| 22 |
+
"eval_zazaki_cer": 0.31815856777493606,
|
| 23 |
+
"eval_zazaki_wer": 0.7622549019607843
|
| 24 |
}
|
train_results.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"epoch": 4.0,
|
| 3 |
-
"total_flos":
|
| 4 |
-
"train_loss":
|
| 5 |
-
"train_runtime":
|
| 6 |
-
"train_samples":
|
| 7 |
-
"train_samples_per_second":
|
| 8 |
-
"train_steps_per_second": 0.
|
| 9 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"epoch": 4.0,
|
| 3 |
+
"total_flos": 5.41529079939072e+18,
|
| 4 |
+
"train_loss": 0.6912094768954486,
|
| 5 |
+
"train_runtime": 12274.8821,
|
| 6 |
+
"train_samples": 20873,
|
| 7 |
+
"train_samples_per_second": 6.802,
|
| 8 |
+
"train_steps_per_second": 0.053
|
| 9 |
}
|
trainer_state.json
CHANGED
|
@@ -3,119 +3,343 @@
|
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
"epoch": 4.0,
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 11 |
{
|
| 12 |
"epoch": 1.0,
|
| 13 |
-
"eval_avg_cer": 0.
|
| 14 |
-
"eval_avg_wer":
|
| 15 |
-
"eval_gilaki_cer":
|
| 16 |
-
"eval_gilaki_wer":
|
| 17 |
-
"eval_hawrami_cer": 0.
|
| 18 |
-
"eval_hawrami_wer":
|
| 19 |
-
"eval_laki_kurdish_cer": 0.
|
| 20 |
-
"eval_laki_kurdish_wer":
|
| 21 |
-
"eval_loss":
|
| 22 |
-
"eval_mazanderani_cer":
|
| 23 |
-
"eval_mazanderani_wer":
|
| 24 |
-
"eval_runtime":
|
| 25 |
-
"eval_samples_per_second": 7.
|
| 26 |
-
"eval_southern_kurdish_cer":
|
| 27 |
-
"eval_southern_kurdish_wer":
|
| 28 |
-
"eval_steps_per_second": 0.
|
| 29 |
-
"eval_talysh_cer": 0.
|
| 30 |
-
"eval_talysh_wer":
|
| 31 |
-
"eval_zazaki_cer": 0.
|
| 32 |
-
"eval_zazaki_wer":
|
| 33 |
-
"step":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 34 |
},
|
| 35 |
{
|
| 36 |
"epoch": 2.0,
|
| 37 |
-
"eval_avg_cer": 0.
|
| 38 |
-
"eval_avg_wer":
|
| 39 |
-
"eval_gilaki_cer": 0.
|
| 40 |
-
"eval_gilaki_wer":
|
| 41 |
-
"eval_hawrami_cer": 0.
|
| 42 |
-
"eval_hawrami_wer":
|
| 43 |
-
"eval_laki_kurdish_cer": 0.
|
| 44 |
-
"eval_laki_kurdish_wer":
|
| 45 |
-
"eval_loss":
|
| 46 |
-
"eval_mazanderani_cer":
|
| 47 |
-
"eval_mazanderani_wer":
|
| 48 |
-
"eval_runtime":
|
| 49 |
-
"eval_samples_per_second": 7.
|
| 50 |
-
"eval_southern_kurdish_cer":
|
| 51 |
-
"eval_southern_kurdish_wer":
|
| 52 |
-
"eval_steps_per_second": 0.
|
| 53 |
-
"eval_talysh_cer": 0.
|
| 54 |
-
"eval_talysh_wer":
|
| 55 |
-
"eval_zazaki_cer": 0.
|
| 56 |
-
"eval_zazaki_wer":
|
| 57 |
-
"step":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 58 |
},
|
| 59 |
{
|
| 60 |
"epoch": 3.0,
|
| 61 |
-
"eval_avg_cer": 0.
|
| 62 |
-
"eval_avg_wer":
|
| 63 |
-
"eval_gilaki_cer": 0.
|
| 64 |
-
"eval_gilaki_wer":
|
| 65 |
-
"eval_hawrami_cer": 0.
