Commit
·
ee2f6cb
1
Parent(s):
1af248d
add .pt and more
Browse files- README.md +1 -1
- config.json +1 -1
- optimizer.pt +3 -0
- rng_state.pth +3 -0
- scheduler.pt +3 -0
- trainer_state.json +726 -0
- training_args.bin +3 -0
README.md
CHANGED
|
@@ -219,7 +219,7 @@ Then you can load this model and run inference.
|
|
| 219 |
from sentence_transformers import SentenceTransformer
|
| 220 |
|
| 221 |
# Download from the 🤗 Hub
|
| 222 |
-
model = SentenceTransformer("
|
| 223 |
# Run inference
|
| 224 |
sentences = [
|
| 225 |
'Thời điểm đánh giá và xếp loại chất lượng hằng năm của công chức, viên chức thuộc Bộ Tài chính được diễn ra trong thời gian nào?',
|
|
|
|
| 219 |
from sentence_transformers import SentenceTransformer
|
| 220 |
|
| 221 |
# Download from the 🤗 Hub
|
| 222 |
+
model = SentenceTransformer("sentence_transformers_model_id")
|
| 223 |
# Run inference
|
| 224 |
sentences = [
|
| 225 |
'Thời điểm đánh giá và xếp loại chất lượng hằng năm của công chức, viên chức thuộc Bộ Tài chính được diễn ra trong thời gian nào?',
|
config.json
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
{
|
| 2 |
-
"_name_or_path": "
|
| 3 |
"architectures": [
|
| 4 |
"BertModel"
|
| 5 |
],
|
|
|
|
| 1 |
{
|
| 2 |
+
"_name_or_path": "sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2",
|
| 3 |
"architectures": [
|
| 4 |
"BertModel"
|
| 5 |
],
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3f971d42ed4d8859ff1dd26a6b7c3b27fa9da7f33fca74bf6d474e34c908ea54
|
| 3 |
+
size 940241530
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6b75d832c9221ae5275522a6baf3fd939bec9a483fa1ec7e29db973616a5ad9a
|
| 3 |
+
size 14244
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:64377763af21ac94eb137e849fbe9f4bcb15667c724f28a26ed2d71dae752c3c
|
| 3 |
+
size 1064
|
trainer_state.json
ADDED
|
@@ -0,0 +1,726 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.6076154469349176,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 49500,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.00613752976701937,
|
| 13 |
+
"grad_norm": 0.7345873117446899,
|
| 14 |
+
"learning_rate": 4.9897707837216344e-05,
|
| 15 |
+
"loss": 1.0473,
|
| 16 |
+
"step": 500
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.01227505953403874,
|
| 20 |
+
"grad_norm": 1.5779420137405396,
|
| 21 |
+
"learning_rate": 4.9795415674432687e-05,
|
| 22 |
+
"loss": 1.0447,
|
| 23 |
+
"step": 1000
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.01841258930105811,
|
| 27 |
+
"grad_norm": 1.1132322549819946,
|
| 28 |
+
"learning_rate": 4.969312351164903e-05,
|
| 29 |
+
"loss": 1.0383,
|
| 30 |
+
"step": 1500
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.02455011906807748,
|
| 34 |
+
"grad_norm": 1.4636728763580322,
|
| 35 |
+
"learning_rate": 4.959083134886538e-05,
|
| 36 |
+
"loss": 1.0395,
|
| 37 |
+
"step": 2000
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.03068764883509685,
|
| 41 |
+
"grad_norm": 2.259823799133301,
|
| 42 |
+
"learning_rate": 4.948853918608172e-05,
|
| 43 |
+
"loss": 1.0436,
|
| 44 |
+
"step": 2500
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.03682517860211622,
|
| 48 |
+
"grad_norm": 3.276669979095459,
|
| 49 |
+
"learning_rate": 4.938624702329807e-05,
|
| 50 |
+
"loss": 1.0375,
|
| 51 |
+
"step": 3000
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.04296270836913559,
|
| 55 |
+
"grad_norm": 2.432560443878174,
|
| 56 |
+
"learning_rate": 4.928395486051441e-05,
|
| 57 |
+
"loss": 1.0189,
|
| 58 |
+
"step": 3500
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.04910023813615496,
