Training in progress, step 1545, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 3664864144
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e735d6651ff979f84f7d64ca5c909cdd9d5d53bc7affb4b04db59bcdb9df0496
|
| 3 |
size 3664864144
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 7330126728
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:243c3f8ae1c3a08e0b668b5bd310c8b6c343d08cd9a9bd43d8d1adaa4d40aeb3
|
| 3 |
size 7330126728
|
last-checkpoint/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:b469b819f05d46728b32e996c035264ffbdf19b66cb9b8b924e9dfb2e5b598d6
|
| 3 |
size 14244
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3938e5834fb6d7b92270fc83a570825167a94c73c9876e613f5b81e1c8c2381e
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -10045,6 +10045,783 @@
|
|
| 10045 |
"learning_rate": 1.4489095429227995e-06,
|
| 10046 |
"loss": 0.0,
|
| 10047 |
"step": 1434
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 10048 |
}
|
| 10049 |
],
|
| 10050 |
"logging_steps": 1,
|
|
@@ -10059,12 +10836,12 @@
|
|
| 10059 |
"should_evaluate": false,
|
| 10060 |
"should_log": false,
|
| 10061 |
"should_save": true,
|
| 10062 |
-
"should_training_stop":
|
| 10063 |
},
|
| 10064 |
"attributes": {}
|
| 10065 |
}
|
| 10066 |
},
|
| 10067 |
-
"total_flos": 4.
|
| 10068 |
"train_batch_size": 4,
|
| 10069 |
"trial_name": null,
|
| 10070 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 1545,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 10045 |
"learning_rate": 1.4489095429227995e-06,
|
| 10046 |
"loss": 0.0,
|
| 10047 |
"step": 1434
|
| 10048 |
+
},
|
| 10049 |
+
{
|
| 10050 |
+
"epoch": 0.9288025889967637,
|
| 10051 |
+
"grad_norm": 3.283307160018012e-05,
|
| 10052 |
+
"learning_rate": 1.4230446531803e-06,
|
| 10053 |
+
"loss": 0.0,
|
| 10054 |
+
"step": 1435
|
| 10055 |
+
},
|
| 10056 |
+
{
|
| 10057 |
+
"epoch": 0.9294498381877022,
|
| 10058 |
+
"grad_norm": 2.6713743864092976e-05,
|
| 10059 |
+
"learning_rate": 1.3974093752209206e-06,
|
| 10060 |
+
"loss": 0.0,
|
| 10061 |
+
"step": 1436
|
| 10062 |
+
},
|
| 10063 |
+
{
|
| 10064 |
+
"epoch": 0.9300970873786408,
|
| 10065 |
+
"grad_norm": 2.7796757422038354e-05,
|
| 10066 |
+
"learning_rate": 1.372003830216545e-06,
|
| 10067 |
+
"loss": 0.0,
|
| 10068 |
+
"step": 1437
|
| 10069 |
+
},
|
| 10070 |
+
{
|
| 10071 |
+
"epoch": 0.9307443365695793,
|
| 10072 |
+
"grad_norm": 2.8533981094369665e-05,
|
| 10073 |
+
"learning_rate": 1.3468281382531866e-06,
|
| 10074 |
+
"loss": 0.0,
|
| 10075 |
+
"step": 1438
|
| 10076 |
+
},
|
| 10077 |
+
{
|
| 10078 |
+
"epoch": 0.9313915857605178,
|
| 10079 |
+
"grad_norm": 2.888152448576875e-05,
|
| 10080 |
+
"learning_rate": 1.321882418330389e-06,
|
| 10081 |
+
"loss": 0.0,
|
| 10082 |
+
"step": 1439
|
| 10083 |
+
},
|
| 10084 |
+
{
|
| 10085 |
+
"epoch": 0.9320388349514563,
|
| 10086 |
+
"grad_norm": 2.925360058725346e-05,
|
| 10087 |
+
"learning_rate": 1.2971667883606652e-06,
|
| 10088 |
+
"loss": 0.0,
|
| 10089 |
+
"step": 1440
|
| 10090 |
+
},
|
| 10091 |
+
{
|
| 10092 |
+
"epoch": 0.9326860841423948,
|
| 10093 |
+
"grad_norm": 2.8393138563842513e-05,
|
| 10094 |
+
"learning_rate": 1.272681365168965e-06,
|
| 10095 |
+
"loss": 0.0,
|
| 10096 |
+
"step": 1441
|
| 10097 |
+
},
|
| 10098 |
+
{
|
| 10099 |
+
"epoch": 0.9333333333333333,
|
| 10100 |
+
"grad_norm": 3.054711487493478e-05,
|
| 10101 |
+
"learning_rate": 1.2484262644920918e-06,
|
| 10102 |
+
"loss": 0.0,
|
| 10103 |
+
"step": 1442
|
| 10104 |
+
},
|
| 10105 |
+
{
|
| 10106 |
+
