Training in progress, step 219, checkpoint
Browse files- checkpoint-219/1_AdvancedWeightedPooling/config.json +3 -3
- checkpoint-219/1_AdvancedWeightedPooling/pytorch_model.bin +2 -2
- checkpoint-219/README.md +0 -0
- checkpoint-219/optimizer.pt +2 -2
- checkpoint-219/rng_state.pth +1 -1
- checkpoint-219/trainer_state.json +1508 -108
- checkpoint-219/training_args.bin +1 -1
checkpoint-219/1_AdvancedWeightedPooling/config.json
CHANGED
|
@@ -3,8 +3,8 @@
|
|
| 3 |
"num_heads": 8,
|
| 4 |
"dropout": 0.05,
|
| 5 |
"bias": true,
|
| 6 |
-
"use_layernorm":
|
| 7 |
-
"use_MLP":
|
| 8 |
-
"MLP_h_size":
|
| 9 |
"use_residual": false
|
| 10 |
}
|
|
|
|
| 3 |
"num_heads": 8,
|
| 4 |
"dropout": 0.05,
|
| 5 |
"bias": true,
|
| 6 |
+
"use_layernorm": true,
|
| 7 |
+
"use_MLP": true,
|
| 8 |
+
"MLP_h_size": 2048,
|
| 9 |
"use_residual": false
|
| 10 |
}
|
checkpoint-219/1_AdvancedWeightedPooling/pytorch_model.bin
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:e4a04620ba1d9784801533cf38956b610797b242832a12878c87866d64167d5d
|
| 3 |
+
size 33595276
|
checkpoint-219/README.md
CHANGED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-219/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 sha256:65463ea38447a4abb9a8a8ad374b85c61ef0eb7fd8597c70d67476cd027eb816
|
| 3 |
+
size 67191488
|
checkpoint-219/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e25d70a0a83e5fa5e858dd1e9293efb8acc5162fe94a4a09c0786dc8cbe38398
|
| 3 |
size 14244
|
checkpoint-219/trainer_state.json
CHANGED
|
@@ -8,262 +8,1662 @@
|
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 11 |
{
|
| 12 |
"epoch": 0.015089163237311385,
|
| 13 |
-
"grad_norm":
|
| 14 |
"learning_rate": 1.8281535648994517e-05,
|
| 15 |
-
"loss": 0.
|
| 16 |
"step": 11
|
| 17 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 18 |
{
|
| 19 |
"epoch": 0.03017832647462277,
|
| 20 |
-
"grad_norm":
|
| 21 |
"learning_rate": 3.839122486288849e-05,
|
| 22 |
-
"loss": 0.
|
| 23 |
"step": 22
|
| 24 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 25 |
{
|
| 26 |
"epoch": 0.04526748971193416,
|
| 27 |
-
"grad_norm":
|
| 28 |
"learning_rate": 5.850091407678245e-05,
|
| 29 |
-
"loss": 0.
|
| 30 |
"step": 33
|
| 31 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 32 |
{
|
| 33 |
"epoch": 0.06035665294924554,
|
| 34 |
-
"grad_norm":
|
| 35 |
"learning_rate": 7.861060329067642e-05,
|
| 36 |
-
"loss": 0.
|
| 37 |
"step": 44
|
| 38 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 39 |
{
|
| 40 |
"epoch": 0.07544581618655692,
|
| 41 |
-
"grad_norm":
|
| 42 |
"learning_rate": 9.872029250457039e-05,
|
| 43 |
-
"loss": 0.
|
| 44 |
"step": 55
|
| 45 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 46 |
{
|
| 47 |
"epoch": 0.09053497942386832,
|
| 48 |
-
"grad_norm":
|
| 49 |
"learning_rate": 0.00011882998171846434,
|
| 50 |
-
"loss": 0.
|
| 51 |
"step": 66
|
| 52 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 53 |
{
|
| 54 |
"epoch": 0.1056241426611797,
|
| 55 |
-
"grad_norm":
|
| 56 |
"learning_rate": 0.00013893967093235832,
|
| 57 |
-
"loss": 0.
|
| 58 |
"step": 77
|
| 59 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 60 |
{
|
| 61 |
"epoch": 0.12071330589849108,
|
| 62 |
-
"grad_norm":
|
| 63 |
"learning_rate": 0.00015904936014625229,
|
| 64 |
-
"loss": 0.
|
| 65 |
"step": 88
|
| 66 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 67 |
{
|
| 68 |
"epoch": 0.13580246913580246,
|
| 69 |
-
"grad_norm":
|
| 70 |
"learning_rate": 0.00017915904936014626,
|
| 71 |
-
"loss": 0.
|
| 72 |
"step": 99
|
| 73 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 74 |
{
|
| 75 |
"epoch": 0.15089163237311384,
|
| 76 |
-
"grad_norm":
|
| 77 |
"learning_rate": 0.00019926873857404023,
|
| 78 |
-
"loss": 0.
|
| 79 |
"step": 110
|
| 80 |
},
|
| 81 |
{
|
| 82 |
"epoch": 0.15089163237311384,
|
| 83 |
-
"eval_Qnli-dev_cosine_accuracy": 0.
|
| 84 |
-
"eval_Qnli-dev_cosine_accuracy_threshold": 0.
|
| 85 |
-
"eval_Qnli-dev_cosine_ap": 0.
|
| 86 |
-
"eval_Qnli-dev_cosine_f1": 0.
|
| 87 |
-
"eval_Qnli-dev_cosine_f1_threshold": 0.
|
| 88 |
-
"eval_Qnli-dev_cosine_precision": 0.
|
| 89 |
-
"eval_Qnli-dev_cosine_recall": 0.
|
| 90 |
-
"eval_allNLI-dev_cosine_accuracy": 0.
|
| 91 |
-
"eval_allNLI-dev_cosine_accuracy_threshold": 0.
|
| 92 |
-
"eval_allNLI-dev_cosine_ap": 0.
|
| 93 |
-
"eval_allNLI-dev_cosine_f1": 0.
|
| 94 |
-
"eval_allNLI-dev_cosine_f1_threshold": 0.
|
| 95 |
-
"eval_allNLI-dev_cosine_precision": 0.
|
| 96 |
-
"eval_allNLI-dev_cosine_recall": 0.
|
| 97 |
-
"eval_sequential_score": 0.
|
| 98 |
-
"eval_sts-test_pearson_cosine": 0.
|
| 99 |
-
"eval_sts-test_spearman_cosine": 0.
|
| 100 |
-
"eval_vitaminc-pairs_loss":
|
| 101 |
-
"eval_vitaminc-pairs_runtime":
|
| 102 |
-
"eval_vitaminc-pairs_samples_per_second":
|
| 103 |
-
"eval_vitaminc-pairs_steps_per_second": 0.
|
| 104 |
"step": 110
|
| 105 |
},
|
| 106 |
{
|
| 107 |
"epoch": 0.15089163237311384,
|
| 108 |
-
"eval_negation-triplets_loss":
|
| 109 |
-
"eval_negation-triplets_runtime":
|
| 110 |
-
"eval_negation-triplets_samples_per_second":
|
| 111 |
-
"eval_negation-triplets_steps_per_second": 0.
|
| 112 |
"step": 110
|
| 113 |
},
|
| 114 |
{
|
| 115 |
"epoch": 0.15089163237311384,
|
| 116 |
-
"eval_scitail-pairs-pos_loss": 0.
|
| 117 |
-
"eval_scitail-pairs-pos_runtime":
|
| 118 |
-
"eval_scitail-pairs-pos_samples_per_second":
|
| 119 |
-
"eval_scitail-pairs-pos_steps_per_second": 0.
|
| 120 |
"step": 110
|
| 121 |
},
|
| 122 |
{
|
| 123 |
"epoch": 0.15089163237311384,
|
| 124 |
-
"eval_scitail-pairs-qa_loss": 0.
|
| 125 |
-
"eval_scitail-pairs-qa_runtime":
|
| 126 |
-
"eval_scitail-pairs-qa_samples_per_second":
|
| 127 |
-
"eval_scitail-pairs-qa_steps_per_second": 0.
|
| 128 |
"step": 110
|
| 129 |
},
|
| 130 |
{
|
| 131 |
"epoch": 0.15089163237311384,
|
| 132 |
-
"eval_xsum-pairs_loss": 0.
|
| 133 |
-
"eval_xsum-pairs_runtime":
|
| 134 |
-
"eval_xsum-pairs_samples_per_second":
|
| 135 |
-
"eval_xsum-pairs_steps_per_second": 0.
|
| 136 |
"step": 110
|
| 137 |
},
|
| 138 |
{
|
| 139 |
"epoch": 0.15089163237311384,
|
| 140 |
-
"eval_sciq_pairs_loss": 0.
|
| 141 |
-
"eval_sciq_pairs_runtime":
|
| 142 |
-
"eval_sciq_pairs_samples_per_second":
|
| 143 |
-
"eval_sciq_pairs_steps_per_second": 0.
|
| 144 |
"step": 110
|
| 145 |
},
|
| 146 |
{
|
| 147 |
"epoch": 0.15089163237311384,
|
| 148 |
-
"eval_qasc_pairs_loss": 0.
|
| 149 |
-
"eval_qasc_pairs_runtime":
|
| 150 |
-
"eval_qasc_pairs_samples_per_second":
|
| 151 |
-
"eval_qasc_pairs_steps_per_second": 0.
|
| 152 |
"step": 110
|
| 153 |
},
|
| 154 |
{
|
| 155 |
"epoch": 0.15089163237311384,
|
| 156 |
-
"eval_openbookqa_pairs_loss": 1.
|
| 157 |
-
"eval_openbookqa_pairs_runtime":
|
| 158 |
-
"eval_openbookqa_pairs_samples_per_second":
|
| 159 |
-
"eval_openbookqa_pairs_steps_per_second": 0.
|
| 160 |
"step": 110
|
| 161 |
},
|
| 162 |
{
|
| 163 |
"epoch": 0.15089163237311384,
|
| 164 |
-
"eval_nq_pairs_loss": 0.
|
| 165 |
-
"eval_nq_pairs_runtime":
|
| 166 |
-
"eval_nq_pairs_samples_per_second":
|
| 167 |
-
"eval_nq_pairs_steps_per_second": 0.
|
| 168 |
"step": 110
|
| 169 |
},
|
| 170 |
{
|
| 171 |
"epoch": 0.15089163237311384,
|
| 172 |
-
"eval_trivia_pairs_loss": 0.
|
| 173 |
-
"eval_trivia_pairs_runtime":
|
| 174 |
-
"eval_trivia_pairs_samples_per_second":
|
| 175 |
-
"eval_trivia_pairs_steps_per_second": 0.
|
| 176 |
"step": 110
|
| 177 |
},
|
| 178 |
{
|
| 179 |
"epoch": 0.15089163237311384,
|
| 180 |
-
"eval_gooaq_pairs_loss": 0.
|
| 181 |
-
"eval_gooaq_pairs_runtime":
|
| 182 |
-
"eval_gooaq_pairs_samples_per_second":
|
| 183 |
-
"eval_gooaq_pairs_steps_per_second": 0.