|
| 66 |
-
"eval_hawrami_wer":
|
| 67 |
-
"eval_laki_kurdish_cer": 0.
|
| 68 |
-
"eval_laki_kurdish_wer":
|
| 69 |
-
"eval_loss":
|
| 70 |
-
"eval_mazanderani_cer":
|
| 71 |
-
"eval_mazanderani_wer":
|
| 72 |
-
"eval_runtime":
|
| 73 |
-
"eval_samples_per_second": 7.
|
| 74 |
-
"eval_southern_kurdish_cer":
|
| 75 |
-
"eval_southern_kurdish_wer":
|
| 76 |
-
"eval_steps_per_second": 0.
|
| 77 |
-
"eval_talysh_cer": 0.
|
| 78 |
-
"eval_talysh_wer":
|
| 79 |
-
"eval_zazaki_cer": 0.
|
| 80 |
-
"eval_zazaki_wer":
|
| 81 |
-
"step":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 82 |
},
|
| 83 |
{
|
| 84 |
"epoch": 4.0,
|
| 85 |
-
"eval_avg_cer": 0.
|
| 86 |
-
"eval_avg_wer":
|
| 87 |
-
"eval_gilaki_cer": 0.
|
| 88 |
-
"eval_gilaki_wer":
|
| 89 |
-
"eval_hawrami_cer": 0.
|
| 90 |
-
"eval_hawrami_wer":
|
| 91 |
-
"eval_laki_kurdish_cer": 0.
|
| 92 |
-
"eval_laki_kurdish_wer":
|
| 93 |
-
"eval_loss":
|
| 94 |
-
"eval_mazanderani_cer": 0.
|
| 95 |
-
"eval_mazanderani_wer":
|
| 96 |
-
"eval_runtime":
|
| 97 |
-
"eval_samples_per_second": 7.
|
| 98 |
-
"eval_southern_kurdish_cer":
|
| 99 |
-
"eval_southern_kurdish_wer":
|
| 100 |
-
"eval_steps_per_second": 0.
|
| 101 |
-
"eval_talysh_cer": 0.
|
| 102 |
-
"eval_talysh_wer":
|
| 103 |
-
"eval_zazaki_cer": 0.
|
| 104 |
-
"eval_zazaki_wer":
|
| 105 |
-
"step":
|
| 106 |
},
|
| 107 |
{
|
| 108 |
"epoch": 4.0,
|
| 109 |
-
"step":
|
| 110 |
-
"total_flos":
|
| 111 |
-
"train_loss":
|
| 112 |
-
"train_runtime":
|
| 113 |
-
"train_samples_per_second":
|
| 114 |
-
"train_steps_per_second": 0.
|
| 115 |
}
|
| 116 |
],
|
| 117 |
"logging_steps": 20,
|
| 118 |
-
"max_steps":
|
| 119 |
"num_input_tokens_seen": 0,
|
| 120 |
"num_train_epochs": 4,
|
| 121 |
"save_steps": 500,
|
|
@@ -131,8 +355,8 @@
|
|
| 131 |
"attributes": {}
|
| 132 |
}
|
| 133 |
},
|
| 134 |
-
"total_flos":
|
| 135 |
-
"train_batch_size":
|
| 136 |
"trial_name": null,
|
| 137 |
"trial_params": null
|
| 138 |
}
|
|
|
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
"epoch": 4.0,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 656,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.12195121951219512,
|
| 13 |
+
"grad_norm": 42.956241607666016,
|
| 14 |
+
"learning_rate": 1.5e-06,
|
| 15 |
+
"loss": 3.946,
|
| 16 |
+
"step": 20
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.24390243902439024,
|
| 20 |
+
"grad_norm": 18.780467987060547,
|
| 21 |
+
"learning_rate": 3.5e-06,
|
| 22 |
+
"loss": 2.801,
|
| 23 |
+
"step": 40
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.36585365853658536,
|
| 27 |
+
"grad_norm": 10.132372856140137,
|
| 28 |
+
"learning_rate": 5.500000000000001e-06,
|
| 29 |
+
"loss": 1.829,
|
| 30 |
+
"step": 60
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.4878048780487805,
|
| 34 |
+
"grad_norm": 6.34807825088501,
|
| 35 |
+