"epoch": 0.9339805825242719,
|
| 10107 |
+
"grad_norm": 2.7200458134757355e-05,
|
| 10108 |
+
"learning_rate": 1.2244016009781701e-06,
|
| 10109 |
+
"loss": 0.0,
|
| 10110 |
+
"step": 1443
|
| 10111 |
+
},
|
| 10112 |
+
{
|
| 10113 |
+
"epoch": 0.9346278317152104,
|
| 10114 |
+
"grad_norm": 2.731495442276355e-05,
|
| 10115 |
+
"learning_rate": 1.2006074881861063e-06,
|
| 10116 |
+
"loss": 0.0,
|
| 10117 |
+
"step": 1444
|
| 10118 |
+
},
|
| 10119 |
+
{
|
| 10120 |
+
"epoch": 0.9352750809061489,
|
| 10121 |
+
"grad_norm": 2.8398326321621425e-05,
|
| 10122 |
+
"learning_rate": 1.1770440385850401e-06,
|
| 10123 |
+
"loss": 0.0,
|
| 10124 |
+
"step": 1445
|
| 10125 |
+
},
|
| 10126 |
+
{
|
| 10127 |
+
"epoch": 0.9359223300970874,
|
| 10128 |
+
"grad_norm": 2.7919622880290262e-05,
|
| 10129 |
+
"learning_rate": 1.1537113635538332e-06,
|
| 10130 |
+
"loss": 0.0,
|
| 10131 |
+
"step": 1446
|
| 10132 |
+
},
|
| 10133 |
+
{
|
| 10134 |
+
"epoch": 0.9365695792880259,
|
| 10135 |
+
"grad_norm": 2.779790520435199e-05,
|
| 10136 |
+
"learning_rate": 1.1306095733805254e-06,
|
| 10137 |
+
"loss": 0.0,
|
| 10138 |
+
"step": 1447
|
| 10139 |
+
},
|
| 10140 |
+
{
|
| 10141 |
+
"epoch": 0.9372168284789644,
|
| 10142 |
+
"grad_norm": 2.8276941520744003e-05,
|
| 10143 |
+
"learning_rate": 1.1077387772618075e-06,
|
| 10144 |
+
"loss": 0.0,
|
| 10145 |
+
"step": 1448
|
| 10146 |
+
},
|
| 10147 |
+
{
|
| 10148 |
+
"epoch": 0.9378640776699029,
|
| 10149 |
+
"grad_norm": 2.8997719709877856e-05,
|
| 10150 |
+
"learning_rate": 1.0850990833025322e-06,
|
| 10151 |
+
"loss": 0.0,
|
| 10152 |
+
"step": 1449
|
| 10153 |
+
},
|
| 10154 |
+
{
|
| 10155 |
+
"epoch": 0.9385113268608414,
|
| 10156 |
+
"grad_norm": 2.779282112896908e-05,
|
| 10157 |
+
"learning_rate": 1.062690598515187e-06,
|
| 10158 |
+
"loss": 0.0,
|
| 10159 |
+
"step": 1450
|
| 10160 |
+
},
|
| 10161 |
+
{
|
| 10162 |
+
"epoch": 0.9391585760517799,
|
| 10163 |
+
"grad_norm": 3.428855779930018e-05,
|
| 10164 |
+
"learning_rate": 1.0405134288193674e-06,
|
| 10165 |
+
"loss": 0.0,
|
| 10166 |
+
"step": 1451
|
| 10167 |
+
},
|
| 10168 |
+
{
|
| 10169 |
+
"epoch": 0.9398058252427185,
|
| 10170 |
+
"grad_norm": 3.5010241845157e-05,
|
| 10171 |
+
"learning_rate": 1.0185676790413213e-06,
|
| 10172 |
+
"loss": 0.0,
|
| 10173 |
+
"step": 1452
|
| 10174 |
+
},
|
| 10175 |
+
{
|
| 10176 |
+
"epoch": 0.940453074433657,
|
| 10177 |
+
"grad_norm": 3.0191284167813137e-05,
|
| 10178 |
+
"learning_rate": 9.968534529134154e-07,
|
| 10179 |
+
"loss": 0.0,
|
| 10180 |
+
"step": 1453
|
| 10181 |
+
},
|
| 10182 |
+
{
|
| 10183 |
+
"epoch": 0.9411003236245955,
|
| 10184 |
+
"grad_norm": 3.0183244234649464e-05,
|
| 10185 |
+
"learning_rate": 9.75370853073665e-07,
|
| 10186 |
+
"loss": 0.0,
|
| 10187 |
+
"step": 1454
|
| 10188 |
+
},
|
| 10189 |
+
{
|
| 10190 |
+
"epoch": 0.941747572815534,
|
| 10191 |
+
"grad_norm": 2.900006620620843e-05,
|
| 10192 |
+
"learning_rate": 9.54119981065238e-07,
|
| 10193 |
+
"loss": 0.0,
|
| 10194 |
+
"step": 1455
|
| 10195 |
+
},
|
| 10196 |
+
{
|
| 10197 |
+
"epoch": 0.9423948220064725,
|
| 10198 |
+
"grad_norm": 4.836772131966427e-05,
|
| 10199 |
+
"learning_rate": 9.3310093733599e-07,
|
| 10200 |
+
"loss": 0.0,
|
| 10201 |
+
"step": 1456
|
| 10202 |
+
},
|
| 10203 |
+
{
|
| 10204 |
+
"epoch": 0.943042071197411,
|
| 10205 |
+
"grad_norm": 3.2600761187495664e-05,
|
| 10206 |
+
"learning_rate": 9.123138212379534e-07,
|
| 10207 |
+
"loss": 0.0,
|
| 10208 |
+
"step": 1457
|
| 10209 |
+
},
|
| 10210 |
+
{
|