|
| 62 |
+
"grad_norm": 2.106665849685669,
|
| 63 |
+
"learning_rate": 4.918166269773075e-05,
|
| 64 |
+
"loss": 1.0282,
|
| 65 |
+
"step": 4000
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.05523776790317433,
|
| 69 |
+
"grad_norm": 10.379436492919922,
|
| 70 |
+
"learning_rate": 4.9079370534947094e-05,
|
| 71 |
+
"loss": 1.0355,
|
| 72 |
+
"step": 4500
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.0613752976701937,
|
| 76 |
+
"grad_norm": 1.7805677652359009,
|
| 77 |
+
"learning_rate": 4.897707837216344e-05,
|
| 78 |
+
"loss": 1.0286,
|
| 79 |
+
"step": 5000
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.06751282743721307,
|
| 83 |
+
"grad_norm": 4.699483394622803,
|
| 84 |
+
"learning_rate": 4.8874786209379785e-05,
|
| 85 |
+
"loss": 1.0264,
|
| 86 |
+
"step": 5500
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.07365035720423244,
|
| 90 |
+
"grad_norm": 3.2066714763641357,
|
| 91 |
+
"learning_rate": 4.877249404659613e-05,
|
| 92 |
+
"loss": 1.0174,
|
| 93 |
+
"step": 6000
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.07978788697125182,
|
| 97 |
+
"grad_norm": 4.304272651672363,
|
| 98 |
+
"learning_rate": 4.867020188381247e-05,
|
| 99 |
+
"loss": 1.0238,
|
| 100 |
+
"step": 6500
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.08592541673827118,
|
| 104 |
+
"grad_norm": 1.9387307167053223,
|
| 105 |
+
"learning_rate": 4.856790972102881e-05,
|
| 106 |
+
"loss": 1.0217,
|
| 107 |
+
"step": 7000
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.09206294650529055,
|
| 111 |
+
"grad_norm": 7.0195488929748535,
|
| 112 |
+
"learning_rate": 4.846561755824516e-05,
|
| 113 |
+
"loss": 1.0203,
|
| 114 |
+
"step": 7500
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.09820047627230992,
|
| 118 |
+
"grad_norm": 5.841785907745361,
|
| 119 |
+
"learning_rate": 4.836332539546151e-05,
|
| 120 |
+
"loss": 1.0201,
|
| 121 |
+
"step": 8000
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.10433800603932929,
|
| 125 |
+
"grad_norm": 4.427718162536621,
|
| 126 |
+
"learning_rate": 4.826103323267785e-05,
|
| 127 |
+
"loss": 1.0266,
|
| 128 |
+
"step": 8500
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.11047553580634865,
|
| 132 |
+
"grad_norm": 2.067218780517578,
|
| 133 |
+
"learning_rate": 4.815874106989419e-05,
|
| 134 |
+
"loss": 1.0379,
|
| 135 |
+
"step": 9000
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.11661306557336804,
|
| 139 |
+
"grad_norm": 4.303891181945801,
|
| 140 |
+
"learning_rate": 4.8056448907110535e-05,
|
| 141 |
+
"loss": 1.0367,
|
| 142 |
+
"step": 9500
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.1227505953403874,
|
| 146 |
+
"grad_norm": 5.363577842712402,
|
| 147 |
+
"learning_rate": 4.795415674432688e-05,
|
| 148 |
+
"loss": 1.0384,
|
| 149 |
+
"step": 10000
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.12888812510740677,
|
| 153 |
+
"grad_norm": 7.8037028312683105,
|
| 154 |
+
"learning_rate": 4.785186458154322e-05,
|
| 155 |
+
"loss": 1.0291,
|
| 156 |
+
"step": 10500
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.13502565487442614,
|
| 160 |
+
"grad_norm": 2.8838307857513428,
|
| 161 |
+
"learning_rate": 4.774957241875956e-05,
|
| 162 |
+
"loss": 1.0362,
|
| 163 |
+
"step": 11000
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.1411631846414455,
|
| 167 |
+
"grad_norm": 18.838226318359375,
|
| 168 |
+
"learning_rate": 4.764728025597591e-05,
|
| 169 |
+
"loss": 1.0354,
|
| 170 |
+
"step": 11500
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.14730071440846487,
