Final training metrics
Browse files- training_metrics.json +132 -0
training_metrics.json
ADDED
|
@@ -0,0 +1,132 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"metrics_history": [
|
| 3 |
+
{
|
| 4 |
+
"epoch": 0.18409425625920472,
|
| 5 |
+
"global_step": 2000,
|
| 6 |
+
"eval_loss": 0.7301858067512512,
|
| 7 |
+
"eval_cosine_accuracy": 0.9443621635437012
|
| 8 |
+
},
|
| 9 |
+
{
|
| 10 |
+
"epoch": 0.36818851251840945,
|
| 11 |
+
"global_step": 4000,
|
| 12 |
+
"eval_loss": 0.7006688117980957,
|
| 13 |
+
"eval_cosine_accuracy": 0.9476411938667297
|
| 14 |
+
},
|
| 15 |
+
{
|
| 16 |
+
"epoch": 0.5522827687776142,
|
| 17 |
+
"global_step": 6000,
|
| 18 |
+
"eval_loss": 0.6458290815353394,
|
| 19 |
+
"eval_cosine_accuracy": 0.9529299736022949
|
| 20 |
+
},
|
| 21 |
+
{
|
| 22 |
+
"epoch": 0.7363770250368189,
|
| 23 |
+
"global_step": 8000,
|
| 24 |
+
"eval_loss": 0.6126231551170349,
|
| 25 |
+
"eval_cosine_accuracy": 0.958430290222168
|
| 26 |
+
},
|
| 27 |
+
{
|
| 28 |
+
"epoch": 0.9204712812960235,
|
| 29 |
+
"global_step": 10000,
|
| 30 |
+
"eval_loss": 0.5929833054542542,
|
| 31 |
+
"eval_cosine_accuracy": 0.9612862467765808
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 1.1044885945548197,
|
| 35 |
+
"global_step": 12000,
|
| 36 |
+
"eval_loss": 0.5854976177215576,
|
| 37 |
+
"eval_cosine_accuracy": 0.9625555276870728
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 1.2884473877851361,
|
| 41 |
+
"global_step": 14000,
|
| 42 |
+
"eval_loss": 0.5530505776405334,
|
| 43 |
+
"eval_cosine_accuracy": 0.9658345580101013
|
| 44 |
+
},
|
| 45 |
+
{
|
| 46 |
+
"epoch": 1.4724061810154525,
|
| 47 |
+
"global_step": 16000,
|
| 48 |
+
"eval_loss": 0.5652992725372314,
|
| 49 |
+
"eval_cosine_accuracy": 0.9668923020362854
|
| 50 |
+
},
|
| 51 |
+
{
|
| 52 |
+
"epoch": 1.656364974245769,
|
| 53 |
+
"global_step": 18000,
|
| 54 |
+
"eval_loss": 0.5359070301055908,
|
| 55 |
+
"eval_cosine_accuracy": 0.9699597954750061
|
| 56 |
+
},
|
| 57 |
+
{
|
| 58 |
+
"epoch": 1.8403237674760855,
|
| 59 |
+
"global_step": 20000,
|
| 60 |
+
"eval_loss": 0.5235444903373718,
|
| 61 |
+
"eval_cosine_accuracy": 0.9698540568351746
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"epoch": 2.0242825607064017,
|
| 65 |
+
"global_step": 22000,
|
| 66 |
+
"eval_loss": 0.5146053433418274,
|
| 67 |
+
"eval_cosine_accuracy": 0.9708060026168823
|
| 68 |
+
},
|
| 69 |
+
{
|
| 70 |
+
"epoch": 2.2082413539367183,
|
| 71 |
+
"global_step": 24000,
|
| 72 |
+
"eval_loss": 0.5143696665763855,
|
| 73 |
+
"eval_cosine_accuracy": 0.9712291359901428
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 2.3922001471670344,
|
| 77 |
+
"global_step": 26000,
|
| 78 |
+
"eval_loss": 0.5076085329055786,
|
| 79 |
+
"eval_cosine_accuracy": 0.9711233377456665
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 2.576158940397351,
|
| 83 |
+
"global_step": 28000,
|
| 84 |
+
"eval_loss": 0.5030118823051453,
|
| 85 |
+
"eval_cosine_accuracy": 0.972815752029419
|
| 86 |
+
},
|
| 87 |
+
{
|
| 88 |
+
"epoch": 2.760117733627667,
|
| 89 |
+
"global_step": 30000,
|
| 90 |
+
"eval_loss": 0.4934147894382477,
|
| 91 |
+
"eval_cosine_accuracy": 0.9729214906692505
|
| 92 |
+
},
|
| 93 |
+
{
|
| 94 |
+
"epoch": 2.944076526857984,
|
| 95 |
+
"global_step": 32000,
|
| 96 |
+
"eval_loss": 0.4958887994289398,
|
| 97 |
+
"eval_cosine_accuracy": 0.9721810817718506
|
| 98 |
+
},
|
| 99 |
+
{
|
| 100 |
+
"epoch": 3.1280353200883004,
|
| 101 |
+
"global_step": 34000,
|
| 102 |
+
"eval_loss": 0.49165186285972595,
|
| 103 |
+
"eval_cosine_accuracy": 0.9727099537849426
|
| 104 |
+
},
|
| 105 |
+
{
|
| 106 |
+
"epoch": 3.3119941133186166,
|
| 107 |
+
"global_step": 36000,
|
| 108 |
+
"eval_loss": 0.48895686864852905,
|
| 109 |
+
"eval_cosine_accuracy": 0.972815752029419
|
| 110 |
+
},
|
| 111 |
+
{
|
| 112 |
+
"epoch": 3.495952906548933,
|
| 113 |
+
"global_step": 38000,
|
| 114 |
+
"eval_loss": 0.4920443296432495,
|
| 115 |
+
"eval_cosine_accuracy": 0.9729214906692505
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 3.6799116997792494,
|
| 119 |
+
"global_step": 40000,
|
| 120 |
+
"eval_loss": 0.49153533577919006,
|
| 121 |
+
"eval_cosine_accuracy": 0.9729214906692505
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 3.863870493009566,
|
| 125 |
+
"global_step": 42000,
|
| 126 |
+
"eval_loss": 0.49136409163475037,
|
| 127 |
+
"eval_cosine_accuracy": 0.9727099537849426
|
| 128 |
+
}
|
| 129 |
+
],
|
| 130 |
+
"total_epochs": 3.997792494481236,
|
| 131 |
+
"total_steps": 43456
|
| 132 |
+
}
|