| 10211 |
+
"epoch": 0.9436893203883495,
|
| 10212 |
+
"grad_norm": 2.8977845431654714e-05,
|
| 10213 |
+
"learning_rate": 8.917587310269315e-07,
|
| 10214 |
+
"loss": 0.0,
|
| 10215 |
+
"step": 1458
|
| 10216 |
+
},
|
| 10217 |
+
{
|
| 10218 |
+
"epoch": 0.944336569579288,
|
| 10219 |
+
"grad_norm": 3.258264041505754e-05,
|
| 10220 |
+
"learning_rate": 8.714357638619608e-07,
|
| 10221 |
+
"loss": 0.0,
|
| 10222 |
+
"step": 1459
|
| 10223 |
+
},
|
| 10224 |
+
{
|
| 10225 |
+
"epoch": 0.9449838187702265,
|
| 10226 |
+
"grad_norm": 3.353532883920707e-05,
|
| 10227 |
+
"learning_rate": 8.513450158049108e-07,
|
| 10228 |
+
"loss": 0.0,
|
| 10229 |
+
"step": 1460
|
| 10230 |
+
},
|
| 10231 |
+
{
|
| 10232 |
+
"epoch": 0.945631067961165,
|
| 10233 |
+
"grad_norm": 3.066585122724064e-05,
|
| 10234 |
+
"learning_rate": 8.314865818200013e-07,
|
| 10235 |
+
"loss": 0.0,
|
| 10236 |
+
"step": 1461
|
| 10237 |
+
},
|
| 10238 |
+
{
|
| 10239 |
+
"epoch": 0.9462783171521035,
|
| 10240 |
+
"grad_norm": 2.8273796488065273e-05,
|
| 10241 |
+
"learning_rate": 8.118605557733417e-07,
|
| 10242 |
+
"loss": 0.0,
|
| 10243 |
+
"step": 1462
|
| 10244 |
+
},
|
| 10245 |
+
{
|
| 10246 |
+
"epoch": 0.946925566343042,
|
| 10247 |
+
"grad_norm": 2.887805203499738e-05,
|
| 10248 |
+
"learning_rate": 7.924670304325199e-07,
|
| 10249 |
+
"loss": 0.0,
|
| 10250 |
+
"step": 1463
|
| 10251 |
+
},
|
| 10252 |
+
{
|
| 10253 |
+
"epoch": 0.9475728155339805,
|
| 10254 |
+
"grad_norm": 3.234121322748251e-05,
|
| 10255 |
+
"learning_rate": 7.733060974661588e-07,
|
| 10256 |
+
"loss": 0.0,
|
| 10257 |
+
"step": 1464
|
| 10258 |
+
},
|
| 10259 |
+
{
|
| 10260 |
+
"epoch": 0.948220064724919,
|
| 10261 |
+
"grad_norm": 2.8881553589599207e-05,
|
| 10262 |
+
"learning_rate": 7.543778474434438e-07,
|
| 10263 |
+
"loss": 0.0,
|
| 10264 |
+
"step": 1465
|
| 10265 |
+
},
|
| 10266 |
+
{
|
| 10267 |
+
"epoch": 0.9488673139158577,
|
| 10268 |
+
"grad_norm": 2.9947719667688943e-05,
|
| 10269 |
+
"learning_rate": 7.356823698337512e-07,
|
| 10270 |
+
"loss": 0.0,
|
| 10271 |
+
"step": 1466
|
| 10272 |
+
},
|
| 10273 |
+
{
|
| 10274 |
+
"epoch": 0.9495145631067962,
|
| 10275 |
+
"grad_norm": 3.2102212571771815e-05,
|
| 10276 |
+
"learning_rate": 7.172197530061708e-07,
|
| 10277 |
+
"loss": 0.0,
|
| 10278 |
+
"step": 1467
|
| 10279 |
+
},
|
| 10280 |
+
{
|
| 10281 |
+
"epoch": 0.9501618122977347,
|
| 10282 |
+
"grad_norm": 2.9960867323097773e-05,
|
| 10283 |
+
"learning_rate": 6.989900842291286e-07,
|
| 10284 |
+
"loss": 0.0,
|
| 10285 |
+
"step": 1468
|
| 10286 |
+
},
|
| 10287 |
+
{
|
| 10288 |
+
"epoch": 0.9508090614886732,
|
| 10289 |
+
"grad_norm": 2.9122256819391623e-05,
|
| 10290 |
+
"learning_rate": 6.809934496699588e-07,
|
| 10291 |
+
"loss": 0.0,
|
| 10292 |
+
"step": 1469
|
| 10293 |
+
},
|
| 10294 |
+
{
|
| 10295 |
+
"epoch": 0.9514563106796117,
|
| 10296 |
+
"grad_norm": 3.0438421163125895e-05,
|
| 10297 |
+
"learning_rate": 6.632299343945103e-07,
|
| 10298 |
+
"loss": 0.0,
|
| 10299 |
+
"step": 1470
|
| 10300 |
+
},
|
| 10301 |
+
{
|
| 10302 |
+
"epoch": 0.9521035598705502,
|
| 10303 |
+
"grad_norm": 3.2355859730159864e-05,
|
| 10304 |
+
"learning_rate": 6.456996223667022e-07,
|
| 10305 |
+
"loss": 0.0,
|
| 10306 |
+
"step": 1471
|
| 10307 |
+
},
|
| 10308 |
+
{
|
| 10309 |
+
"epoch": 0.9527508090614887,
|
| 10310 |
+
"grad_norm": 2.8882805054308847e-05,
|
| 10311 |
+
"learning_rate": 6.28402596448191e-07,
|
| 10312 |
+
"loss": 0.0,
|
| 10313 |
+
"step": 1472
|
| 10314 |
+
},
|
| 10315 |