|
| 184 |
"step": 110
|
| 185 |
},
|
| 186 |
{
|
| 187 |
"epoch": 0.15089163237311384,
|
| 188 |
-
"eval_paws-pos_loss": 0.
|
| 189 |
-
"eval_paws-pos_runtime":
|
| 190 |
-
"eval_paws-pos_samples_per_second":
|
| 191 |
-
"eval_paws-pos_steps_per_second": 0.
|
| 192 |
"step": 110
|
| 193 |
},
|
| 194 |
{
|
| 195 |
"epoch": 0.15089163237311384,
|
| 196 |
-
"eval_global_dataset_loss": 0.
|
| 197 |
-
"eval_global_dataset_runtime":
|
| 198 |
-
"eval_global_dataset_samples_per_second":
|
| 199 |
-
"eval_global_dataset_steps_per_second": 0.
|
| 200 |
"step": 110
|
| 201 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 202 |
{
|
| 203 |
"epoch": 0.16598079561042525,
|
| 204 |
-
"grad_norm":
|
| 205 |
"learning_rate": 0.00021937842778793417,
|
| 206 |
-
"loss": 0.
|
| 207 |
"step": 121
|
| 208 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 209 |
{
|
| 210 |
"epoch": 0.18106995884773663,
|
| 211 |
-
"grad_norm":
|
| 212 |
"learning_rate": 0.00023948811700182814,
|
| 213 |
-
"loss": 0.
|
| 214 |
"step": 132
|
| 215 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 216 |
{
|
| 217 |
"epoch": 0.19615912208504802,
|
| 218 |
-
"grad_norm":
|
| 219 |
"learning_rate": 0.0002595978062157221,
|
| 220 |
-
"loss": 0.
|
| 221 |
"step": 143
|
| 222 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 223 |
{
|
| 224 |
"epoch": 0.2112482853223594,
|
| 225 |
-
"grad_norm":
|
| 226 |
"learning_rate": 0.0002797074954296161,
|
| 227 |
-
"loss": 0.
|
| 228 |
"step": 154
|
| 229 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 230 |
{
|
| 231 |
"epoch": 0.22633744855967078,
|
| 232 |
-
"grad_norm":
|
| 233 |
"learning_rate": 0.00029981718464351003,
|
| 234 |
-
"loss": 0.
|
| 235 |
"step": 165
|
| 236 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 237 |
{
|
| 238 |
"epoch": 0.24142661179698216,
|
| 239 |
-
"grad_norm": 0.
|
| 240 |
"learning_rate": 0.000319926873857404,
|
| 241 |
-
"loss": 0.
|
| 242 |
"step": 176
|
| 243 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 244 |
{
|
| 245 |
"epoch": 0.25651577503429357,
|
| 246 |
-
"grad_norm":
|
| 247 |
"learning_rate": 0.00034003656307129797,
|
| 248 |
-
"loss": 0.
|
| 249 |
"step": 187
|
| 250 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 251 |
{
|
| 252 |
"epoch": 0.2716049382716049,
|
| 253 |
-
"grad_norm":
|
| 254 |
"learning_rate": 0.00036014625228519197,
|
| 255 |
-
"loss": 0.
|
| 256 |
"step": 198
|
| 257 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 258 |
{
|
| 259 |
"epoch": 0.28669410150891633,
|
| 260 |
-
"grad_norm":
|
| 261 |
"learning_rate": 0.0003802559414990859,
|
| 262 |
-
"loss": 0.
|
| 263 |
"step": 209
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 264 |
}
|
| 265 |
],
|
| 266 |
-
"logging_steps":
|
| 267 |
"max_steps": 2187,
|
| 268 |
"num_input_tokens_seen": 0,
|
| 269 |
"num_train_epochs": 3,
|
|
|
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0013717421124828531,
|
| 13 |
+
"grad_norm": 13.75906753540039,
|
| 14 |
+
"learning_rate": 1.8281535648994516e-06,
|
| 15 |
+
"loss": 1.2468,
|
| 16 |
+
"step": 1
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.0027434842249657062,
|
| 20 |
+
"grad_norm": 14.05058765411377,
|
| 21 |
+
"learning_rate": 3.6563071297989032e-06,
|
| 22 |
+
"loss": 1.4692,
|
| 23 |
+
"step": 2
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.00411522633744856,
|
| 27 |
+
"grad_norm": 13.991771697998047,
|
| 28 |
+
"learning_rate": 5.484460694698355e-06,
|
| 29 |
+
"loss": 1.2457,
|
| 30 |
+
"step": 3
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.0054869684499314125,
|
| 34 |
+
"grad_norm": 13.429465293884277,
|
| 35 |
+
"learning_rate": 7.3126142595978065e-06,
|
| 36 |
+
"loss": 1.1859,
|
| 37 |
+
"step": 4
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.006858710562414266,
|
| 41 |
+
"grad_norm": 12.403002738952637,
|
| 42 |
+
"learning_rate": 9.140767824497258e-06,
|
| 43 |
+
"loss": 1.2404,
|
| 44 |
+
"step": 5
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.00823045267489712,
|
| 48 |
+
"grad_norm": 1.5524662733078003,
|
| 49 |
+
"learning_rate": 1.096892138939671e-05,
|
| 50 |
+
"loss": 0.042,
|
| 51 |
+
"step": 6
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.009602194787379973,
|
| 55 |
+
"grad_norm": 10.1494140625,
|
| 56 |
+
"learning_rate": 1.2797074954296162e-05,
|
| 57 |
+
"loss": 0.8856,
|
| 58 |
+
"step": 7
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.010973936899862825,
|
| 62 |
+
"grad_norm": Infinity,
|
| 63 |
+
"learning_rate": 1.2797074954296162e-05,
|
| 64 |
+
"loss": 1.5417,
|
| 65 |
+
"step": 8
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.012345679012345678,
|
| 69 |
+
"grad_norm": 0.4553964138031006,
|
| 70 |
+
"learning_rate": 1.4625228519195613e-05,
|
| 71 |
+
"loss": 0.023,
|
| 72 |
+
"step": 9
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.013717421124828532,
|
| 76 |
+
"grad_norm": 10.332369804382324,
|
| 77 |
+
"learning_rate": 1.6453382084095062e-05,
|
| 78 |
+
"loss": 0.8655,
|
| 79 |
+
"step": 10
|
| 80 |
+
},
|
| 81 |
{
|
| 82 |
"epoch": 0.015089163237311385,
|
| 83 |
+
"grad_norm": 7.544310092926025,
|
| 84 |
"learning_rate": 1.8281535648994517e-05,
|
| 85 |
+
"loss": 0.5894,
|
| 86 |
"step": 11
|
| 87 |
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.01646090534979424,
|
| 90 |
+
"grad_norm": 8.16427230834961,
|
| 91 |
+
"learning_rate": 2.0109689213893968e-05,
|
| 92 |
+
"loss": 0.7053,
|
| 93 |
+
"step": 12
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.01783264746227709,
|
| 97 |
+
"grad_norm": 7.403252601623535,
|
| 98 |
+
"learning_rate": 2.193784277879342e-05,
|
| 99 |
+
"loss": 0.5857,
|
| 100 |
+
"step": 13
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.019204389574759947,
|
| 104 |
+
"grad_norm": 8.974674224853516,
|
| 105 |
+
"learning_rate": 2.376599634369287e-05,
|
| 106 |
+
"loss": 0.8375,
|
| 107 |
+
"step": 14
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.0205761316872428,
|
| 111 |
+
"grad_norm": 13.417745590209961,
|
| 112 |
+
"learning_rate": 2.5594149908592324e-05,
|
| 113 |
+
"loss": 0.9043,
|
| 114 |
+
"step": 15
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.02194787379972565,
|
| 118 |
+
"grad_norm": 12.881294250488281,
|
| 119 |
+
"learning_rate": 2.742230347349177e-05,
|
| 120 |
+
"loss": 0.8756,
|
| 121 |
+
"step": 16
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.023319615912208505,
|
| 125 |
+
"grad_norm": 7.439205169677734,
|
| 126 |
+
"learning_rate": 2.9250457038391226e-05,
|
| 127 |
+
"loss": 0.5076,
|
| 128 |
+
"step": 17
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.024691358024691357,
|
| 132 |
+
"grad_norm": 8.46964168548584,
|
| 133 |
+
"learning_rate": 3.107861060329068e-05,
|
| 134 |
+
"loss": 0.4757,
|
| 135 |
+
"step": 18
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.02606310013717421,
|
| 139 |
+
"grad_norm": 17.02773666381836,
|
| 140 |
+
"learning_rate": 3.2906764168190124e-05,
|
| 141 |
+
"loss": 0.9993,
|
| 142 |
+
"step": 19
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.027434842249657063,
|
| 146 |
+
"grad_norm": 6.2668776512146,
|
| 147 |
+
"learning_rate": 3.473491773308958e-05,
|
| 148 |
+
"loss": 0.2622,
|
| 149 |
+
"step": 20
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.02880658436213992,
|
| 153 |
+
"grad_norm": 8.273824691772461,
|
| 154 |
+
"learning_rate": 3.656307129798903e-05,
|
| 155 |
+
"loss": 0.3497,
|
| 156 |
+
"step": 21
|
| 157 |
+
},
|
| 158 |
{
|
| 159 |
"epoch": 0.03017832647462277,
|
| 160 |
+
"grad_norm": 5.5460944175720215,
|
| 161 |
"learning_rate": 3.839122486288849e-05,
|
| 162 |
+
"loss": 0.2514,
|
| 163 |
"step": 22
|
| 164 |
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.03155006858710562,
|
| 167 |
+
"grad_norm": 4.283128261566162,
|
| 168 |
+
"learning_rate": 4.0219378427787935e-05,
|
| 169 |
+
"loss": 0.1673,
|
| 170 |
+
"step": 23
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.03292181069958848,
|
| 174 |
+
"grad_norm": 4.708792209625244,
|
| 175 |
+
"learning_rate": 4.204753199268738e-05,
|
| 176 |
+
"loss": 0.203,
|
| 177 |
+
"step": 24
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.03429355281207133,
|
| 181 |
+