"learning_rate": 7.500000000000001e-06,
|
| 36 |
+
"loss": 1.1664,
|
| 37 |
+
"step": 80
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.6097560975609756,
|
| 41 |
+
"grad_norm": 3.893784523010254,
|
| 42 |
+
"learning_rate": 9.5e-06,
|
| 43 |
+
"loss": 0.8395,
|
| 44 |
+
"step": 100
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.7317073170731707,
|
| 48 |
+
"grad_norm": 3.6245577335357666,
|
| 49 |
+
"learning_rate": 9.73021582733813e-06,
|
| 50 |
+
"loss": 0.6985,
|
| 51 |
+
"step": 120
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.8536585365853658,
|
| 55 |
+
"grad_norm": 3.2853732109069824,
|
| 56 |
+
"learning_rate": 9.370503597122303e-06,
|
| 57 |
+
"loss": 0.6198,
|
| 58 |
+
"step": 140
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.975609756097561,
|
| 62 |
+
"grad_norm": 3.1876819133758545,
|
| 63 |
+
"learning_rate": 9.010791366906476e-06,
|
| 64 |
+
"loss": 0.5933,
|
| 65 |
+
"step": 160
|
| 66 |
+
},
|
| 67 |
{
|
| 68 |
"epoch": 1.0,
|
| 69 |
+
"eval_avg_cer": 0.35838623267570763,
|
| 70 |
+
"eval_avg_wer": 0.8217081943985028,
|
| 71 |
+
"eval_gilaki_cer": 0.38532785111447737,
|
| 72 |
+
"eval_gilaki_wer": 0.9969254419677172,
|
| 73 |
+
"eval_hawrami_cer": 0.11622886617633012,
|
| 74 |
+
"eval_hawrami_wer": 0.5257112184648417,
|
| 75 |
+
"eval_laki_kurdish_cer": 0.24613632666341304,
|
| 76 |
+
"eval_laki_kurdish_wer": 0.7856850715746422,
|
| 77 |
+
"eval_loss": 0.7336705923080444,
|
| 78 |
+
"eval_mazanderani_cer": 0.3138505136379738,
|
| 79 |
+
"eval_mazanderani_wer": 0.8884360672440142,
|
| 80 |
+
"eval_runtime": 457.0137,
|
| 81 |
+
"eval_samples_per_second": 7.129,
|
| 82 |
+
"eval_southern_kurdish_cer": 0.21365623482062313,
|
| 83 |
+
"eval_southern_kurdish_wer": 0.6900034831069314,
|
| 84 |
+
"eval_steps_per_second": 0.057,
|
| 85 |
+
"eval_talysh_cer": 0.5,
|
| 86 |
+
"eval_talysh_wer": 0.9166666666666666,
|
| 87 |
+
"eval_zazaki_cer": 0.7335038363171356,
|
| 88 |
+
"eval_zazaki_wer": 0.9485294117647058,
|
| 89 |
+
"step": 164
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 1.0975609756097562,
|
| 93 |
+
"grad_norm": 3.3254923820495605,
|
| 94 |
+
"learning_rate": 8.651079136690648e-06,
|
| 95 |
+
"loss": 0.5337,
|
| 96 |
+
"step": 180
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 1.2195121951219512,
|
| 100 |
+
"grad_norm": 3.416600227355957,
|
| 101 |
+
"learning_rate": 8.29136690647482e-06,
|
| 102 |
+
"loss": 0.5054,
|
| 103 |
+
"step": 200
|
| 104 |
+
},
|
| 105 |
+
{
|
| 106 |
+
"epoch": 1.3414634146341464,
|
| 107 |
+
"grad_norm": 3.544034481048584,
|
| 108 |
+
"learning_rate": 7.931654676258992e-06,
|
| 109 |
+
"loss": 0.4915,
|
| 110 |
+
"step": 220
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 1.4634146341463414,
|
| 114 |
+
"grad_norm": 2.7927117347717285,
|
| 115 |
+
"learning_rate": 7.571942446043166e-06,