+
{
|
| 10316 |
+
"epoch": 0.9533980582524272,
|
| 10317 |
+
"grad_norm": 2.912832860602066e-05,
|
| 10318 |
+
"learning_rate": 6.113389383979151e-07,
|
| 10319 |
+
"loss": 0.0,
|
| 10320 |
+
"step": 1473
|
| 10321 |
+
},
|
| 10322 |
+
{
|
| 10323 |
+
"epoch": 0.9540453074433657,
|
| 10324 |
+
"grad_norm": 2.939105615951121e-05,
|
| 10325 |
+
"learning_rate": 5.945087288717622e-07,
|
| 10326 |
+
"loss": 0.0,
|
| 10327 |
+
"step": 1474
|
| 10328 |
+
},
|
| 10329 |
+
{
|
| 10330 |
+
"epoch": 0.9546925566343042,
|
| 10331 |
+
"grad_norm": 3.0436318411375396e-05,
|
| 10332 |
+
"learning_rate": 5.779120474221522e-07,
|
| 10333 |
+
"loss": 0.0,
|
| 10334 |
+
"step": 1475
|
| 10335 |
+
},
|
| 10336 |
+
{
|
| 10337 |
+
"epoch": 0.9553398058252427,
|
| 10338 |
+
"grad_norm": 2.996389230247587e-05,
|
| 10339 |
+
"learning_rate": 5.615489724976664e-07,
|
| 10340 |
+
"loss": 0.0,
|
| 10341 |
+
"step": 1476
|
| 10342 |
+
},
|
| 10343 |
+
{
|
| 10344 |
+
"epoch": 0.9559870550161812,
|
| 10345 |
+
"grad_norm": 3.117434971500188e-05,
|
| 10346 |
+
"learning_rate": 5.454195814427021e-07,
|
| 10347 |
+
"loss": 0.0,
|
| 10348 |
+
"step": 1477
|
| 10349 |
+
},
|
| 10350 |
+
{
|
| 10351 |
+
"epoch": 0.9566343042071197,
|
| 10352 |
+
"grad_norm": 2.7828933525597677e-05,
|
| 10353 |
+
"learning_rate": 5.295239504970739e-07,
|
| 10354 |
+
"loss": 0.0,
|
| 10355 |
+
"step": 1478
|
| 10356 |
+
},
|
| 10357 |
+
{
|
| 10358 |
+
"epoch": 0.9572815533980582,
|
| 10359 |
+
"grad_norm": 2.7228026738157496e-05,
|
| 10360 |
+
"learning_rate": 5.138621547956635e-07,
|
| 10361 |
+
"loss": 0.0,
|
| 10362 |
+
"step": 1479
|
| 10363 |
+
},
|
| 10364 |
+
{
|
| 10365 |
+
"epoch": 0.9579288025889967,
|
| 10366 |
+
"grad_norm": 2.887405025830958e-05,
|
| 10367 |
+
"learning_rate": 4.984342683680809e-07,
|
| 10368 |
+
"loss": 0.0,
|
| 10369 |
+
"step": 1480
|
| 10370 |
+
},
|
| 10371 |
+
{
|
| 10372 |
+
"epoch": 0.9585760517799353,
|
| 10373 |
+
"grad_norm": 2.721715827647131e-05,
|
| 10374 |
+
"learning_rate": 4.832403641383044e-07,
|
| 10375 |
+
"loss": 0.0,
|
| 10376 |
+
"step": 1481
|
| 10377 |
+
},
|
| 10378 |
+
{
|
| 10379 |
+
"epoch": 0.9592233009708738,
|
| 10380 |
+
"grad_norm": 2.828839569701813e-05,
|
| 10381 |
+
"learning_rate": 4.6828051392431847e-07,
|
| 10382 |
+
"loss": 0.0,
|
| 10383 |
+
"step": 1482
|
| 10384 |
+
},
|
| 10385 |
+
{
|
| 10386 |
+
"epoch": 0.9598705501618123,
|
| 10387 |
+
"grad_norm": 2.8300539270276204e-05,
|
| 10388 |
+
"learning_rate": 4.535547884378044e-07,
|
| 10389 |
+
"loss": 0.0,
|
| 10390 |
+
"step": 1483
|
| 10391 |
+
},
|
| 10392 |
+
{
|
| 10393 |
+
"epoch": 0.9605177993527508,
|
| 10394 |
+
"grad_norm": 2.900650360970758e-05,
|
| 10395 |
+
"learning_rate": 4.390632572837783e-07,
|
| 10396 |
+
"loss": 0.0,
|
| 10397 |
+
"step": 1484
|
| 10398 |
+
},
|
| 10399 |
+
{
|
| 10400 |
+
"epoch": 0.9611650485436893,
|
| 10401 |
+
"grad_norm": 2.971368121507112e-05,
|
| 10402 |
+
"learning_rate": 4.2480598896028624e-07,
|
| 10403 |
+
"loss": 0.0,
|
| 10404 |
+
"step": 1485
|
| 10405 |
+
},
|
| 10406 |
+
{
|
| 10407 |
+
"epoch": 0.9618122977346278,
|
| 10408 |
+
"grad_norm": 3.18863385473378e-05,
|
| 10409 |
+
"learning_rate": 4.1078305085807124e-07,
|
| 10410 |
+
"loss": 0.0,
|
| 10411 |
+
"step": 1486
|
| 10412 |
+
},
|
| 10413 |
+
{
|
| 10414 |
+
"epoch": 0.9624595469255663,
|
| 10415 |
+
"grad_norm": 2.9233349778223783e-05,
|
| 10416 |
+
"learning_rate": 3.9699450926022896e-07,
|
| 10417 |
+
"loss": 0.0,
|
| 10418 |
+
"step": 1487
|
| 10419 |