"grad_norm": 14.491021156311035,
|
| 182 |
+
"learning_rate": 4.387568555758684e-05,
|
| 183 |
+
"loss": 0.698,
|
| 184 |
+
"step": 25
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.03566529492455418,
|
| 188 |
+
"grad_norm": 7.903520584106445,
|
| 189 |
+
"learning_rate": 4.570383912248629e-05,
|
| 190 |
+
"loss": 0.3401,
|
| 191 |
+
"step": 26
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.037037037037037035,
|
| 195 |
+
"grad_norm": 7.333080291748047,
|
| 196 |
+
"learning_rate": 4.753199268738574e-05,
|
| 197 |
+
"loss": 0.2185,
|
| 198 |
+
"step": 27
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.038408779149519894,
|
| 202 |
+
"grad_norm": 8.625358581542969,
|
| 203 |
+
"learning_rate": 4.936014625228519e-05,
|
| 204 |
+
"loss": 0.4424,
|
| 205 |
+
"step": 28
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.039780521262002745,
|
| 209 |
+
"grad_norm": 1.5588488578796387,
|
| 210 |
+
"learning_rate": 5.118829981718465e-05,
|
| 211 |
+
"loss": 0.0381,
|
| 212 |
+
"step": 29
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.0411522633744856,
|
| 216 |
+
"grad_norm": 12.401138305664062,
|
| 217 |
+
"learning_rate": 5.3016453382084095e-05,
|
| 218 |
+
"loss": 0.8215,
|
| 219 |
+
"step": 30
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.04252400548696845,
|
| 223 |
+
"grad_norm": 5.405845642089844,
|
| 224 |
+
"learning_rate": 5.484460694698354e-05,
|
| 225 |
+
"loss": 0.1542,
|
| 226 |
+
"step": 31
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.0438957475994513,
|
| 230 |
+
"grad_norm": 8.558808326721191,
|
| 231 |
+
"learning_rate": 5.6672760511883e-05,
|
| 232 |
+
"loss": 0.6893,
|
| 233 |
+
"step": 32
|
| 234 |
+
},
|
| 235 |
{
|
| 236 |
"epoch": 0.04526748971193416,
|
| 237 |
+
"grad_norm": 7.206741809844971,
|
| 238 |
"learning_rate": 5.850091407678245e-05,
|
| 239 |
+
"loss": 0.3773,
|
| 240 |
"step": 33
|
| 241 |
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.04663923182441701,
|
| 244 |
+
"grad_norm": 8.300729751586914,
|
| 245 |
+
"learning_rate": 6.0329067641681906e-05,
|
| 246 |
+
"loss": 0.538,
|
| 247 |
+
"step": 34
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.04801097393689986,
|
| 251 |
+
"grad_norm": 0.2500181496143341,
|
| 252 |
+
"learning_rate": 6.215722120658135e-05,
|
| 253 |
+
"loss": 0.0073,
|
| 254 |
+
"step": 35
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.04938271604938271,
|
| 258 |
+
"grad_norm": 17.457223892211914,
|
| 259 |
+
"learning_rate": 6.398537477148081e-05,
|
| 260 |
+
"loss": 2.378,
|
| 261 |
+
"step": 36
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.05075445816186557,
|
| 265 |
+
"grad_norm": 10.884990692138672,
|
| 266 |
+
"learning_rate": 6.581352833638025e-05,
|
| 267 |
+
"loss": 0.5949,
|
| 268 |
+
"step": 37
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.05212620027434842,
|
| 272 |
+
"grad_norm": 10.013723373413086,
|
| 273 |
+
"learning_rate": 6.764168190127972e-05,
|
| 274 |
+
"loss": 0.7071,
|
| 275 |
+
"step": 38
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.053497942386831275,
|
| 279 |
+
"grad_norm": 4.653324604034424,
|
| 280 |
+
"learning_rate": 6.946983546617916e-05,
|
| 281 |
+
"loss": 0.1607,
|
| 282 |
+
"step": 39
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.05486968449931413,
|
| 286 |
+
"grad_norm": 9.527400970458984,
|
| 287 |
+
"learning_rate": 7.129798903107861e-05,
|
| 288 |
+
"loss": 0.7735,
|
| 289 |
+
"step": 40
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.056241426611796985,
|
| 293 |
+
"grad_norm": 12.477531433105469,
|
| 294 |
+
"learning_rate": 7.312614259597807e-05,
|
| 295 |
+
"loss": 0.7594,
|
| 296 |
+
"step": 41
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.05761316872427984,
|
| 300 |
+
"grad_norm": 5.369799613952637,
|
| 301 |
+
"learning_rate": 7.495429616087751e-05,
|
| 302 |
+
"loss": 0.3569,
|
| 303 |
+
"step": 42
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.05898491083676269,
|
| 307 |
+
"grad_norm": 5.1385908126831055,
|
| 308 |
+
"learning_rate": 7.678244972577697e-05,
|
| 309 |
+
"loss": 0.2454,
|
| 310 |
+
"step": 43
|
| 311 |
+
},
|
| 312 |
{
|
| 313 |
"epoch": 0.06035665294924554,
|
| 314 |
+
"grad_norm": 6.1807708740234375,
|
| 315 |
"learning_rate": 7.861060329067642e-05,
|
| 316 |
+
"loss": 0.2723,
|
| 317 |
"step": 44
|
| 318 |
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.06172839506172839,
|
| 321 |
+
"grad_norm": 7.941879749298096,
|
| 322 |
+
"learning_rate": 8.043875685557587e-05,
|
| 323 |
+
"loss": 0.5338,
|
| 324 |
+
"step": 45
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.06310013717421124,
|
| 328 |
+
"grad_norm": 5.015410423278809,
|
| 329 |
+
"learning_rate": 8.226691042047532e-05,
|
| 330 |
+
"loss": 0.1891,
|
| 331 |
+
"step": 46
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.0644718792866941,
|
| 335 |
+
"grad_norm": 7.299699306488037,
|
| 336 |
+
"learning_rate": 8.409506398537477e-05,
|
| 337 |
+
"loss": 0.3647,
|
| 338 |
+
"step": 47
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.06584362139917696,
|
| 342 |
+
"grad_norm": 8.421393394470215,
|
| 343 |
+
"learning_rate": 8.592321755027423e-05,
|
| 344 |
+
"loss": 0.383,
|
| 345 |
+
"step": 48
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.06721536351165981,
|
| 349 |
+
"grad_norm": 5.5915937423706055,
|
| 350 |
+
"learning_rate": 8.775137111517367e-05,
|
| 351 |
+
"loss": 0.2353,
|
| 352 |
+
"step": 49
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.06858710562414266,
|
| 356 |
+
"grad_norm": 8.187829971313477,
|
| 357 |
+
"learning_rate": 8.957952468007313e-05,
|
| 358 |
+
"loss": 0.5541,
|
| 359 |
+
"step": 50
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.06995884773662552,
|
| 363 |
+
"grad_norm": 6.386786460876465,
|
| 364 |
+
"learning_rate": 9.140767824497258e-05,
|
| 365 |
+
"loss": 0.4908,
|
| 366 |
+
"step": 51
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.07133058984910837,
|
| 370 |
+
"grad_norm": 8.64050006866455,
|
| 371 |
+
"learning_rate": 9.323583180987204e-05,
|
| 372 |
+
"loss": 0.586,
|
| 373 |
+
"step": 52
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.07270233196159122,
|
| 377 |
+
"grad_norm": 5.879551410675049,
|
| 378 |
+
"learning_rate": 9.506398537477148e-05,
|
| 379 |
+
"loss": 0.2241,
|
| 380 |
+
"step": 53
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.07407407407407407,
|
| 384 |
+
"grad_norm": 7.824138164520264,
|
| 385 |
+
"learning_rate": 9.689213893967093e-05,
|
| 386 |
+
"loss": 0.6046,
|
| 387 |
+
"step": 54
|
| 388 |
+
},
|
| 389 |
{
|
| 390 |
"epoch": 0.07544581618655692,
|
| 391 |
+
"grad_norm": 6.351109504699707,
|
| 392 |
"learning_rate": 9.872029250457039e-05,
|
| 393 |
+
"loss": 0.231,
|
| 394 |
"step": 55
|
| 395 |
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.07681755829903979,
|
| 398 |
+
"grad_norm": 9.437410354614258,
|
| 399 |
+
"learning_rate": 0.00010054844606946984,
|
| 400 |
+
"loss": 0.7105,
|
| 401 |
+
"step": 56
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.07818930041152264,
|
| 405 |
+
"grad_norm": 8.40911865234375,
|
| 406 |
+
"learning_rate": 0.0001023765996343693,
|
| 407 |
+
"loss": 0.5591,
|
| 408 |
+
"step": 57
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.07956104252400549,
|
| 412 |
+
"grad_norm": 7.631382942199707,
|
| 413 |
+
"learning_rate": 0.00010420475319926874,
|
| 414 |
+
"loss": 0.5194,
|
| 415 |
+
"step": 58
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.08093278463648834,
|
| 419 |
+
"grad_norm": 5.773220062255859,
|
| 420 |
+
"learning_rate": 0.00010603290676416819,
|
| 421 |
+
"loss": 0.3297,
|
| 422 |
+
"step": 59
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.0823045267489712,
|
| 426 |
+
"grad_norm": 1.3606321811676025,
|
| 427 |
+
"learning_rate": 0.00010786106032906765,
|
| 428 |
+
"loss": 0.0299,
|
| 429 |
+
"step": 60
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.08367626886145405,
|
| 433 |
+
"grad_norm": 7.216275215148926,
|
| 434 |
+
"learning_rate": 0.00010968921389396709,
|
| 435 |
+
"loss": 0.3514,
|
| 436 |
+
"step": 61
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 0.0850480109739369,
|
| 440 |
+
"grad_norm": 4.70477294921875,