|
| 116 |
+
"loss": 0.47,
|
| 117 |
+
"step": 240
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 1.5853658536585367,
|
| 121 |
+
"grad_norm": 3.406677007675171,
|
| 122 |
+
"learning_rate": 7.212230215827338e-06,
|
| 123 |
+
"loss": 0.4681,
|
| 124 |
+
"step": 260
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 1.7073170731707317,
|
| 128 |
+
"grad_norm": 2.7277846336364746,
|
| 129 |
+
"learning_rate": 6.852517985611511e-06,
|
| 130 |
+
"loss": 0.4653,
|
| 131 |
+
"step": 280
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 1.8292682926829267,
|
| 135 |
+
"grad_norm": 2.8852195739746094,
|
| 136 |
+
"learning_rate": 6.4928057553956835e-06,
|
| 137 |
+
"loss": 0.4281,
|
| 138 |
+
"step": 300
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"epoch": 1.951219512195122,
|
| 142 |
+
"grad_norm": 3.2276437282562256,
|
| 143 |
+
"learning_rate": 6.133093525179856e-06,
|
| 144 |
+
"loss": 0.4436,
|
| 145 |
+
"step": 320
|
| 146 |
},
|
| 147 |
{
|
| 148 |
"epoch": 2.0,
|
| 149 |
+
"eval_avg_cer": 0.2926722864574391,
|
| 150 |
+
"eval_avg_wer": 0.7377513408917141,
|
| 151 |
+
"eval_gilaki_cer": 0.38182211642501623,
|
| 152 |
+
"eval_gilaki_wer": 0.9723289777094543,
|
| 153 |
+
"eval_hawrami_cer": 0.09332314452192186,
|
| 154 |
+
"eval_hawrami_wer": 0.44186795491143316,
|
| 155 |
+
"eval_laki_kurdish_cer": 0.1845615747519115,
|
| 156 |
+
"eval_laki_kurdish_wer": 0.6486707566462168,
|
| 157 |
+
"eval_loss": 0.5930325984954834,
|
| 158 |
+
"eval_mazanderani_cer": 0.29389538316212066,
|
| 159 |
+
"eval_mazanderani_wer": 0.7514009169638308,
|
| 160 |
+
"eval_runtime": 456.2009,
|
| 161 |
+
"eval_samples_per_second": 7.142,
|
| 162 |
+
"eval_southern_kurdish_cer": 0.17975851779890362,
|
| 163 |
+
"eval_southern_kurdish_wer": 0.5950888192267503,
|
| 164 |
+
"eval_steps_per_second": 0.057,
|
| 165 |
+
"eval_talysh_cer": 0.5,
|
| 166 |
+
"eval_talysh_wer": 0.9166666666666666,
|
| 167 |
+
"eval_zazaki_cer": 0.4153452685421995,
|
| 168 |
+
"eval_zazaki_wer": 0.8382352941176471,
|
| 169 |
+
"step": 328
|
| 170 |
+
},
|
| 171 |
+
{
|
| 172 |
+
"epoch": 2.073170731707317,
|
| 173 |
+
"grad_norm": 3.5604612827301025,
|
| 174 |
+
"learning_rate": 5.773381294964029e-06,
|
| 175 |
+
"loss": 0.4201,
|
| 176 |
+
"step": 340
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 2.1951219512195124,
|
| 180 |
+
"grad_norm": 2.8034543991088867,
|
| 181 |
+
"learning_rate": 5.413669064748202e-06,
|
| 182 |
+
"loss": 0.407,
|
| 183 |
+
"step": 360
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 2.317073170731707,
|
| 187 |
+
"grad_norm": 3.1773579120635986,
|
| 188 |
+
"learning_rate": 5.053956834532374e-06,
|
| 189 |
+
"loss": 0.3865,
|
| 190 |
+
"step": 380
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 2.4390243902439024,
|
| 194 |
+
"grad_norm": 2.6149208545684814,
|
| 195 |