+
},
|
| 10420 |
+
{
|
| 10421 |
+
"epoch": 0.9631067961165048,
|
| 10422 |
+
"grad_norm": 2.9725508284172975e-05,
|
| 10423 |
+
"learning_rate": 3.8344042934195246e-07,
|
| 10424 |
+
"loss": 0.0,
|
| 10425 |
+
"step": 1488
|
| 10426 |
+
},
|
| 10427 |
+
{
|
| 10428 |
+
"epoch": 0.9637540453074434,
|
| 10429 |
+
"grad_norm": 2.8876263968413696e-05,
|
| 10430 |
+
"learning_rate": 3.7012087517016567e-07,
|
| 10431 |
+
"loss": 0.0,
|
| 10432 |
+
"step": 1489
|
| 10433 |
+
},
|
| 10434 |
+
{
|
| 10435 |
+
"epoch": 0.9644012944983819,
|
| 10436 |
+
"grad_norm": 2.9116605219314806e-05,
|
| 10437 |
+
"learning_rate": 3.570359097032516e-07,
|
| 10438 |
+
"loss": 0.0,
|
| 10439 |
+
"step": 1490
|
| 10440 |
+
},
|
| 10441 |
+
{
|
| 10442 |
+
"epoch": 0.9650485436893204,
|
| 10443 |
+
"grad_norm": 2.7797896109404974e-05,
|
| 10444 |
+
"learning_rate": 3.441855947907524e-07,
|
| 10445 |
+
"loss": 0.0,
|
| 10446 |
+
"step": 1491
|
| 10447 |
+
},
|
| 10448 |
+
{
|
| 10449 |
+
"epoch": 0.9656957928802589,
|
| 10450 |
+
"grad_norm": 2.8875567295472138e-05,
|
| 10451 |
+
"learning_rate": 3.315699911730641e-07,
|
| 10452 |
+
"loss": 0.0,
|
| 10453 |
+
"step": 1492
|
| 10454 |
+
},
|
| 10455 |
+
{
|
| 10456 |
+
"epoch": 0.9663430420711974,
|
| 10457 |
+
"grad_norm": 2.9713237381656654e-05,
|
| 10458 |
+
"learning_rate": 3.1918915848115903e-07,
|
| 10459 |
+
"loss": 0.0,
|
| 10460 |
+
"step": 1493
|
| 10461 |
+
},
|
| 10462 |
+
{
|
| 10463 |
+
"epoch": 0.9669902912621359,
|
| 10464 |
+
"grad_norm": 2.8407523132045753e-05,
|
| 10465 |
+
"learning_rate": 3.0704315523631953e-07,
|
| 10466 |
+
"loss": 0.0,
|
| 10467 |
+
"step": 1494
|
| 10468 |
+
},
|
| 10469 |
+
{
|
| 10470 |
+
"epoch": 0.9676375404530745,
|
| 10471 |
+
"grad_norm": 2.7202309865970165e-05,
|
| 10472 |
+
"learning_rate": 2.9513203884981577e-07,
|
| 10473 |
+
"loss": 0.0,
|
| 10474 |
+
"step": 1495
|
| 10475 |
+
},
|
| 10476 |
+
{
|
| 10477 |
+
"epoch": 0.968284789644013,
|
| 10478 |
+
"grad_norm": 2.7205003789276816e-05,
|
| 10479 |
+
"learning_rate": 2.8345586562268934e-07,
|
| 10480 |
+
"loss": 0.0,
|
| 10481 |
+
"step": 1496
|
| 10482 |
+
},
|
| 10483 |
+
{
|
| 10484 |
+
"epoch": 0.9689320388349515,
|
| 10485 |
+
"grad_norm": 2.9845659810234793e-05,
|
| 10486 |
+
"learning_rate": 2.7201469074544795e-07,
|
| 10487 |
+
"loss": 0.0,
|
| 10488 |
+
"step": 1497
|
| 10489 |
+
},
|
| 10490 |
+
{
|
| 10491 |
+
"epoch": 0.96957928802589,
|
| 10492 |
+
"grad_norm": 2.86362374026794e-05,
|
| 10493 |
+
"learning_rate": 2.608085682978212e-07,
|
| 10494 |
+
"loss": 0.0,
|
| 10495 |
+
"step": 1498
|
| 10496 |
+
},
|
| 10497 |
+
{
|
| 10498 |
+
"epoch": 0.9702265372168285,
|
| 10499 |
+
"grad_norm": 2.9241513402666897e-05,
|
| 10500 |
+
"learning_rate": 2.498375512484941e-07,
|
| 10501 |
+
"loss": 0.0,
|
| 10502 |
+
"step": 1499
|
| 10503 |
+
},
|
| 10504 |
+
{
|
| 10505 |
+
"epoch": 0.970873786407767,
|
| 10506 |
+
"grad_norm": 3.067816942348145e-05,
|
| 10507 |
+
"learning_rate": 2.3910169145487936e-07,
|
| 10508 |
+
"loss": 0.0,
|
| 10509 |
+
"step": 1500
|
| 10510 |
+
},
|
| 10511 |
+
{
|
| 10512 |
+
"epoch": 0.9715210355987055,
|
| 10513 |
+
"grad_norm": 2.9163411454646848e-05,
|
| 10514 |
+
"learning_rate": 2.2860103966284019e-07,
|
| 10515 |
+
"loss": 0.0,
|
| 10516 |
+
"step": 1501
|
| 10517 |
+
},
|
| 10518 |
+
{
|
| 10519 |
+
"epoch": 0.972168284789644,
|
| 10520 |
+
"grad_norm": 3.45797925547231e-05,
|
| 10521 |
+
"learning_rate": 2.183356455064789e-07,
|
| 10522 |
+
"loss": 0.0,
|
| 10523 |
+