|
| 441 |
+
"learning_rate": 0.00011151736745886655,
|
| 442 |
+
"loss": 0.1932,
|
| 443 |
+
"step": 62
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.08641975308641975,
|
| 447 |
+
"grad_norm": 6.754104137420654,
|
| 448 |
+
"learning_rate": 0.000113345521023766,
|
| 449 |
+
"loss": 0.4035,
|
| 450 |
+
"step": 63
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.0877914951989026,
|
| 454 |
+
"grad_norm": 0.19067375361919403,
|
| 455 |
+
"learning_rate": 0.00011517367458866546,
|
| 456 |
+
"loss": 0.0094,
|
| 457 |
+
"step": 64
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.08916323731138547,
|
| 461 |
+
"grad_norm": 1.1715893745422363,
|
| 462 |
+
"learning_rate": 0.0001170018281535649,
|
| 463 |
+
"loss": 0.0148,
|
| 464 |
+
"step": 65
|
| 465 |
+
},
|
| 466 |
{
|
| 467 |
"epoch": 0.09053497942386832,
|
| 468 |
+
"grad_norm": 1.6287739276885986,
|
| 469 |
"learning_rate": 0.00011882998171846434,
|
| 470 |
+
"loss": 0.0231,
|
| 471 |
"step": 66
|
| 472 |
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.09190672153635117,
|
| 475 |
+
"grad_norm": 7.027708053588867,
|
| 476 |
+
"learning_rate": 0.00012065813528336381,
|
| 477 |
+
"loss": 0.3204,
|
| 478 |
+
"step": 67
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 0.09327846364883402,
|
| 482 |
+
"grad_norm": 7.248253345489502,
|
| 483 |
+
"learning_rate": 0.00012248628884826325,
|
| 484 |
+
"loss": 0.3011,
|
| 485 |
+
"step": 68
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.09465020576131687,
|
| 489 |
+
"grad_norm": 9.592718124389648,
|
| 490 |
+
"learning_rate": 0.0001243144424131627,
|
| 491 |
+
"loss": 0.3871,
|
| 492 |
+
"step": 69
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 0.09602194787379972,
|
| 496 |
+
"grad_norm": 5.128874778747559,
|
| 497 |
+
"learning_rate": 0.00012614259597806216,
|
| 498 |
+
"loss": 0.1823,
|
| 499 |
+
"step": 70
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 0.09739368998628258,
|
| 503 |
+
"grad_norm": 6.496853351593018,
|
| 504 |
+
"learning_rate": 0.00012797074954296162,
|
| 505 |
+
"loss": 0.3572,
|
| 506 |
+
"step": 71
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.09876543209876543,
|
| 510 |
+
"grad_norm": 6.564659118652344,
|
| 511 |
+
"learning_rate": 0.00012979890310786104,
|
| 512 |
+
"loss": 0.5289,
|
| 513 |
+
"step": 72
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.10013717421124829,
|
| 517 |
+
"grad_norm": 6.480371952056885,
|
| 518 |
+
"learning_rate": 0.0001316270566727605,
|
| 519 |
+
"loss": 0.3223,
|
| 520 |
+
"step": 73
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.10150891632373114,
|
| 524 |
+
"grad_norm": 7.222306728363037,
|
| 525 |
+
"learning_rate": 0.00013345521023765998,
|
| 526 |
+
"loss": 0.3247,
|
| 527 |
+
"step": 74
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.102880658436214,
|
| 531 |
+
"grad_norm": 5.406076431274414,
|
| 532 |
+
"learning_rate": 0.00013528336380255943,
|
| 533 |
+
"loss": 0.2133,
|
| 534 |
+
"step": 75
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.10425240054869685,
|
| 538 |
+
"grad_norm": 11.029163360595703,
|
| 539 |
+
"learning_rate": 0.00013711151736745886,
|
| 540 |
+
"loss": 0.8249,
|
| 541 |
+
"step": 76
|
| 542 |
+
},
|
| 543 |
{
|
| 544 |
"epoch": 0.1056241426611797,
|
| 545 |
+
"grad_norm": 7.284115314483643,
|
| 546 |
"learning_rate": 0.00013893967093235832,
|
| 547 |
+
"loss": 0.4341,
|
| 548 |
"step": 77
|
| 549 |
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.10699588477366255,
|
| 552 |
+
"grad_norm": 6.240738868713379,
|
| 553 |
+
"learning_rate": 0.00014076782449725777,
|
| 554 |
+
"loss": 0.2932,
|
| 555 |
+
"step": 78
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.1083676268861454,
|
| 559 |
+
"grad_norm": 0.1745665967464447,
|
| 560 |
+
"learning_rate": 0.00014259597806215722,
|
| 561 |
+
"loss": 0.0099,
|
| 562 |
+
"step": 79
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 0.10973936899862825,
|
| 566 |
+
"grad_norm": 5.460353851318359,
|
| 567 |
+
"learning_rate": 0.00014442413162705668,
|
| 568 |
+
"loss": 0.3348,
|
| 569 |
+
"step": 80
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.1111111111111111,
|
| 573 |
+
"grad_norm": 8.869246482849121,
|
| 574 |
+
"learning_rate": 0.00014625228519195613,
|
| 575 |
+
"loss": 0.6405,
|
| 576 |
+
"step": 81
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 0.11248285322359397,
|
| 580 |
+
"grad_norm": 4.475996971130371,
|
| 581 |
+
"learning_rate": 0.0001480804387568556,
|
| 582 |
+
"loss": 0.1536,
|
| 583 |
+
"step": 82
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 0.11385459533607682,
|
| 587 |
+
"grad_norm": 2.700299024581909,
|
| 588 |
+
"learning_rate": 0.00014990859232175501,
|
| 589 |
+
"loss": 0.1299,
|
| 590 |
+
"step": 83
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.11522633744855967,
|
| 594 |
+
"grad_norm": 7.5515618324279785,
|
| 595 |
+
"learning_rate": 0.00015173674588665447,
|
| 596 |
+
"loss": 0.5863,
|
| 597 |
+
"step": 84
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.11659807956104253,
|
| 601 |
+
"grad_norm": 9.869407653808594,
|
| 602 |
+
"learning_rate": 0.00015356489945155395,
|
| 603 |
+
"loss": 0.7205,
|
| 604 |
+
"step": 85
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.11796982167352538,
|
| 608 |
+
"grad_norm": 8.208423614501953,
|
| 609 |
+
"learning_rate": 0.00015539305301645338,
|
| 610 |
+
"loss": 0.4052,
|
| 611 |
+
"step": 86
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.11934156378600823,
|
| 615 |
+
"grad_norm": 6.408420562744141,
|
| 616 |
+
"learning_rate": 0.00015722120658135283,
|
| 617 |
+
"loss": 0.3953,
|
| 618 |
+
"step": 87
|
| 619 |
+
},
|
| 620 |
{
|
| 621 |
"epoch": 0.12071330589849108,
|
| 622 |
+
"grad_norm": 7.050099849700928,
|
| 623 |
"learning_rate": 0.00015904936014625229,
|
| 624 |
+
"loss": 0.5598,
|
| 625 |
"step": 88
|
| 626 |
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.12208504801097393,
|
| 629 |
+
"grad_norm": 5.326991558074951,
|
| 630 |
+
"learning_rate": 0.00016087751371115174,
|
| 631 |
+
"loss": 0.2856,
|
| 632 |
+
"step": 89
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.12345679012345678,
|
| 636 |
+
"grad_norm": 4.510193347930908,
|
| 637 |
+
"learning_rate": 0.0001627056672760512,
|
| 638 |
+
"loss": 0.2277,
|
| 639 |
+
"step": 90
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.12482853223593965,
|
| 643 |
+
"grad_norm": 5.769596576690674,
|
| 644 |
+
"learning_rate": 0.00016453382084095065,
|
| 645 |
+
"loss": 0.3296,
|
| 646 |
+
"step": 91
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 0.1262002743484225,
|
| 650 |
+
"grad_norm": 6.066390037536621,
|
| 651 |
+
"learning_rate": 0.0001663619744058501,
|
| 652 |
+
"loss": 0.3079,
|
| 653 |
+
"step": 92
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 0.12757201646090535,
|
| 657 |
+
"grad_norm": 6.80173921585083,
|
| 658 |
+
"learning_rate": 0.00016819012797074953,
|
| 659 |
+
"loss": 0.4867,
|
| 660 |
+
"step": 93
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.1289437585733882,
|
| 664 |
+
"grad_norm": 6.219693183898926,
|
| 665 |
+
"learning_rate": 0.00017001828153564899,
|
| 666 |
+
"loss": 0.4319,
|
| 667 |
+
"step": 94
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 0.13031550068587106,
|
| 671 |
+
"grad_norm": 5.316290855407715,
|
| 672 |
+
"learning_rate": 0.00017184643510054847,
|
| 673 |
+
"loss": 0.2952,
|
| 674 |
+
"step": 95
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 0.13168724279835392,
|
| 678 |
+
"grad_norm": 6.86447811126709,
|
| 679 |
+
"learning_rate": 0.00017367458866544792,
|
| 680 |
+
"loss": 0.5531,
|
| 681 |
+
"step": 96
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.13305898491083676,
|
| 685 |
+
"grad_norm": 1.2648167610168457,
|
| 686 |
+
"learning_rate": 0.00017550274223034735,
|
| 687 |
+
"loss": 0.0296,
|
| 688 |
+
"step": 97
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.13443072702331962,
|
| 692 |
+
"grad_norm": 8.14661979675293,
|
| 693 |
+
"learning_rate": 0.0001773308957952468,
|
| 694 |
+
"loss": 0.8536,
|
| 695 |
+
"step": 98
|
| 696 |
+
},
|
| 697 |
{
|
| 698 |
"epoch": 0.13580246913580246,
|
| 699 |
+
"grad_norm": 8.927884101867676,