|
| 174 |
+
"grad_norm": 1.8900924921035767,
|
| 175 |
+
"learning_rate": 4.754498809319225e-05,
|
| 176 |
+
"loss": 1.0204,
|
| 177 |
+
"step": 12000
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.15343824417548424,
|
| 181 |
+
"grad_norm": 8.636752128601074,
|
| 182 |
+
"learning_rate": 4.74426959304086e-05,
|
| 183 |
+
"loss": 1.0401,
|
| 184 |
+
"step": 12500
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.15957577394250363,
|
| 188 |
+
"grad_norm": 3.1361780166625977,
|
| 189 |
+
"learning_rate": 4.734040376762494e-05,
|
| 190 |
+
"loss": 1.0237,
|
| 191 |
+
"step": 13000
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.165713303709523,
|
| 195 |
+
"grad_norm": 6.990635871887207,
|
| 196 |
+
"learning_rate": 4.7238111604841285e-05,
|
| 197 |
+
"loss": 1.0271,
|
| 198 |
+
"step": 13500
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.17185083347654237,
|
| 202 |
+
"grad_norm": 2.6320559978485107,
|
| 203 |
+
"learning_rate": 4.713581944205763e-05,
|
| 204 |
+
"loss": 1.0235,
|
| 205 |
+
"step": 14000
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.17798836324356174,
|
| 209 |
+
"grad_norm": 5.751638889312744,
|
| 210 |
+
"learning_rate": 4.7033527279273976e-05,
|
| 211 |
+
"loss": 1.0329,
|
| 212 |
+
"step": 14500
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.1841258930105811,
|
| 216 |
+
"grad_norm": 10.333562850952148,
|
| 217 |
+
"learning_rate": 4.693123511649032e-05,
|
| 218 |
+
"loss": 1.0474,
|
| 219 |
+
"step": 15000
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.19026342277760047,
|
| 223 |
+
"grad_norm": 9.57418155670166,
|
| 224 |
+
"learning_rate": 4.682894295370666e-05,
|
| 225 |
+
"loss": 1.0547,
|
| 226 |
+
"step": 15500
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.19640095254461984,
|
| 230 |
+
"grad_norm": 6.002443313598633,
|
| 231 |
+
"learning_rate": 4.6726650790923e-05,
|
| 232 |
+
"loss": 1.0557,
|
| 233 |
+
"step": 16000
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.2025384823116392,
|
| 237 |
+
"grad_norm": 17.77951431274414,
|
| 238 |
+
"learning_rate": 4.6624358628139344e-05,
|
| 239 |
+
"loss": 1.0626,
|
| 240 |
+
"step": 16500
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.20867601207865857,
|
| 244 |
+
"grad_norm": 10.126110076904297,
|
| 245 |
+
"learning_rate": 4.652206646535569e-05,
|
| 246 |
+
"loss": 1.0551,
|
| 247 |
+
"step": 17000
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.21481354184567794,
|
| 251 |
+
"grad_norm": 3.7376019954681396,
|
| 252 |
+
"learning_rate": 4.641977430257204e-05,
|
| 253 |
+
"loss": 1.0526,
|
| 254 |
+
"step": 17500
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.2209510716126973,
|
| 258 |
+
"grad_norm": 19.0529842376709,
|
| 259 |
+
"learning_rate": 4.6317482139788384e-05,
|
| 260 |
+
"loss": 1.125,
|
| 261 |
+
"step": 18000
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.2270886013797167,
|
| 265 |
+
"grad_norm": 1.9759446382522583,
|
| 266 |
+
"learning_rate": 4.6215189977004726e-05,
|
| 267 |
+
"loss": 1.2996,
|
| 268 |
+
"step": 18500
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.23322613114673607,
|
| 272 |
+
"grad_norm": 2.7964680194854736,
|
| 273 |
+
"learning_rate": 4.611289781422107e-05,
|
| 274 |
+
"loss": 1.0703,
|
| 275 |
+
"step": 19000
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.23936366091375544,
|
| 279 |
+
"grad_norm": 18.77906036376953,
|
| 280 |
+
"learning_rate": 4.601060565143741e-05,
|
| 281 |
+
"loss": 1.0601,
|
| 282 |
+
"step": 19500
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.2455011906807748,