+
"learning_rate": 4.6942446043165475e-06,
|
| 196 |
+
"loss": 0.4109,
|
| 197 |
+
"step": 400
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 2.5609756097560976,
|
| 201 |
+
"grad_norm": 2.5704352855682373,
|
| 202 |
+
"learning_rate": 4.33453237410072e-06,
|
| 203 |
+
"loss": 0.3879,
|
| 204 |
+
"step": 420
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 2.682926829268293,
|
| 208 |
+
"grad_norm": 2.887918472290039,
|
| 209 |
+
"learning_rate": 3.974820143884892e-06,
|
| 210 |
+
"loss": 0.3831,
|
| 211 |
+
"step": 440
|
| 212 |
+
},
|
| 213 |
+
{
|
| 214 |
+
"epoch": 2.8048780487804876,
|
| 215 |
+
"grad_norm": 2.9402122497558594,
|
| 216 |
+
"learning_rate": 3.6151079136690647e-06,
|
| 217 |
+
"loss": 0.3946,
|
| 218 |
+
"step": 460
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 2.926829268292683,
|
| 222 |
+
"grad_norm": 2.4520623683929443,
|
| 223 |
+
"learning_rate": 3.2553956834532375e-06,
|
| 224 |
+
"loss": 0.387,
|
| 225 |
+
"step": 480
|
| 226 |
},
|
| 227 |
{
|
| 228 |
"epoch": 3.0,
|
| 229 |
+
"eval_avg_cer": 0.2707400949444132,
|
| 230 |
+
"eval_avg_wer": 0.7061799250572226,
|
| 231 |
+
"eval_gilaki_cer": 0.3735122267907379,
|
| 232 |
+
"eval_gilaki_wer": 0.9523443504996156,
|
| 233 |
+
"eval_hawrami_cer": 0.08913936383608749,
|
| 234 |
+
"eval_hawrami_wer": 0.42297369833601717,
|
| 235 |
+
"eval_laki_kurdish_cer": 0.1678054335448186,
|
| 236 |
+
"eval_laki_kurdish_wer": 0.6094069529652352,
|
| 237 |
+
"eval_loss": 0.5509017109870911,
|
| 238 |
+
"eval_mazanderani_cer": 0.24359428503955602,
|
| 239 |
+
"eval_mazanderani_wer": 0.6821192052980133,
|
| 240 |
+
"eval_runtime": 459.5728,
|
| 241 |
+
"eval_samples_per_second": 7.089,
|
| 242 |
+
"eval_southern_kurdish_cer": 0.1871140101311498,
|
| 243 |
+
"eval_southern_kurdish_wer": 0.5705329153605015,
|
| 244 |
+
"eval_steps_per_second": 0.057,
|
| 245 |
+
"eval_talysh_cer": 0.5,
|
| 246 |
+
"eval_talysh_wer": 0.9166666666666666,
|
| 247 |
+
"eval_zazaki_cer": 0.3340153452685422,
|
| 248 |
+
"eval_zazaki_wer": 0.7892156862745098,
|
| 249 |
+
"step": 492
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 3.048780487804878,
|
| 253 |
+
"grad_norm": 2.502338409423828,
|
| 254 |
+
"learning_rate": 2.89568345323741e-06,
|
| 255 |
+
"loss": 0.3788,
|
| 256 |
+
"step": 500
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"epoch": 3.1707317073170733,
|
| 260 |
+
"grad_norm": 2.474639654159546,
|
| 261 |
+
"learning_rate": 2.5359712230215827e-06,
|
| 262 |
+
"loss": 0.3829,
|
| 263 |
+
"step": 520
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 3.292682926829268,
|
| 267 |
+
"grad_norm": 2.5425171852111816,
|
| 268 |
+
"learning_rate": 2.1762589928057555e-06,
|
| 269 |
+
"loss": 0.3557,
|
| 270 |
+
"step": 540
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 3.4146341463414633,
|
| 274 |