|
| 700 |
"learning_rate": 0.00017915904936014626,
|
| 701 |
+
"loss": 0.4879,
|
| 702 |
"step": 99
|
| 703 |
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 0.13717421124828533,
|
| 706 |
+
"grad_norm": 9.555243492126465,
|
| 707 |
+
"learning_rate": 0.00018098720292504568,
|
| 708 |
+
"loss": 0.67,
|
| 709 |
+
"step": 100
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 0.13854595336076816,
|
| 713 |
+
"grad_norm": 7.783656120300293,
|
| 714 |
+
"learning_rate": 0.00018281535648994517,
|
| 715 |
+
"loss": 0.4813,
|
| 716 |
+
"step": 101
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 0.13991769547325103,
|
| 720 |
+
"grad_norm": 0.5169872641563416,
|
| 721 |
+
"learning_rate": 0.00018464351005484462,
|
| 722 |
+
"loss": 0.0488,
|
| 723 |
+
"step": 102
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 0.1412894375857339,
|
| 727 |
+
"grad_norm": 6.967692852020264,
|
| 728 |
+
"learning_rate": 0.00018647166361974407,
|
| 729 |
+
"loss": 0.5388,
|
| 730 |
+
"step": 103
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.14266117969821673,
|
| 734 |
+
"grad_norm": 6.324373245239258,
|
| 735 |
+
"learning_rate": 0.0001882998171846435,
|
| 736 |
+
"loss": 0.376,
|
| 737 |
+
"step": 104
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.1440329218106996,
|
| 741 |
+
"grad_norm": 0.7642683982849121,
|
| 742 |
+
"learning_rate": 0.00019012797074954296,
|
| 743 |
+
"loss": 0.017,
|
| 744 |
+
"step": 105
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.14540466392318244,
|
| 748 |
+
"grad_norm": 8.600672721862793,
|
| 749 |
+
"learning_rate": 0.00019195612431444244,
|
| 750 |
+
"loss": 0.7542,
|
| 751 |
+
"step": 106
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 0.1467764060356653,
|
| 755 |
+
"grad_norm": 7.111880302429199,
|
| 756 |
+
"learning_rate": 0.00019378427787934186,
|
| 757 |
+
"loss": 0.4063,
|
| 758 |
+
"step": 107
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 0.14814814814814814,
|
| 762 |
+
"grad_norm": 6.075577735900879,
|
| 763 |
+
"learning_rate": 0.00019561243144424132,
|
| 764 |
+
"loss": 0.3658,
|
| 765 |
+
"step": 108
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.149519890260631,
|
| 769 |
+
"grad_norm": 6.12313175201416,
|
| 770 |
+
"learning_rate": 0.00019744058500914077,
|
| 771 |
+
"loss": 0.4389,
|
| 772 |
+
"step": 109
|
| 773 |
+
},
|
| 774 |
{
|
| 775 |
"epoch": 0.15089163237311384,
|
| 776 |
+
"grad_norm": 5.813235759735107,
|
| 777 |
"learning_rate": 0.00019926873857404023,
|
| 778 |
+
"loss": 0.3803,
|
| 779 |
"step": 110
|
| 780 |
},
|
| 781 |
{
|
| 782 |
"epoch": 0.15089163237311384,
|
| 783 |
+
"eval_Qnli-dev_cosine_accuracy": 0.705078125,
|
| 784 |
+
"eval_Qnli-dev_cosine_accuracy_threshold": 0.6866907477378845,
|
| 785 |
+
"eval_Qnli-dev_cosine_ap": 0.7567018413685389,
|
| 786 |
+
"eval_Qnli-dev_cosine_f1": 0.6931818181818182,
|
| 787 |
+
"eval_Qnli-dev_cosine_f1_threshold": 0.6343963146209717,
|
| 788 |
+
"eval_Qnli-dev_cosine_precision": 0.6267123287671232,
|
| 789 |
+
"eval_Qnli-dev_cosine_recall": 0.7754237288135594,
|
| 790 |
+
"eval_allNLI-dev_cosine_accuracy": 0.76953125,
|
| 791 |
+
"eval_allNLI-dev_cosine_accuracy_threshold": 0.7752166986465454,
|
| 792 |
+
"eval_allNLI-dev_cosine_ap": 0.6627175481841632,
|
| 793 |
+
"eval_allNLI-dev_cosine_f1": 0.6624737945492662,
|
| 794 |
+
"eval_allNLI-dev_cosine_f1_threshold": 0.6564935445785522,
|
| 795 |
+
"eval_allNLI-dev_cosine_precision": 0.5197368421052632,
|
| 796 |
+
"eval_allNLI-dev_cosine_recall": 0.9132947976878613,
|
| 797 |
+
"eval_sequential_score": 0.7567018413685389,
|
| 798 |
+
"eval_sts-test_pearson_cosine": 0.9026620207137961,
|
| 799 |
+
"eval_sts-test_spearman_cosine": 0.913678627606199,
|
| 800 |
+
"eval_vitaminc-pairs_loss": 2.009296178817749,
|
| 801 |
+
"eval_vitaminc-pairs_runtime": 14.3224,
|
| 802 |
+
"eval_vitaminc-pairs_samples_per_second": 8.937,
|
| 803 |
+
"eval_vitaminc-pairs_steps_per_second": 0.07,
|
| 804 |
"step": 110
|
| 805 |
},
|
| 806 |
{
|
| 807 |
"epoch": 0.15089163237311384,
|
| 808 |
+
"eval_negation-triplets_loss": 1.59572434425354,
|
| 809 |
+
"eval_negation-triplets_runtime": 1.1528,
|
| 810 |
+
"eval_negation-triplets_samples_per_second": 111.029,
|
| 811 |
+
"eval_negation-triplets_steps_per_second": 0.867,
|
| 812 |
"step": 110
|
| 813 |
},
|
| 814 |
{
|
| 815 |
"epoch": 0.15089163237311384,
|
| 816 |
+
"eval_scitail-pairs-pos_loss": 0.061776161193847656,
|
| 817 |
+
"eval_scitail-pairs-pos_runtime": 1.5728,
|
| 818 |
+
"eval_scitail-pairs-pos_samples_per_second": 81.383,
|
| 819 |
+
"eval_scitail-pairs-pos_steps_per_second": 0.636,
|
| 820 |
"step": 110
|
| 821 |
},
|
| 822 |
{
|
| 823 |
"epoch": 0.15089163237311384,
|
| 824 |
+
"eval_scitail-pairs-qa_loss": 0.009187542833387852,
|
| 825 |
+
"eval_scitail-pairs-qa_runtime": 1.2102,
|
| 826 |
+
"eval_scitail-pairs-qa_samples_per_second": 105.771,
|
| 827 |
+
"eval_scitail-pairs-qa_steps_per_second": 0.826,
|
| 828 |
"step": 110
|
| 829 |
},
|
| 830 |
{
|
| 831 |
"epoch": 0.15089163237311384,
|
| 832 |
+
"eval_xsum-pairs_loss": 0.37210211157798767,
|
| 833 |
+
"eval_xsum-pairs_runtime": 6.2854,
|
| 834 |
+
"eval_xsum-pairs_samples_per_second": 20.365,
|
| 835 |
+
"eval_xsum-pairs_steps_per_second": 0.159,
|
| 836 |
"step": 110
|
| 837 |
},
|
| 838 |
{
|
| 839 |
"epoch": 0.15089163237311384,
|
| 840 |
+
"eval_sciq_pairs_loss": 0.04122849553823471,
|
| 841 |
+
"eval_sciq_pairs_runtime": 8.8116,
|
| 842 |
+
"eval_sciq_pairs_samples_per_second": 14.526,
|
| 843 |
+
"eval_sciq_pairs_steps_per_second": 0.113,
|
| 844 |
"step": 110
|
| 845 |
},
|
| 846 |
{
|
| 847 |
"epoch": 0.15089163237311384,
|
| 848 |
+
"eval_qasc_pairs_loss": 0.4748501479625702,
|
| 849 |
+
"eval_qasc_pairs_runtime": 1.3827,
|
| 850 |
+
"eval_qasc_pairs_samples_per_second": 92.573,
|
| 851 |
+
"eval_qasc_pairs_steps_per_second": 0.723,
|
| 852 |
"step": 110
|
| 853 |
},
|
| 854 |
{
|
| 855 |
"epoch": 0.15089163237311384,
|
| 856 |
+
"eval_openbookqa_pairs_loss": 1.1540580987930298,
|
| 857 |
+
"eval_openbookqa_pairs_runtime": 1.1788,
|
| 858 |
+
"eval_openbookqa_pairs_samples_per_second": 108.581,
|
| 859 |
+
"eval_openbookqa_pairs_steps_per_second": 0.848,
|
| 860 |
"step": 110
|
| 861 |
},
|
| 862 |
{
|
| 863 |
"epoch": 0.15089163237311384,
|
| 864 |
+
"eval_nq_pairs_loss": 0.2363465428352356,
|
| 865 |
+
"eval_nq_pairs_runtime": 7.8515,
|
| 866 |
+
"eval_nq_pairs_samples_per_second": 16.303,
|
| 867 |
+
"eval_nq_pairs_steps_per_second": 0.127,
|
| 868 |
"step": 110
|
| 869 |
},
|
| 870 |
{
|
| 871 |
"epoch": 0.15089163237311384,
|
| 872 |
+
"eval_trivia_pairs_loss": 0.6520176529884338,
|
| 873 |
+
"eval_trivia_pairs_runtime": 8.9067,
|
| 874 |
+
"eval_trivia_pairs_samples_per_second": 14.371,
|
| 875 |
+
"eval_trivia_pairs_steps_per_second": 0.112,
|
| 876 |
"step": 110
|
| 877 |
},
|
| 878 |
{
|
| 879 |
"epoch": 0.15089163237311384,
|
| 880 |
+
"eval_gooaq_pairs_loss": 0.22620199620723724,
|
| 881 |
+
"eval_gooaq_pairs_runtime": 2.067,
|
| 882 |
+
"eval_gooaq_pairs_samples_per_second": 61.924,
|
| 883 |
+
"eval_gooaq_pairs_steps_per_second": 0.484,
|
| 884 |
"step": 110
|
| 885 |
},
|
| 886 |
{
|
| 887 |
"epoch": 0.15089163237311384,
|
| 888 |
+
"eval_paws-pos_loss": 0.02822125516831875,
|
| 889 |
+
"eval_paws-pos_runtime": 1.5117,
|
| 890 |
+
"eval_paws-pos_samples_per_second": 84.672,
|
| 891 |
+
"eval_paws-pos_steps_per_second": 0.662,
|
| 892 |
"step": 110
|
| 893 |
},
|
| 894 |
{
|
| 895 |
"epoch": 0.15089163237311384,
|
| 896 |
+
"eval_global_dataset_loss": 0.30668479204177856,
|
| 897 |
+
"eval_global_dataset_runtime": 33.2591,
|
| 898 |
+
"eval_global_dataset_samples_per_second": 11.546,
|
| 899 |
+
"eval_global_dataset_steps_per_second": 0.06,
|
| 900 |
"step": 110
|
| 901 |
},
|
| 902 |
+
{
|
| 903 |
+
"epoch": 0.1522633744855967,
|
| 904 |
+
"grad_norm": 4.30504846572876,
|
| 905 |
+
"learning_rate": 0.00020109689213893968,
|
| 906 |
+
"loss": 0.2478,
|
| 907 |
+
"step": 111
|
| 908 |
+
},
|
| 909 |
+
{
|
| 910 |
+
"epoch": 0.15363511659807957,
|
| 911 |
+
"grad_norm": 6.559568881988525,
|
| 912 |
+
"learning_rate": 0.00020292504570383914,
|
| 913 |
+
"loss": 0.8402,
|
| 914 |
+
"step": 112