|
| 286 |
+
"grad_norm": 3.129147529602051,
|
| 287 |
+
"learning_rate": 4.590831348865375e-05,
|
| 288 |
+
"loss": 1.0835,
|
| 289 |
+
"step": 20000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.2516387204477942,
|
| 293 |
+
"grad_norm": 3.018688917160034,
|
| 294 |
+
"learning_rate": 4.58060213258701e-05,
|
| 295 |
+
"loss": 1.0583,
|
| 296 |
+
"step": 20500
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.25777625021481354,
|
| 300 |
+
"grad_norm": 32.16483688354492,
|
| 301 |
+
"learning_rate": 4.570372916308644e-05,
|
| 302 |
+
"loss": 1.141,
|
| 303 |
+
"step": 21000
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.2639137799818329,
|
| 307 |
+
"grad_norm": 8.431964874267578,
|
| 308 |
+
"learning_rate": 4.5601437000302785e-05,
|
| 309 |
+
"loss": 1.0802,
|
| 310 |
+
"step": 21500
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.2700513097488523,
|
| 314 |
+
"grad_norm": 14.990534782409668,
|
| 315 |
+
"learning_rate": 4.5499144837519134e-05,
|
| 316 |
+
"loss": 1.0589,
|
| 317 |
+
"step": 22000
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.27618883951587164,
|
| 321 |
+
"grad_norm": 6.134013652801514,
|
| 322 |
+
"learning_rate": 4.5396852674735476e-05,
|
| 323 |
+
"loss": 1.086,
|
| 324 |
+
"step": 22500
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.282326369282891,
|
| 328 |
+
"grad_norm": 15.587114334106445,
|
| 329 |
+
"learning_rate": 4.529456051195182e-05,
|
| 330 |
+
"loss": 1.0743,
|
| 331 |
+
"step": 23000
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.2884638990499104,
|
| 335 |
+
"grad_norm": 20.36590003967285,
|
| 336 |
+
"learning_rate": 4.519226834916816e-05,
|
| 337 |
+
"loss": 1.0605,
|
| 338 |
+
"step": 23500
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.29460142881692974,
|
| 342 |
+
"grad_norm": 6.840169906616211,
|
| 343 |
+
"learning_rate": 4.508997618638451e-05,
|
| 344 |
+
"loss": 1.0602,
|
| 345 |
+
"step": 24000
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.3007389585839491,
|
| 349 |
+
"grad_norm": 22.10498046875,
|
| 350 |
+
"learning_rate": 4.498768402360085e-05,
|
| 351 |
+
"loss": 1.0732,
|
| 352 |
+
"step": 24500
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.3068764883509685,
|
| 356 |
+
"grad_norm": 2.3044190406799316,
|
| 357 |
+
"learning_rate": 4.488539186081719e-05,
|
| 358 |
+
"loss": 1.0614,
|
| 359 |
+
"step": 25000
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.31301401811798785,
|
| 363 |
+
"grad_norm": 19.358158111572266,
|
| 364 |
+
"learning_rate": 4.4783099698033535e-05,
|
| 365 |
+
"loss": 1.0666,
|
| 366 |
+
"step": 25500
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.31915154788500727,
|
| 370 |
+
"grad_norm": 47.77769088745117,
|
| 371 |
+
"learning_rate": 4.468080753524988e-05,
|
| 372 |
+
"loss": 1.0669,
|
| 373 |
+
"step": 26000
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.32528907765202664,
|
| 377 |
+
"grad_norm": 31.26322364807129,
|
| 378 |
+
"learning_rate": 4.4578515372466226e-05,
|
| 379 |
+
"loss": 1.0627,
|
| 380 |
+
"step": 26500
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.331426607419046,
|
| 384 |
+
"grad_norm": 31.237651824951172,
|
| 385 |
+
"learning_rate": 4.4476223209682575e-05,
|
| 386 |
+
"loss": 1.0659,
|
| 387 |
+
"step": 27000
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.33756413718606537,
|
| 391 |
+
"grad_norm": 10.651726722717285,
|
| 392 |
+
"learning_rate": 4.437393104689892e-05,
|
| 393 |
+
"loss": 1.07,
|
| 394 |
+
"step": 27500