"step": 1502
|
| 10524 |
+
},
|
| 10525 |
+
{
|
| 10526 |
+
"epoch": 0.9728155339805825,
|
| 10527 |
+
"grad_norm": 2.8640923119382933e-05,
|
| 10528 |
+
"learning_rate": 2.0830555750788738e-07,
|
| 10529 |
+
"loss": 0.0,
|
| 10530 |
+
"step": 1503
|
| 10531 |
+
},
|
| 10532 |
+
{
|
| 10533 |
+
"epoch": 0.973462783171521,
|
| 10534 |
+
"grad_norm": 3.0909341148799285e-05,
|
| 10535 |
+
"learning_rate": 1.9851082307691948e-07,
|
| 10536 |
+
"loss": 0.0,
|
| 10537 |
+
"step": 1504
|
| 10538 |
+
},
|
| 10539 |
+
{
|
| 10540 |
+
"epoch": 0.9741100323624595,
|
| 10541 |
+
"grad_norm": 3.211084185750224e-05,
|
| 10542 |
+
"learning_rate": 1.889514885109689e-07,
|
| 10543 |
+
"loss": 0.0,
|
| 10544 |
+
"step": 1505
|
| 10545 |
+
},
|
| 10546 |
+
{
|
| 10547 |
+
"epoch": 0.974757281553398,
|
| 10548 |
+
"grad_norm": 2.8628259315155447e-05,
|
| 10549 |
+
"learning_rate": 1.7962759899474713e-07,
|
| 10550 |
+
"loss": 0.0,
|
| 10551 |
+
"step": 1506
|
| 10552 |
+
},
|
| 10553 |
+
{
|
| 10554 |
+
"epoch": 0.9754045307443365,
|
| 10555 |
+
"grad_norm": 2.9366896342253312e-05,
|
| 10556 |
+
"learning_rate": 1.7053919860007816e-07,
|
| 10557 |
+
"loss": 0.0,
|
| 10558 |
+
"step": 1507
|
| 10559 |
+
},
|
| 10560 |
+
{
|
| 10561 |
+
"epoch": 0.976051779935275,
|
| 10562 |
+
"grad_norm": 2.9947226721560583e-05,
|
| 10563 |
+
"learning_rate": 1.6168633028568747e-07,
|
| 10564 |
+
"loss": 0.0,
|
| 10565 |
+
"step": 1508
|
| 10566 |
+
},
|
| 10567 |
+
{
|
| 10568 |
+
"epoch": 0.9766990291262136,
|
| 10569 |
+
"grad_norm": 3.185419336659834e-05,
|
| 10570 |
+
"learning_rate": 1.5306903589698552e-07,
|
| 10571 |
+
"loss": 0.0,
|
| 10572 |
+
"step": 1509
|
| 10573 |
+
},
|
| 10574 |
+
{
|
| 10575 |
+
"epoch": 0.9773462783171522,
|
| 10576 |
+
"grad_norm": 3.955503780161962e-05,
|
| 10577 |
+
"learning_rate": 1.4468735616587904e-07,
|
| 10578 |
+
"loss": 0.0,
|
| 10579 |
+
"step": 1510
|
| 10580 |
+
},
|
| 10581 |
+
{
|
| 10582 |
+
"epoch": 0.9779935275080907,
|
| 10583 |
+
"grad_norm": 3.210457362001762e-05,
|
| 10584 |
+
"learning_rate": 1.3654133071059893e-07,
|
| 10585 |
+
"loss": 0.0,
|
| 10586 |
+
"step": 1511
|
| 10587 |
+
},
|
| 10588 |
+
{
|
| 10589 |
+
"epoch": 0.9786407766990292,
|
| 10590 |
+
"grad_norm": 3.211268631275743e-05,
|
| 10591 |
+
"learning_rate": 1.2863099803547274e-07,
|
| 10592 |
+
"loss": 0.0,
|
| 10593 |
+
"step": 1512
|
| 10594 |
+
},
|
| 10595 |
+
{
|
| 10596 |
+
"epoch": 0.9792880258899677,
|
| 10597 |
+
"grad_norm": 3.0189445169526152e-05,
|
| 10598 |
+
"learning_rate": 1.2095639553077466e-07,
|
| 10599 |
+
"loss": 0.0,
|
| 10600 |
+
"step": 1513
|
| 10601 |
+
},
|
| 10602 |
+
{
|
| 10603 |
+
"epoch": 0.9799352750809062,
|
| 10604 |
+
"grad_norm": 2.912877607741393e-05,
|
| 10605 |
+
"learning_rate": 1.1351755947253684e-07,
|
| 10606 |
+
"loss": 0.0,
|
| 10607 |
+
"step": 1514
|
| 10608 |
+
},
|
| 10609 |
+
{
|
| 10610 |
+
"epoch": 0.9805825242718447,
|
| 10611 |
+
"grad_norm": 2.8272341296542436e-05,
|
| 10612 |
+
"learning_rate": 1.0631452502237737e-07,
|
| 10613 |
+
"loss": 0.0,
|
| 10614 |
+
"step": 1515
|
| 10615 |
+
},
|
| 10616 |
+
{
|
| 10617 |
+
"epoch": 0.9812297734627832,
|
| 10618 |
+
"grad_norm": 2.7096859412267804e-05,
|
| 10619 |
+
"learning_rate": 9.934732622734477e-08,
|
| 10620 |
+
"loss": 0.0,
|
| 10621 |
+
"step": 1516
|
| 10622 |
+
},
|
| 10623 |
+
{
|
| 10624 |
+
"epoch": 0.9818770226537217,
|
| 10625 |
+
"grad_norm": 3.0457478715106845e-05,
|
| 10626 |
+
"learning_rate": 9.261599601972926e-08,
|
| 10627 |
+
"loss": 0.0,