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 0.1550068587105624,
|
| 918 |
+
"grad_norm": 5.812280654907227,
|
| 919 |
+
"learning_rate": 0.0002047531992687386,
|
| 920 |
+
"loss": 0.6608,
|
| 921 |
+
"step": 113
|
| 922 |
+
},
|
| 923 |
+
{
|
| 924 |
+
"epoch": 0.15637860082304528,
|
| 925 |
+
"grad_norm": 2.0805885791778564,
|
| 926 |
+
"learning_rate": 0.00020658135283363802,
|
| 927 |
+
"loss": 0.0934,
|
| 928 |
+
"step": 114
|
| 929 |
+
},
|
| 930 |
+
{
|
| 931 |
+
"epoch": 0.15775034293552812,
|
| 932 |
+
"grad_norm": 5.199294090270996,
|
| 933 |
+
"learning_rate": 0.00020840950639853747,
|
| 934 |
+
"loss": 0.3907,
|
| 935 |
+
"step": 115
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 0.15912208504801098,
|
| 939 |
+
"grad_norm": 6.3685078620910645,
|
| 940 |
+
"learning_rate": 0.00021023765996343693,
|
| 941 |
+
"loss": 0.449,
|
| 942 |
+
"step": 116
|
| 943 |
+
},
|
| 944 |
+
{
|
| 945 |
+
"epoch": 0.16049382716049382,
|
| 946 |
+
"grad_norm": 6.4199652671813965,
|
| 947 |
+
"learning_rate": 0.00021206581352833638,
|
| 948 |
+
"loss": 0.4041,
|
| 949 |
+
"step": 117
|
| 950 |
+
},
|
| 951 |
+
{
|
| 952 |
+
"epoch": 0.16186556927297668,
|
| 953 |
+
"grad_norm": 6.015898704528809,
|
| 954 |
+
"learning_rate": 0.00021389396709323584,
|
| 955 |
+
"loss": 0.6749,
|
| 956 |
+
"step": 118
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 0.16323731138545952,
|
| 960 |
+
"grad_norm": 7.721911430358887,
|
| 961 |
+
"learning_rate": 0.0002157221206581353,
|
| 962 |
+
"loss": 0.4847,
|
| 963 |
+
"step": 119
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 0.1646090534979424,
|
| 967 |
+
"grad_norm": 1.8774610757827759,
|
| 968 |
+
"learning_rate": 0.00021755027422303474,
|
| 969 |
+
"loss": 0.0526,
|
| 970 |
+
"step": 120
|
| 971 |
+
},
|
| 972 |
{
|
| 973 |
"epoch": 0.16598079561042525,
|
| 974 |
+
"grad_norm": 8.094359397888184,
|
| 975 |
"learning_rate": 0.00021937842778793417,
|
| 976 |
+
"loss": 0.6795,
|
| 977 |
"step": 121
|
| 978 |
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 0.1673525377229081,
|
| 981 |
+
"grad_norm": 0.33090323209762573,
|
| 982 |
+
"learning_rate": 0.00022120658135283365,
|
| 983 |
+
"loss": 0.0064,
|
| 984 |
+
"step": 122
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 0.16872427983539096,
|
| 988 |
+
"grad_norm": 7.3609418869018555,
|
| 989 |
+
"learning_rate": 0.0002230347349177331,
|
| 990 |
+
"loss": 0.5918,
|
| 991 |
+
"step": 123
|
| 992 |
+
},
|
| 993 |
+
{
|
| 994 |
+
"epoch": 0.1700960219478738,
|
| 995 |
+
"grad_norm": 6.189216613769531,
|
| 996 |
+
"learning_rate": 0.00022486288848263253,
|
| 997 |
+
"loss": 0.3544,
|
| 998 |
+
"step": 124
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 0.17146776406035666,
|
| 1002 |
+
"grad_norm": 5.588890075683594,
|
| 1003 |
+
"learning_rate": 0.000226691042047532,
|
| 1004 |
+
"loss": 0.3849,
|
| 1005 |
+
"step": 125
|
| 1006 |
+
},
|
| 1007 |
+
{
|
| 1008 |
+
"epoch": 0.1728395061728395,
|
| 1009 |
+
"grad_norm": 3.4582345485687256,
|
| 1010 |
+
"learning_rate": 0.00022851919561243144,
|
| 1011 |
+
"loss": 0.2051,
|
| 1012 |
+
"step": 126
|
| 1013 |
+
},
|
| 1014 |
+
{
|
| 1015 |
+
"epoch": 0.17421124828532236,
|
| 1016 |
+
"grad_norm": 4.075862407684326,
|
| 1017 |
+
"learning_rate": 0.00023034734917733092,
|
| 1018 |
+
"loss": 0.2129,
|
| 1019 |
+
"step": 127
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 0.1755829903978052,
|
| 1023 |
+
"grad_norm": 15.110091209411621,
|
| 1024 |
+
"learning_rate": 0.00023217550274223035,
|
| 1025 |
+
"loss": 2.7937,
|
| 1026 |
+
"step": 128
|
| 1027 |
+
},
|
| 1028 |
+
{
|
| 1029 |
+
"epoch": 0.17695473251028807,
|
| 1030 |
+
"grad_norm": 0.35791516304016113,
|
| 1031 |
+
"learning_rate": 0.0002340036563071298,
|
| 1032 |
+
"loss": 0.0166,
|
| 1033 |
+
"step": 129
|
| 1034 |
+
},
|
| 1035 |
+
{
|
| 1036 |
+
"epoch": 0.17832647462277093,
|
| 1037 |
+
"grad_norm": 7.5200090408325195,
|
| 1038 |
+
"learning_rate": 0.00023583180987202926,
|
| 1039 |
+
"loss": 0.7856,
|
| 1040 |
+
"step": 130
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 0.17969821673525377,
|
| 1044 |
+
"grad_norm": 6.566864490509033,
|
| 1045 |
+
"learning_rate": 0.0002376599634369287,
|
| 1046 |
+
"loss": 0.8368,
|
| 1047 |
+
"step": 131
|
| 1048 |
+
},
|
| 1049 |
{
|
| 1050 |
"epoch": 0.18106995884773663,
|
| 1051 |
+
"grad_norm": 4.958701133728027,
|
| 1052 |
"learning_rate": 0.00023948811700182814,
|
| 1053 |
+
"loss": 0.3813,
|
| 1054 |
"step": 132
|
| 1055 |
},
|
| 1056 |
+
{
|
| 1057 |
+
"epoch": 0.18244170096021947,
|
| 1058 |
+
"grad_norm": 5.745133876800537,
|
| 1059 |
+
"learning_rate": 0.00024131627056672762,
|
| 1060 |
+
"loss": 0.5695,
|
| 1061 |
+
"step": 133
|
| 1062 |
+
},
|
| 1063 |
+
{
|
| 1064 |
+
"epoch": 0.18381344307270234,
|
| 1065 |
+
"grad_norm": 4.952736854553223,
|
| 1066 |
+
"learning_rate": 0.00024314442413162708,
|
| 1067 |
+
"loss": 0.351,
|
| 1068 |
+
"step": 134
|
| 1069 |
+
},
|
| 1070 |
+
{
|
| 1071 |
+
"epoch": 0.18518518518518517,
|
| 1072 |
+
"grad_norm": 5.733601093292236,
|
| 1073 |
+
"learning_rate": 0.0002449725776965265,
|
| 1074 |
+
"loss": 0.3821,
|
| 1075 |
+
"step": 135
|
| 1076 |
+
},
|
| 1077 |
+
{
|
| 1078 |
+
"epoch": 0.18655692729766804,
|
| 1079 |
+
"grad_norm": 5.019097328186035,
|
| 1080 |
+
"learning_rate": 0.00024680073126142596,
|
| 1081 |
+
"loss": 0.3249,
|
| 1082 |
+
"step": 136
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 0.18792866941015088,
|
| 1086 |
+
"grad_norm": 5.300777912139893,
|
| 1087 |
+
"learning_rate": 0.0002486288848263254,
|
| 1088 |
+
"loss": 0.3404,
|
| 1089 |
+
"step": 137
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 0.18930041152263374,
|
| 1093 |
+
"grad_norm": 4.518141269683838,
|
| 1094 |
+
"learning_rate": 0.00025045703839122487,
|
| 1095 |
+
"loss": 0.4535,
|
| 1096 |
+
"step": 138
|
| 1097 |
+
},
|
| 1098 |
+
{
|
| 1099 |
+
"epoch": 0.1906721536351166,
|
| 1100 |
+
"grad_norm": 1.0158088207244873,
|
| 1101 |
+
"learning_rate": 0.0002522851919561243,
|
| 1102 |
+
"loss": 0.0577,
|
| 1103 |
+
"step": 139
|
| 1104 |
+
},
|
| 1105 |
+
{
|
| 1106 |
+
"epoch": 0.19204389574759945,
|
| 1107 |
+
"grad_norm": 5.966796398162842,
|
| 1108 |
+
"learning_rate": 0.0002541133455210238,
|
| 1109 |
+
"loss": 0.7431,
|
| 1110 |
+
"step": 140
|
| 1111 |
+
},
|
| 1112 |
+
{
|
| 1113 |
+
"epoch": 0.1934156378600823,
|
| 1114 |
+
"grad_norm": 6.123642921447754,
|
| 1115 |
+
"learning_rate": 0.00025594149908592323,
|
| 1116 |
+
"loss": 0.6778,
|
| 1117 |
+
"step": 141
|
| 1118 |
+
},
|
| 1119 |
+
{
|
| 1120 |
+
"epoch": 0.19478737997256515,
|
| 1121 |
+
"grad_norm": 5.842874050140381,
|
| 1122 |
+
"learning_rate": 0.0002577696526508227,
|
| 1123 |
+
"loss": 0.5436,
|
| 1124 |
+
"step": 142
|
| 1125 |
+
},
|
| 1126 |
{
|
| 1127 |
"epoch": 0.19615912208504802,
|
| 1128 |
+
"grad_norm": 4.759068012237549,
|
| 1129 |
"learning_rate": 0.0002595978062157221,
|
| 1130 |
+
"loss": 0.3582,
|
| 1131 |
"step": 143
|
| 1132 |
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 0.19753086419753085,
|
| 1135 |
+
"grad_norm": 4.080338478088379,
|
| 1136 |
+
"learning_rate": 0.00026142595978062154,
|
| 1137 |
+
"loss": 0.316,
|
| 1138 |
+
"step": 144
|
| 1139 |
+
},
|
| 1140 |
+
{
|
| 1141 |
+
"epoch": 0.19890260631001372,
|
| 1142 |
+
"grad_norm": 4.1391448974609375,
|
| 1143 |
+
"learning_rate": 0.000263254113345521,
|
| 1144 |
+
"loss": 0.4446,
|
| 1145 |
+
"step": 145
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"epoch": 0.20027434842249658,
|
| 1149 |
+
"grad_norm": 5.856256008148193,
|
| 1150 |
+
"learning_rate": 0.0002650822669104205,
|
| 1151 |
+
"loss": 0.7792,
|
| 1152 |
+
"step": 146
|
| 1153 |
+
},
|
| 1154 |
+
{
|
| 1155 |
+
"epoch": 0.20164609053497942,
|
| 1156 |
+
"grad_norm": 7.747331142425537,
|
| 1157 |
+
"learning_rate": 0.00026691042047531996,
|
| 1158 |
+
"loss": 1.1147,
|
| 1159 |
+
"step": 147
|
| 1160 |
+
},
|
| 1161 |
+
{
|
| 1162 |
+
"epoch": 0.2030178326474623,
|
| 1163 |
+
"grad_norm": 6.825289249420166,
|
| 1164 |
+
"learning_rate": 0.0002687385740402194,
|
| 1165 |
+
"loss": 0.8267,
|
| 1166 |
+
"step": 148
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 0.20438957475994513,