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.34370166695308474,
|
| 398 |
+
"grad_norm": 10.767593383789062,
|
| 399 |
+
"learning_rate": 4.427163888411526e-05,
|
| 400 |
+
"loss": 1.0783,
|
| 401 |
+
"step": 28000
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.3498391967201041,
|
| 405 |
+
"grad_norm": 13.177048683166504,
|
| 406 |
+
"learning_rate": 4.41693467213316e-05,
|
| 407 |
+
"loss": 1.078,
|
| 408 |
+
"step": 28500
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.3559767264871235,
|
| 412 |
+
"grad_norm": 41.34538269042969,
|
| 413 |
+
"learning_rate": 4.406705455854794e-05,
|
| 414 |
+
"loss": 1.0832,
|
| 415 |
+
"step": 29000
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.36211425625414284,
|
| 419 |
+
"grad_norm": 2.813678026199341,
|
| 420 |
+
"learning_rate": 4.3964762395764285e-05,
|
| 421 |
+
"loss": 1.0695,
|
| 422 |
+
"step": 29500
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.3682517860211622,
|
| 426 |
+
"grad_norm": 32.63833236694336,
|
| 427 |
+
"learning_rate": 4.3862470232980634e-05,
|
| 428 |
+
"loss": 1.0714,
|
| 429 |
+
"step": 30000
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.3743893157881816,
|
| 433 |
+
"grad_norm": 6.144928932189941,
|
| 434 |
+
"learning_rate": 4.3760178070196976e-05,
|
| 435 |
+
"loss": 1.3794,
|
| 436 |
+
"step": 30500
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 0.38052684555520094,
|
| 440 |
+
"grad_norm": 21.691205978393555,
|
| 441 |
+
"learning_rate": 4.365788590741332e-05,
|
| 442 |
+
"loss": 1.0838,
|
| 443 |
+
"step": 31000
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.3866643753222203,
|
| 447 |
+
"grad_norm": 5.830935955047607,
|
| 448 |
+
"learning_rate": 4.355559374462967e-05,
|
| 449 |
+
"loss": 1.0541,
|
| 450 |
+
"step": 31500
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.3928019050892397,
|
| 454 |
+
"grad_norm": 5.59116792678833,
|
| 455 |
+
"learning_rate": 4.345330158184601e-05,
|
| 456 |
+
"loss": 1.0799,
|
| 457 |
+
"step": 32000
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.39893943485625905,
|
| 461 |
+
"grad_norm": 5.144099712371826,
|
| 462 |
+
"learning_rate": 4.335100941906235e-05,
|
| 463 |
+
"loss": 1.0622,
|
| 464 |
+
"step": 32500
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.4050769646232784,
|
| 468 |
+
"grad_norm": 30.603145599365234,
|
| 469 |
+
"learning_rate": 4.324871725627869e-05,
|
| 470 |
+
"loss": 1.0597,
|
| 471 |
+
"step": 33000
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.4112144943902978,
|
| 475 |
+
"grad_norm": 14.602336883544922,
|
| 476 |
+
"learning_rate": 4.314642509349504e-05,
|
| 477 |
+
"loss": 1.0731,
|
| 478 |
+
"step": 33500
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 0.41735202415731715,
|
| 482 |
+
"grad_norm": 11.165698051452637,
|
| 483 |
+
"learning_rate": 4.3044132930711384e-05,
|
| 484 |
+
"loss": 1.0871,
|
| 485 |
+
"step": 34000
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.4234895539243365,
|
| 489 |
+
"grad_norm": 7.897899627685547,
|
| 490 |
+
"learning_rate": 4.2941840767927726e-05,
|
| 491 |
+
"loss": 1.0535,
|
| 492 |
+
"step": 34500
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 0.4296270836913559,
|
| 496 |
+
"grad_norm": 34.061134338378906,
|
| 497 |
+
"learning_rate": 4.283954860514407e-05,
|
| 498 |
+
"loss": 1.3215,
|
| 499 |
+
"step": 35000
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 0.43576461345837525,
|
| 503 |
+
"grad_norm": 5.050889015197754,
|
| 504 |
+
"learning_rate": 4.273725644236041e-05,
|
| 505 |
+
"loss": 1.1501,
|
| 506 |
+