|
| 10628 |
+
"step": 1517
|
| 10629 |
+
},
|
| 10630 |
+
{
|
| 10631 |
+
"epoch": 0.9825242718446602,
|
| 10632 |
+
"grad_norm": 2.9363456633291207e-05,
|
| 10633 |
+
"learning_rate": 8.612056621694064e-08,
|
| 10634 |
+
"loss": 0.0,
|
| 10635 |
+
"step": 1518
|
| 10636 |
+
},
|
| 10637 |
+
{
|
| 10638 |
+
"epoch": 0.9831715210355987,
|
| 10639 |
+
"grad_norm": 2.9489487133105285e-05,
|
| 10640 |
+
"learning_rate": 7.986106752134737e-08,
|
| 10641 |
+
"loss": 0.0,
|
| 10642 |
+
"step": 1519
|
| 10643 |
+
},
|
| 10644 |
+
{
|
| 10645 |
+
"epoch": 0.9838187702265372,
|
| 10646 |
+
"grad_norm": 3.068981095566414e-05,
|
| 10647 |
+
"learning_rate": 7.383752952010992e-08,
|
| 10648 |
+
"loss": 0.0,
|
| 10649 |
+
"step": 1520
|
| 10650 |
+
},
|
| 10651 |
+
{
|
| 10652 |
+
"epoch": 0.9844660194174757,
|
| 10653 |
+
"grad_norm": 2.996692455781158e-05,
|
| 10654 |
+
"learning_rate": 6.80499806850754e-08,
|
| 10655 |
+
"loss": 0.0,
|
| 10656 |
+
"step": 1521
|
| 10657 |
+
},
|
| 10658 |
+
{
|
| 10659 |
+
"epoch": 0.9851132686084142,
|
| 10660 |
+
"grad_norm": 3.164894587825984e-05,
|
| 10661 |
+
"learning_rate": 6.249844837261654e-08,
|
| 10662 |
+
"loss": 0.0,
|
| 10663 |
+
"step": 1522
|
| 10664 |
+
},
|
| 10665 |
+
{
|
| 10666 |
+
"epoch": 0.9857605177993527,
|
| 10667 |
+
"grad_norm": 2.961757309094537e-05,
|
| 10668 |
+
"learning_rate": 5.718295882350955e-08,
|
| 10669 |
+
"loss": 0.0,
|
| 10670 |
+
"step": 1523
|
| 10671 |
+
},
|
| 10672 |
+
{
|
| 10673 |
+
"epoch": 0.9864077669902913,
|
| 10674 |
+
"grad_norm": 3.163444489473477e-05,
|
| 10675 |
+
"learning_rate": 5.2103537162817576e-08,
|
| 10676 |
+
"loss": 0.0,
|
| 10677 |
+
"step": 1524
|
| 10678 |
+
},
|
| 10679 |
+
{
|
| 10680 |
+
"epoch": 0.9870550161812298,
|
| 10681 |
+
"grad_norm": 2.9721826649620198e-05,
|
| 10682 |
+
"learning_rate": 4.7260207399774105e-08,
|
| 10683 |
+
"loss": 0.0,
|
| 10684 |
+
"step": 1525
|
| 10685 |
+
},
|
| 10686 |
+
{
|
| 10687 |
+
"epoch": 0.9877022653721683,
|
| 10688 |
+
"grad_norm": 3.056014247704297e-05,
|
| 10689 |
+
"learning_rate": 4.265299242764975e-08,
|
| 10690 |
+
"loss": 0.0,
|
| 10691 |
+
"step": 1526
|
| 10692 |
+
},
|
| 10693 |
+
{
|
| 10694 |
+
"epoch": 0.9883495145631068,
|
| 10695 |
+
"grad_norm": 3.068634759983979e-05,
|
| 10696 |
+
"learning_rate": 3.8281914023657886e-08,
|
| 10697 |
+
"loss": 0.0,
|
| 10698 |
+
"step": 1527
|
| 10699 |
+
},
|
| 10700 |
+
{
|
| 10701 |
+
"epoch": 0.9889967637540453,
|
| 10702 |
+
"grad_norm": 2.780274735414423e-05,
|
| 10703 |
+
"learning_rate": 3.4146992848854695e-08,
|
| 10704 |
+
"loss": 0.0,
|
| 10705 |
+
"step": 1528
|
| 10706 |
+
},
|
| 10707 |
+
{
|
| 10708 |
+
"epoch": 0.9896440129449838,
|
| 10709 |
+
"grad_norm": 2.7697720724972896e-05,
|
| 10710 |
+
"learning_rate": 3.0248248448033757e-08,
|
| 10711 |
+
"loss": 0.0,
|
| 10712 |
+
"step": 1529
|
| 10713 |
+
},
|
| 10714 |
+
{
|
| 10715 |
+
"epoch": 0.9902912621359223,
|
| 10716 |
+
"grad_norm": 3.008269595738966e-05,
|
| 10717 |
+
"learning_rate": 2.6585699249642716e-08,
|
| 10718 |
+
"loss": 0.0,
|
| 10719 |
+
"step": 1530
|
| 10720 |
+
},
|
| 10721 |
+
{
|
| 10722 |
+
"epoch": 0.9909385113268608,
|
| 10723 |
+
"grad_norm": 2.8287571694818325e-05,
|
| 10724 |
+
"learning_rate": 2.3159362565677857e-08,
|
| 10725 |
+
"loss": 0.0,
|
| 10726 |
+
"step": 1531
|
| 10727 |
+
},
|
| 10728 |
+
{
|
| 10729 |
+
"epoch": 0.9915857605177993,
|
| 10730 |
+
"grad_norm": 2.7803509510704316e-05,
|
| 10731 |
+
"learning_rate": 1.996925459162857e-08,
|
| 10732 |