+
"grad_norm": 2.498109817504883,
|
| 275 |
+
"learning_rate": 1.8165467625899283e-06,
|
| 276 |
+
"loss": 0.3607,
|
| 277 |
+
"step": 560
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"epoch": 3.5365853658536586,
|
| 281 |
+
"grad_norm": 2.560028076171875,
|
| 282 |
+
"learning_rate": 1.4568345323741009e-06,
|
| 283 |
+
"loss": 0.3544,
|
| 284 |
+
"step": 580
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 3.658536585365854,
|
| 288 |
+
"grad_norm": 2.3217689990997314,
|
| 289 |
+
"learning_rate": 1.0971223021582735e-06,
|
| 290 |
+
"loss": 0.3614,
|
| 291 |
+
"step": 600
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 3.7804878048780486,
|
| 295 |
+
"grad_norm": 2.47906231880188,
|
| 296 |
+
"learning_rate": 7.37410071942446e-07,
|
| 297 |
+
"loss": 0.3608,
|
| 298 |
+
"step": 620
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 3.902439024390244,
|
| 302 |
+
"grad_norm": 2.4336979389190674,
|
| 303 |
+
"learning_rate": 3.7769784172661875e-07,
|
| 304 |
+
"loss": 0.3575,
|
| 305 |
+
"step": 640
|
| 306 |
},
|
| 307 |
{
|
| 308 |
"epoch": 4.0,
|
| 309 |
+
"eval_avg_cer": 0.2678389158869433,
|
| 310 |
+
"eval_avg_wer": 0.7003291969579674,
|
| 311 |
+
"eval_gilaki_cer": 0.38182211642501623,
|
| 312 |
+
"eval_gilaki_wer": 0.9606072252113759,
|
| 313 |
+
"eval_hawrami_cer": 0.08648390486197345,
|
| 314 |
+
"eval_hawrami_wer": 0.4144927536231884,
|
| 315 |
+
"eval_laki_kurdish_cer": 0.16528387831462502,
|
| 316 |
+
"eval_laki_kurdish_wer": 0.603680981595092,
|
| 317 |
+
"eval_loss": 0.5420735478401184,
|
| 318 |
+
"eval_mazanderani_cer": 0.2510331798323297,
|
| 319 |
+
"eval_mazanderani_wer": 0.6836474783494652,
|
| 320 |
+
"eval_runtime": 457.2923,
|
| 321 |
+
"eval_samples_per_second": 7.125,
|
| 322 |
+
"eval_southern_kurdish_cer": 0.17209076399972242,
|
| 323 |
+
"eval_southern_kurdish_wer": 0.5609543712991989,
|
| 324 |
+
"eval_steps_per_second": 0.057,
|
| 325 |
+
"eval_talysh_cer": 0.5,
|
| 326 |
+
"eval_talysh_wer": 0.9166666666666666,
|
| 327 |
+
"eval_zazaki_cer": 0.31815856777493606,
|
| 328 |
+
"eval_zazaki_wer": 0.7622549019607843,
|
| 329 |
+
"step": 656
|
| 330 |
},
|
| 331 |
{
|
| 332 |
"epoch": 4.0,
|
| 333 |
+
"step": 656,
|
| 334 |
+
"total_flos": 5.41529079939072e+18,
|
| 335 |
+
"train_loss": 0.6912094768954486,
|
| 336 |
+
"train_runtime": 12274.8821,
|
| 337 |
+
"train_samples_per_second": 6.802,
|
| 338 |
+
"train_steps_per_second": 0.053
|
| 339 |
}
|
| 340 |
],
|
| 341 |
"logging_steps": 20,
|
| 342 |
+
"max_steps": 656,
|
| 343 |
"num_input_tokens_seen": 0,
|
| 344 |
"num_train_epochs": 4,
|
| 345 |
"save_steps": 500,
|
|
|
|
| 355 |
"attributes": {}
|
| 356 |
}
|
| 357 |
},
|
| 358 |
+
"total_flos": 5.41529079939072e+18,
|
| 359 |
+
"train_batch_size": 128,
|
| 360 |
"trial_name": null,
|
| 361 |
"trial_params": null
|
| 362 |
}
|