"step": 35500
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.4419021432253946,
|
| 510 |
+
"grad_norm": 11.23651123046875,
|
| 511 |
+
"learning_rate": 4.263496427957676e-05,
|
| 512 |
+
"loss": 1.1088,
|
| 513 |
+
"step": 36000
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.44803967299241404,
|
| 517 |
+
"grad_norm": 11.963942527770996,
|
| 518 |
+
"learning_rate": 4.25326721167931e-05,
|
| 519 |
+
"loss": 1.0844,
|
| 520 |
+
"step": 36500
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.4541772027594334,
|
| 524 |
+
"grad_norm": 19.694747924804688,
|
| 525 |
+
"learning_rate": 4.243037995400945e-05,
|
| 526 |
+
"loss": 1.0981,
|
| 527 |
+
"step": 37000
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.4603147325264528,
|
| 531 |
+
"grad_norm": 32.396446228027344,
|
| 532 |
+
"learning_rate": 4.232808779122579e-05,
|
| 533 |
+
"loss": 1.0856,
|
| 534 |
+
"step": 37500
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.46645226229347214,
|
| 538 |
+
"grad_norm": 30.794092178344727,
|
| 539 |
+
"learning_rate": 4.2225795628442134e-05,
|
| 540 |
+
"loss": 1.0956,
|
| 541 |
+
"step": 38000
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 0.4725897920604915,
|
| 545 |
+
"grad_norm": 25.20745086669922,
|
| 546 |
+
"learning_rate": 4.2123503465658476e-05,
|
| 547 |
+
"loss": 1.0813,
|
| 548 |
+
"step": 38500
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.4787273218275109,
|
| 552 |
+
"grad_norm": 13.46246337890625,
|
| 553 |
+
"learning_rate": 4.202121130287482e-05,
|
| 554 |
+
"loss": 1.0843,
|
| 555 |
+
"step": 39000
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.48486485159453024,
|
| 559 |
+
"grad_norm": 9.27540397644043,
|
| 560 |
+
"learning_rate": 4.191891914009117e-05,
|
| 561 |
+
"loss": 1.1053,
|
| 562 |
+
"step": 39500
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 0.4910023813615496,
|
| 566 |
+
"grad_norm": 3.5075111389160156,
|
| 567 |
+
"learning_rate": 4.181662697730751e-05,
|
| 568 |
+
"loss": 1.092,
|
| 569 |
+
"step": 40000
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.497139911128569,
|
| 573 |
+
"grad_norm": 6.219689846038818,
|
| 574 |
+
"learning_rate": 4.171433481452385e-05,
|
| 575 |
+
"loss": 1.081,
|
| 576 |
+
"step": 40500
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 0.5032774408955883,
|
| 580 |
+
"grad_norm": 16.255443572998047,
|
| 581 |
+
"learning_rate": 4.161204265174019e-05,
|
| 582 |
+
"loss": 1.0919,
|
| 583 |
+
"step": 41000
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 0.5094149706626078,
|
| 587 |
+
"grad_norm": 32.27680206298828,
|
| 588 |
+
"learning_rate": 4.150975048895654e-05,
|
| 589 |
+
"loss": 1.0681,
|
| 590 |
+
"step": 41500
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.5155525004296271,
|
| 594 |
+
"grad_norm": 107.78765869140625,
|
| 595 |
+
"learning_rate": 4.1407458326172884e-05,
|
| 596 |
+
"loss": 1.0826,
|
| 597 |
+
"step": 42000
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.5216900301966465,
|
| 601 |
+
"grad_norm": 8.029512405395508,
|
| 602 |
+
"learning_rate": 4.1305166163389226e-05,
|
| 603 |
+
"loss": 1.0809,
|
| 604 |
+
"step": 42500
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.5278275599636658,
|
| 608 |
+
"grad_norm": 44.91604232788086,
|
| 609 |
+
"learning_rate": 4.1202874000605575e-05,
|
| 610 |
+
"loss": 1.093,
|
| 611 |
+
"step": 43000
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.5339650897306852,
|
| 615 |
+
"grad_norm": 11.860628128051758,
|
| 616 |
+
"learning_rate": 4.110058183782192e-05,
|
| 617 |
+