+
"loss": 0.0,
|
| 10733 |
+
"step": 1532
|
| 10734 |
+
},
|
| 10735 |
+
{
|
| 10736 |
+
"epoch": 0.9922330097087378,
|
| 10737 |
+
"grad_norm": 2.8281427148613147e-05,
|
| 10738 |
+
"learning_rate": 1.7015390406377453e-08,
|
| 10739 |
+
"loss": 0.0,
|
| 10740 |
+
"step": 1533
|
| 10741 |
+
},
|
| 10742 |
+
{
|
| 10743 |
+
"epoch": 0.9928802588996763,
|
| 10744 |
+
"grad_norm": 3.0304252504720353e-05,
|
| 10745 |
+
"learning_rate": 1.4297783972144763e-08,
|
| 10746 |
+
"loss": 0.0,
|
| 10747 |
+
"step": 1534
|
| 10748 |
+
},
|
| 10749 |
+
{
|
| 10750 |
+
"epoch": 0.9935275080906149,
|
| 10751 |
+
"grad_norm": 2.8403026590240188e-05,
|
| 10752 |
+
"learning_rate": 1.181644813441074e-08,
|
| 10753 |
+
"loss": 0.0,
|
| 10754 |
+
"step": 1535
|
| 10755 |
+
},
|
| 10756 |
+
{
|
| 10757 |
+
"epoch": 0.9941747572815534,
|
| 10758 |
+
"grad_norm": 3.0313709430629387e-05,
|
| 10759 |
+
"learning_rate": 9.571394621865626e-09,
|
| 10760 |
+
"loss": 0.0,
|
| 10761 |
+
"step": 1536
|
| 10762 |
+
},
|
| 10763 |
+
{
|
| 10764 |
+
"epoch": 0.9948220064724919,
|
| 10765 |
+
"grad_norm": 2.7807334845419973e-05,
|
| 10766 |
+
"learning_rate": 7.562634046348604e-09,
|
| 10767 |
+
"loss": 0.0,
|
| 10768 |
+
"step": 1537
|
| 10769 |
+
},
|
| 10770 |
+
{
|
| 10771 |
+
"epoch": 0.9954692556634305,
|
| 10772 |
+
"grad_norm": 2.8401254894561134e-05,
|
| 10773 |
+
"learning_rate": 5.790175902786743e-09,
|
| 10774 |
+
"loss": 0.0,
|
| 10775 |
+
"step": 1538
|
| 10776 |
+
},
|
| 10777 |
+
{
|
| 10778 |
+
"epoch": 0.996116504854369,
|
| 10779 |
+
"grad_norm": 2.612316529848613e-05,
|
| 10780 |
+
"learning_rate": 4.254028569183888e-09,
|
| 10781 |
+
"loss": 0.0,
|
| 10782 |
+
"step": 1539
|
| 10783 |
+
},
|
| 10784 |
+
{
|
| 10785 |
+
"epoch": 0.9967637540453075,
|
| 10786 |
+
"grad_norm": 2.9241215088404715e-05,
|
| 10787 |
+
"learning_rate": 2.9541993065373976e-09,
|
| 10788 |
+
"loss": 0.0,
|
| 10789 |
+
"step": 1540
|
| 10790 |
+
},
|
| 10791 |
+
{
|
| 10792 |
+
"epoch": 0.997411003236246,
|
| 10793 |
+
"grad_norm": 2.732668144744821e-05,
|
| 10794 |
+
"learning_rate": 1.8906942588325927e-09,
|
| 10795 |
+
"loss": 0.0,
|
| 10796 |
+
"step": 1541
|
| 10797 |
+
},
|
| 10798 |
+
{
|
| 10799 |
+
"epoch": 0.9980582524271845,
|
| 10800 |
+
"grad_norm": 2.7786840291810222e-05,
|
| 10801 |
+
"learning_rate": 1.063518453009449e-09,
|
| 10802 |
+
"loss": 0.0,
|
| 10803 |
+
"step": 1542
|
| 10804 |
+
},
|
| 10805 |
+
{
|
| 10806 |
+
"epoch": 0.998705501618123,
|
| 10807 |
+
"grad_norm": 2.719917392823845e-05,
|
| 10808 |
+
"learning_rate": 4.726757989348407e-10,
|
| 10809 |
+
"loss": 0.0,
|
| 10810 |
+
"step": 1543
|
| 10811 |
+
},
|
| 10812 |
+
{
|
| 10813 |
+
"epoch": 0.9993527508090615,
|
| 10814 |
+
"grad_norm": 2.7324498660163954e-05,
|
| 10815 |
+
"learning_rate": 1.1816908937478664e-10,
|
| 10816 |
+
"loss": 0.0,
|
| 10817 |
+
"step": 1544
|
| 10818 |
+
},
|
| 10819 |
+
{
|
| 10820 |
+
"epoch": 1.0,
|
| 10821 |
+
"grad_norm": 2.888041126425378e-05,
|
| 10822 |
+
"learning_rate": 0.0,
|
| 10823 |
+
"loss": 0.0,
|
| 10824 |
+
"step": 1545
|
| 10825 |
}
|
| 10826 |
],
|
| 10827 |
"logging_steps": 1,
|
|
|
|
| 10836 |
"should_evaluate": false,
|
| 10837 |
"should_log": false,
|
| 10838 |
"should_save": true,
|
| 10839 |
+
"should_training_stop": true
|
| 10840 |
},
|
| 10841 |
"attributes": {}
|
| 10842 |
}
|
| 10843 |
},
|
| 10844 |
+
"total_flos": 4.528157851813675e+18,
|
| 10845 |
"train_batch_size": 4,
|
| 10846 |
"trial_name": null,
|
| 10847 |
"trial_params": null
|