|
| 1170 |
+
"grad_norm": 7.336719512939453,
|
| 1171 |
+
"learning_rate": 0.00027056672760511887,
|
| 1172 |
+
"loss": 0.8149,
|
| 1173 |
+
"step": 149
|
| 1174 |
+
},
|
| 1175 |
+
{
|
| 1176 |
+
"epoch": 0.205761316872428,
|
| 1177 |
+
"grad_norm": 6.731626510620117,
|
| 1178 |
+
"learning_rate": 0.00027239488117001827,
|
| 1179 |
+
"loss": 0.942,
|
| 1180 |
+
"step": 150
|
| 1181 |
+
},
|
| 1182 |
+
{
|
| 1183 |
+
"epoch": 0.20713305898491083,
|
| 1184 |
+
"grad_norm": 10.727692604064941,
|
| 1185 |
+
"learning_rate": 0.0002742230347349177,
|
| 1186 |
+
"loss": 2.4865,
|
| 1187 |
+
"step": 151
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"epoch": 0.2085048010973937,
|
| 1191 |
+
"grad_norm": 8.583380699157715,
|
| 1192 |
+
"learning_rate": 0.0002760511882998172,
|
| 1193 |
+
"loss": 1.0715,
|
| 1194 |
+
"step": 152
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 0.20987654320987653,
|
| 1198 |
+
"grad_norm": 6.236877918243408,
|
| 1199 |
+
"learning_rate": 0.00027787934186471663,
|
| 1200 |
+
"loss": 0.6219,
|
| 1201 |
+
"step": 153
|
| 1202 |
+
},
|
| 1203 |
{
|
| 1204 |
"epoch": 0.2112482853223594,
|
| 1205 |
+
"grad_norm": 6.254538536071777,
|
| 1206 |
"learning_rate": 0.0002797074954296161,
|
| 1207 |
+
"loss": 0.8705,
|
| 1208 |
"step": 154
|
| 1209 |
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 0.21262002743484226,
|
| 1212 |
+
"grad_norm": 3.0917959213256836,
|
| 1213 |
+
"learning_rate": 0.00028153564899451554,
|
| 1214 |
+
"loss": 0.2407,
|
| 1215 |
+
"step": 155
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 0.2139917695473251,
|
| 1219 |
+
"grad_norm": 4.438024997711182,
|
| 1220 |
+
"learning_rate": 0.000283363802559415,
|
| 1221 |
+
"loss": 0.4925,
|
| 1222 |
+
"step": 156
|
| 1223 |
+
},
|
| 1224 |
+
{
|
| 1225 |
+
"epoch": 0.21536351165980797,
|
| 1226 |
+
"grad_norm": 0.43344631791114807,
|
| 1227 |
+
"learning_rate": 0.00028519195612431445,
|
| 1228 |
+
"loss": 0.0316,
|
| 1229 |
+
"step": 157
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 0.2167352537722908,
|
| 1233 |
+
"grad_norm": 5.73934268951416,
|
| 1234 |
+
"learning_rate": 0.0002870201096892139,
|
| 1235 |
+
"loss": 0.3935,
|
| 1236 |
+
"step": 158
|
| 1237 |
+
},
|
| 1238 |
+
{
|
| 1239 |
+
"epoch": 0.21810699588477367,
|
| 1240 |
+
"grad_norm": 4.532804012298584,
|
| 1241 |
+
"learning_rate": 0.00028884826325411336,
|
| 1242 |
+
"loss": 0.2083,
|
| 1243 |
+
"step": 159
|
| 1244 |
+
},
|
| 1245 |
+
{
|
| 1246 |
+
"epoch": 0.2194787379972565,
|
| 1247 |
+
"grad_norm": 4.846848487854004,
|
| 1248 |
+
"learning_rate": 0.0002906764168190128,
|
| 1249 |
+
"loss": 0.2798,
|
| 1250 |
+
"step": 160
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 0.22085048010973937,
|
| 1254 |
+
"grad_norm": 7.060863018035889,
|
| 1255 |
+
"learning_rate": 0.00029250457038391227,
|
| 1256 |
+
"loss": 0.8777,
|
| 1257 |
+
"step": 161
|
| 1258 |
+
},
|
| 1259 |
+
{
|
| 1260 |
+
"epoch": 0.2222222222222222,
|
| 1261 |
+
"grad_norm": 0.012754157185554504,
|
| 1262 |
+
"learning_rate": 0.0002943327239488117,
|
| 1263 |
+
"loss": 0.0002,
|
| 1264 |
+
"step": 162
|
| 1265 |
+
},
|
| 1266 |
+
{
|
| 1267 |
+
"epoch": 0.22359396433470508,
|
| 1268 |
+
"grad_norm": 4.094379901885986,
|
| 1269 |
+
"learning_rate": 0.0002961608775137112,
|
| 1270 |
+
"loss": 0.2736,
|
| 1271 |
+
"step": 163
|
| 1272 |
+
},
|
| 1273 |
+
{
|
| 1274 |
+
"epoch": 0.22496570644718794,
|
| 1275 |
+
"grad_norm": 10.741785049438477,
|
| 1276 |
+
"learning_rate": 0.0002979890310786106,
|
| 1277 |
+
"loss": 2.4185,
|
| 1278 |
+
"step": 164
|
| 1279 |
+
},
|
| 1280 |
{
|
| 1281 |
"epoch": 0.22633744855967078,
|
| 1282 |
+
"grad_norm": 4.820891380310059,
|
| 1283 |
"learning_rate": 0.00029981718464351003,
|
| 1284 |
+
"loss": 0.7767,
|
| 1285 |
"step": 165
|
| 1286 |
},
|
| 1287 |
+
{
|
| 1288 |
+
"epoch": 0.22770919067215364,
|
| 1289 |
+
"grad_norm": 6.423076152801514,
|
| 1290 |
+
"learning_rate": 0.0003016453382084095,
|
| 1291 |
+
"loss": 0.7971,
|
| 1292 |
+
"step": 166
|
| 1293 |
+
},
|
| 1294 |
+
{
|
| 1295 |
+
"epoch": 0.22908093278463648,
|
| 1296 |
+
"grad_norm": 4.492727756500244,
|
| 1297 |
+
"learning_rate": 0.00030347349177330894,
|
| 1298 |
+
"loss": 0.4535,
|
| 1299 |
+
"step": 167
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 0.23045267489711935,
|
| 1303 |
+
"grad_norm": 5.301379680633545,
|
| 1304 |
+
"learning_rate": 0.00030530164533820845,
|
| 1305 |
+
"loss": 0.6654,
|
| 1306 |
+
"step": 168
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 0.23182441700960219,
|
| 1310 |
+
"grad_norm": 5.155853748321533,
|
| 1311 |
+
"learning_rate": 0.0003071297989031079,
|
| 1312 |
+
"loss": 0.3985,
|
| 1313 |
+
"step": 169
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 0.23319615912208505,
|
| 1317 |
+
"grad_norm": 0.4378865361213684,
|
| 1318 |
+
"learning_rate": 0.00030895795246800735,
|
| 1319 |
+
"loss": 0.0338,
|
| 1320 |
+
"step": 170
|
| 1321 |
+
},
|
| 1322 |
+
{
|
| 1323 |
+
"epoch": 0.2345679012345679,
|
| 1324 |
+
"grad_norm": 4.022473335266113,
|
| 1325 |
+
"learning_rate": 0.00031078610603290675,
|
| 1326 |
+
"loss": 0.1834,
|
| 1327 |
+
"step": 171
|
| 1328 |
+
},
|
| 1329 |
+
{
|
| 1330 |
+
"epoch": 0.23593964334705075,
|
| 1331 |
+
"grad_norm": 7.863429069519043,
|
| 1332 |
+
"learning_rate": 0.0003126142595978062,
|
| 1333 |
+
"loss": 0.603,
|
| 1334 |
+
"step": 172
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 0.23731138545953362,
|
| 1338 |
+
"grad_norm": 8.951998710632324,
|
| 1339 |
+
"learning_rate": 0.00031444241316270566,
|
| 1340 |
+
"loss": 0.7871,
|
| 1341 |
+
"step": 173
|
| 1342 |
+
},
|
| 1343 |
+
{
|
| 1344 |
+
"epoch": 0.23868312757201646,
|
| 1345 |
+
"grad_norm": 6.265102386474609,
|
| 1346 |
+
"learning_rate": 0.0003162705667276051,
|
| 1347 |
+
"loss": 0.4304,
|
| 1348 |
+
"step": 174
|
| 1349 |
+
},
|
| 1350 |
+
{
|
| 1351 |
+
"epoch": 0.24005486968449932,
|
| 1352 |
+
"grad_norm": 6.6486005783081055,
|
| 1353 |
+
"learning_rate": 0.00031809872029250457,
|
| 1354 |
+
"loss": 0.649,
|
| 1355 |
+
"step": 175
|
| 1356 |
+
},
|
| 1357 |
{
|
| 1358 |
"epoch": 0.24142661179698216,
|
| 1359 |
+
"grad_norm": 0.47100114822387695,
|
| 1360 |
"learning_rate": 0.000319926873857404,
|
| 1361 |
+
"loss": 0.048,
|
| 1362 |
"step": 176
|
| 1363 |
},
|
| 1364 |
+
{
|
| 1365 |
+
"epoch": 0.24279835390946503,
|
| 1366 |
+
"grad_norm": 4.884115695953369,
|
| 1367 |
+
"learning_rate": 0.0003217550274223035,
|
| 1368 |
+
"loss": 0.4079,
|
| 1369 |
+
"step": 177
|
| 1370 |
+
},
|
| 1371 |
+
{
|
| 1372 |
+
"epoch": 0.24417009602194786,
|
| 1373 |
+
"grad_norm": 4.508667469024658,
|
| 1374 |
+
"learning_rate": 0.0003235831809872029,
|
| 1375 |
+
"loss": 0.4627,
|
| 1376 |
+
"step": 178
|
| 1377 |
+
},
|
| 1378 |
+
{
|
| 1379 |
+
"epoch": 0.24554183813443073,
|
| 1380 |
+
"grad_norm": 3.22367262840271,
|
| 1381 |
+
"learning_rate": 0.0003254113345521024,
|
| 1382 |
+
"loss": 0.3703,
|
| 1383 |
+
"step": 179
|
| 1384 |
+
},
|
| 1385 |
+
{
|
| 1386 |
+
"epoch": 0.24691358024691357,
|
| 1387 |
+
"grad_norm": 7.695303916931152,
|
| 1388 |
+
"learning_rate": 0.00032723948811700184,
|
| 1389 |
+
"loss": 0.8343,
|
| 1390 |
+
"step": 180
|
| 1391 |
+
},
|
| 1392 |
+
{
|
| 1393 |
+
"epoch": 0.24828532235939643,
|
| 1394 |
+
"grad_norm": 7.249318599700928,
|
| 1395 |
+
"learning_rate": 0.0003290676416819013,
|
| 1396 |
+
"loss": 0.692,
|
| 1397 |
+
"step": 181
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 0.2496570644718793,
|
| 1401 |
+
"grad_norm": 11.686202049255371,
|
| 1402 |
+
"learning_rate": 0.00033089579524680075,
|
| 1403 |
+
"loss": 2.7071,
|
| 1404 |
+
"step": 182
|
| 1405 |
+
},
|
| 1406 |
+
{
|
| 1407 |
+
"epoch": 0.25102880658436216,
|
| 1408 |
+
"grad_norm": 6.061092376708984,
|
| 1409 |
+
"learning_rate": 0.0003327239488117002,
|
| 1410 |
+
"loss": 0.8451,
|
| 1411 |
+
"step": 183
|
| 1412 |
+
},
|
| 1413 |
+
{
|
| 1414 |
+
"epoch": 0.252400548696845,
|
| 1415 |
+
"grad_norm": 5.932607650756836,
|
| 1416 |
+
"learning_rate": 0.00033455210237659966,
|
| 1417 |
+
"loss": 0.635,
|
| 1418 |
+
"step": 184
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 0.25377229080932784,
|
| 1422 |
+