"loss": 1.0709,
|
| 618 |
+
"step": 43500
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.5401026194977046,
|
| 622 |
+
"grad_norm": 30.72344398498535,
|
| 623 |
+
"learning_rate": 4.099828967503826e-05,
|
| 624 |
+
"loss": 1.0623,
|
| 625 |
+
"step": 44000
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.546240149264724,
|
| 629 |
+
"grad_norm": 19.160911560058594,
|
| 630 |
+
"learning_rate": 4.08959975122546e-05,
|
| 631 |
+
"loss": 1.0801,
|
| 632 |
+
"step": 44500
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.5523776790317433,
|
| 636 |
+
"grad_norm": 12.648110389709473,
|
| 637 |
+
"learning_rate": 4.079370534947094e-05,
|
| 638 |
+
"loss": 1.0833,
|
| 639 |
+
"step": 45000
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.5585152087987627,
|
| 643 |
+
"grad_norm": 25.488170623779297,
|
| 644 |
+
"learning_rate": 4.0691413186687285e-05,
|
| 645 |
+
"loss": 1.0816,
|
| 646 |
+
"step": 45500
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 0.564652738565782,
|
| 650 |
+
"grad_norm": 8.982975006103516,
|
| 651 |
+
"learning_rate": 4.0589121023903634e-05,
|
| 652 |
+
"loss": 1.0697,
|
| 653 |
+
"step": 46000
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 0.5707902683328014,
|
| 657 |
+
"grad_norm": 12.776642799377441,
|
| 658 |
+
"learning_rate": 4.048682886111998e-05,
|
| 659 |
+
"loss": 1.0864,
|
| 660 |
+
"step": 46500
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.5769277980998208,
|
| 664 |
+
"grad_norm": 70.14471435546875,
|
| 665 |
+
"learning_rate": 4.0384536698336325e-05,
|
| 666 |
+
"loss": 1.0744,
|
| 667 |
+
"step": 47000
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 0.5830653278668402,
|
| 671 |
+
"grad_norm": 26.72732162475586,
|
| 672 |
+
"learning_rate": 4.028224453555267e-05,
|
| 673 |
+
"loss": 1.0897,
|
| 674 |
+
"step": 47500
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 0.5892028576338595,
|
| 678 |
+
"grad_norm": 22.69961166381836,
|
| 679 |
+
"learning_rate": 4.017995237276901e-05,
|
| 680 |
+
"loss": 1.0727,
|
| 681 |
+
"step": 48000
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.5953403874008789,
|
| 685 |
+
"grad_norm": 6.76914119720459,
|
| 686 |
+
"learning_rate": 4.007766020998535e-05,
|
| 687 |
+
"loss": 1.0621,
|
| 688 |
+
"step": 48500
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.6014779171678982,
|
| 692 |
+
"grad_norm": 41.32302474975586,
|
| 693 |
+
"learning_rate": 3.99753680472017e-05,
|
| 694 |
+
"loss": 1.0582,
|
| 695 |
+
"step": 49000
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 0.6076154469349176,
|
| 699 |
+
"grad_norm": 47.626644134521484,
|
| 700 |
+
"learning_rate": 3.987307588441804e-05,
|
| 701 |
+
"loss": 1.0681,
|
| 702 |
+
"step": 49500
|
| 703 |
+
}
|
| 704 |
+
],
|
| 705 |
+
"logging_steps": 500,
|
| 706 |
+
"max_steps": 244398,
|
| 707 |
+
"num_input_tokens_seen": 0,
|
| 708 |
+
"num_train_epochs": 3,
|
| 709 |
+
"save_steps": 500,
|
| 710 |
+
"stateful_callbacks": {
|
| 711 |
+
"TrainerControl": {
|
| 712 |
+
"args": {
|
| 713 |
+
"should_epoch_stop": false,
|
| 714 |
+
"should_evaluate": false,
|
| 715 |
+
"should_log": false,
|
| 716 |
+
"should_save": true,
|
| 717 |
+
"should_training_stop": false
|
| 718 |
+
},
|
| 719 |
+
"attributes": {}
|
| 720 |
+
}
|
| 721 |
+
},
|
| 722 |
+
"total_flos": 0.0,
|
| 723 |
+
"train_batch_size": 8,
|
| 724 |
+
"trial_name": null,
|
| 725 |
+
"trial_params": null
|
| 726 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:320a99fb4b91f8525d40d71b170ac4e34508c4a26597a496c07d42c089137f1c
|
| 3 |
+
size 5432
|