"grad_norm": 3.491114616394043,
|
| 1423 |
+
"learning_rate": 0.00033638025594149906,
|
| 1424 |
+
"loss": 0.312,
|
| 1425 |
+
"step": 185
|
| 1426 |
+
},
|
| 1427 |
+
{
|
| 1428 |
+
"epoch": 0.2551440329218107,
|
| 1429 |
+
"grad_norm": 6.4914164543151855,
|
| 1430 |
+
"learning_rate": 0.0003382084095063985,
|
| 1431 |
+
"loss": 0.6996,
|
| 1432 |
+
"step": 186
|
| 1433 |
+
},
|
| 1434 |
{
|
| 1435 |
"epoch": 0.25651577503429357,
|
| 1436 |
+
"grad_norm": 6.15857458114624,
|
| 1437 |
"learning_rate": 0.00034003656307129797,
|
| 1438 |
+
"loss": 0.4432,
|
| 1439 |
"step": 187
|
| 1440 |
},
|
| 1441 |
+
{
|
| 1442 |
+
"epoch": 0.2578875171467764,
|
| 1443 |
+
"grad_norm": 4.767185211181641,
|
| 1444 |
+
"learning_rate": 0.0003418647166361974,
|
| 1445 |
+
"loss": 0.375,
|
| 1446 |
+
"step": 188
|
| 1447 |
+
},
|
| 1448 |
+
{
|
| 1449 |
+
"epoch": 0.25925925925925924,
|
| 1450 |
+
"grad_norm": 7.944342613220215,
|
| 1451 |
+
"learning_rate": 0.00034369287020109693,
|
| 1452 |
+
"loss": 0.9366,
|
| 1453 |
+
"step": 189
|
| 1454 |
+
},
|
| 1455 |
+
{
|
| 1456 |
+
"epoch": 0.2606310013717421,
|
| 1457 |
+
"grad_norm": 6.573953628540039,
|
| 1458 |
+
"learning_rate": 0.0003455210237659964,
|
| 1459 |
+
"loss": 0.755,
|
| 1460 |
+
"step": 190
|
| 1461 |
+
},
|
| 1462 |
+
{
|
| 1463 |
+
"epoch": 0.262002743484225,
|
| 1464 |
+
"grad_norm": 4.173367023468018,
|
| 1465 |
+
"learning_rate": 0.00034734917733089584,
|
| 1466 |
+
"loss": 0.6068,
|
| 1467 |
+
"step": 191
|
| 1468 |
+
},
|
| 1469 |
+
{
|
| 1470 |
+
"epoch": 0.26337448559670784,
|
| 1471 |
+
"grad_norm": 5.26171875,
|
| 1472 |
+
"learning_rate": 0.00034917733089579524,
|
| 1473 |
+
"loss": 0.5336,
|
| 1474 |
+
"step": 192
|
| 1475 |
+
},
|
| 1476 |
+
{
|
| 1477 |
+
"epoch": 0.26474622770919065,
|
| 1478 |
+
"grad_norm": 6.669304370880127,
|
| 1479 |
+
"learning_rate": 0.0003510054844606947,
|
| 1480 |
+
"loss": 0.8783,
|
| 1481 |
+
"step": 193
|
| 1482 |
+
},
|
| 1483 |
+
{
|
| 1484 |
+
"epoch": 0.2661179698216735,
|
| 1485 |
+
"grad_norm": 4.4192938804626465,
|
| 1486 |
+
"learning_rate": 0.00035283363802559415,
|
| 1487 |
+
"loss": 0.3576,
|
| 1488 |
+
"step": 194
|
| 1489 |
+
},
|
| 1490 |
+
{
|
| 1491 |
+
"epoch": 0.2674897119341564,
|
| 1492 |
+
"grad_norm": 10.117819786071777,
|
| 1493 |
+
"learning_rate": 0.0003546617915904936,
|
| 1494 |
+
"loss": 2.1854,
|
| 1495 |
+
"step": 195
|
| 1496 |
+
},
|
| 1497 |
+
{
|
| 1498 |
+
"epoch": 0.26886145404663925,
|
| 1499 |
+
"grad_norm": 5.256247520446777,
|
| 1500 |
+
"learning_rate": 0.00035648994515539306,
|
| 1501 |
+
"loss": 0.7835,
|
| 1502 |
+
"step": 196
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 0.27023319615912206,
|
| 1506 |
+
"grad_norm": 5.784887313842773,
|
| 1507 |
+
"learning_rate": 0.0003583180987202925,
|
| 1508 |
+
"loss": 0.5668,
|
| 1509 |
+
"step": 197
|
| 1510 |
+
},
|
| 1511 |
{
|
| 1512 |
"epoch": 0.2716049382716049,
|
| 1513 |
+
"grad_norm": 4.977567672729492,
|
| 1514 |
"learning_rate": 0.00036014625228519197,
|
| 1515 |
+
"loss": 0.7033,
|
| 1516 |
"step": 198
|
| 1517 |
},
|
| 1518 |
+
{
|
| 1519 |
+
"epoch": 0.2729766803840878,
|
| 1520 |
+
"grad_norm": 0.011424711905419827,
|
| 1521 |
+
"learning_rate": 0.00036197440585009137,
|
| 1522 |
+
"loss": 0.0002,
|
| 1523 |
+
"step": 199
|
| 1524 |
+
},
|
| 1525 |
+
{
|
| 1526 |
+
"epoch": 0.27434842249657065,
|
| 1527 |
+
"grad_norm": 5.805008411407471,
|
| 1528 |
+
"learning_rate": 0.0003638025594149909,
|
| 1529 |
+
"loss": 0.5791,
|
| 1530 |
+
"step": 200
|
| 1531 |
+
},
|
| 1532 |
+
{
|
| 1533 |
+
"epoch": 0.2757201646090535,
|
| 1534 |
+
"grad_norm": 3.8826043605804443,
|
| 1535 |
+
"learning_rate": 0.00036563071297989033,
|
| 1536 |
+
"loss": 0.2697,
|
| 1537 |
+
"step": 201
|
| 1538 |
+
},
|
| 1539 |
+
{
|
| 1540 |
+
"epoch": 0.27709190672153633,
|
| 1541 |
+
"grad_norm": 6.563521385192871,
|
| 1542 |
+
"learning_rate": 0.0003674588665447898,
|
| 1543 |
+
"loss": 0.6261,
|
| 1544 |
+
"step": 202
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 0.2784636488340192,
|
| 1548 |
+
"grad_norm": 4.584529399871826,
|
| 1549 |
+
"learning_rate": 0.00036928702010968924,
|
| 1550 |
+
"loss": 0.3253,
|
| 1551 |
+
"step": 203
|
| 1552 |
+
},
|
| 1553 |
+
{
|
| 1554 |
+
"epoch": 0.27983539094650206,
|
| 1555 |
+
"grad_norm": 6.636009216308594,
|
| 1556 |
+
"learning_rate": 0.0003711151736745887,
|
| 1557 |
+
"loss": 0.8323,
|
| 1558 |
+
"step": 204
|
| 1559 |
+
},
|
| 1560 |
+
{
|
| 1561 |
+
"epoch": 0.2812071330589849,
|
| 1562 |
+
"grad_norm": 5.0911359786987305,
|
| 1563 |
+
"learning_rate": 0.00037294332723948815,
|
| 1564 |
+
"loss": 0.4472,
|
| 1565 |
+
"step": 205
|
| 1566 |
+
},
|
| 1567 |
+
{
|
| 1568 |
+
"epoch": 0.2825788751714678,
|
| 1569 |
+
"grad_norm": 3.9219255447387695,
|
| 1570 |
+
"learning_rate": 0.00037477148080438755,
|
| 1571 |
+
"loss": 0.3342,
|
| 1572 |
+
"step": 206
|
| 1573 |
+
},
|
| 1574 |
+
{
|
| 1575 |
+
"epoch": 0.2839506172839506,
|
| 1576 |
+
"grad_norm": 5.114777565002441,
|
| 1577 |
+
"learning_rate": 0.000376599634369287,
|
| 1578 |
+
"loss": 0.6313,
|
| 1579 |
+
"step": 207
|
| 1580 |
+
},
|
| 1581 |
+
{
|
| 1582 |
+
"epoch": 0.28532235939643347,
|
| 1583 |
+
"grad_norm": 0.3298715353012085,
|
| 1584 |
+
"learning_rate": 0.00037842778793418646,
|
| 1585 |
+
"loss": 0.059,
|
| 1586 |
+
"step": 208
|
| 1587 |
+
},
|
| 1588 |
{
|
| 1589 |
"epoch": 0.28669410150891633,
|
| 1590 |
+
"grad_norm": 1.5965046882629395,
|
| 1591 |
"learning_rate": 0.0003802559414990859,
|
| 1592 |
+
"loss": 0.1195,
|
| 1593 |
"step": 209
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 0.2880658436213992,
|
| 1597 |
+
"grad_norm": 0.39121323823928833,
|
| 1598 |
+
"learning_rate": 0.00038208409506398537,
|
| 1599 |
+
"loss": 0.0296,
|
| 1600 |
+
"step": 210
|
| 1601 |
+
},
|
| 1602 |
+
{
|
| 1603 |
+
"epoch": 0.289437585733882,
|
| 1604 |
+
"grad_norm": 4.317224025726318,
|
| 1605 |
+
"learning_rate": 0.0003839122486288849,
|
| 1606 |
+
"loss": 0.5316,
|
| 1607 |
+
"step": 211
|
| 1608 |
+
},
|
| 1609 |
+
{
|
| 1610 |
+
"epoch": 0.2908093278463649,
|
| 1611 |
+
"grad_norm": 4.000308036804199,
|
| 1612 |
+
"learning_rate": 0.00038574040219378433,
|
| 1613 |
+
"loss": 0.5201,
|
| 1614 |
+
"step": 212
|
| 1615 |
+
},
|
| 1616 |
+
{
|
| 1617 |
+
"epoch": 0.29218106995884774,
|
| 1618 |
+
"grad_norm": 6.2192301750183105,
|
| 1619 |
+
"learning_rate": 0.00038756855575868373,
|
| 1620 |
+
"loss": 0.6602,
|
| 1621 |
+
"step": 213
|
| 1622 |
+
},
|
| 1623 |
+
{
|
| 1624 |
+
"epoch": 0.2935528120713306,
|
| 1625 |
+
"grad_norm": 6.702320098876953,
|
| 1626 |
+
"learning_rate": 0.0003893967093235832,
|
| 1627 |
+
"loss": 0.9578,
|
| 1628 |
+
"step": 214
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 0.29492455418381347,
|
| 1632 |
+
"grad_norm": 3.9136242866516113,
|
| 1633 |
+
"learning_rate": 0.00039122486288848264,
|
| 1634 |
+
"loss": 0.2089,
|
| 1635 |
+
"step": 215
|
| 1636 |
+
},
|
| 1637 |
+
{
|
| 1638 |
+
"epoch": 0.2962962962962963,
|
| 1639 |
+
"grad_norm": 6.901303768157959,
|
| 1640 |
+
"learning_rate": 0.0003930530164533821,
|
| 1641 |
+
"loss": 1.2112,
|
| 1642 |
+
"step": 216
|
| 1643 |
+
},
|
| 1644 |
+
{
|
| 1645 |
+
"epoch": 0.29766803840877915,
|
| 1646 |
+
"grad_norm": 4.04884672164917,
|
| 1647 |
+
"learning_rate": 0.00039488117001828155,
|
| 1648 |
+
"loss": 0.3294,
|
| 1649 |
+
"step": 217
|
| 1650 |
+
},
|
| 1651 |
+
{
|
| 1652 |
+
"epoch": 0.299039780521262,
|
| 1653 |
+
"grad_norm": 5.46201753616333,
|
| 1654 |
+
"learning_rate": 0.000396709323583181,
|
| 1655 |
+
"loss": 0.867,
|
| 1656 |
+
"step": 218
|
| 1657 |
+
},
|
| 1658 |
+
{
|
| 1659 |
+
"epoch": 0.3004115226337449,
|
| 1660 |
+
"grad_norm": 5.559458255767822,
|
| 1661 |
+
"learning_rate": 0.00039853747714808046,
|
| 1662 |
+
"loss": 1.1745,
|
| 1663 |
+
"step": 219
|
| 1664 |
}
|
| 1665 |
],
|
| 1666 |
+
"logging_steps": 1,
|
| 1667 |
"max_steps": 2187,
|
| 1668 |
"num_input_tokens_seen": 0,
|
| 1669 |
"num_train_epochs": 3,
|
checkpoint-219/training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5880
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:845321cafd8553e30026f9c6fea66eb5820d6f58e9ea4a0042d4e1d71fd924e8
|
| 3 |
size 5880
|