Training in progress, step 937, 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 4102239984
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:0ec5d4e85b4b2b01c9ffcd604688835ae1a7f5fa20569b216a0e590222072e86
|
3 |
size 4102239984
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 8204830696
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:864849a05c066555e793597b6fab85cc6da3e2acf75b0179a472d8000c4b1b7c
|
3 |
size 8204830696
|
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:d8b60c837d596a3ced4d864bdd9ac9e022e64500c8a0c3de7b1a5515bed7f61b
|
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 sha256:2982127b3bd380c10a83e88ec6cdb13087dd3bcf10319900ecad520c33cb08c2
|
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,
|
@@ -5026,6 +5026,1546 @@
|
|
5026 |
"learning_rate": 1.6099602899673083e-05,
|
5027 |
"loss": 0.0007,
|
5028 |
"step": 717
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
5029 |
}
|
5030 |
],
|
5031 |
"logging_steps": 1,
|
@@ -5040,12 +6580,12 @@
|
|
5040 |
"should_evaluate": false,
|
5041 |
"should_log": false,
|
5042 |
"should_save": true,
|
5043 |
-
"should_training_stop":
|
5044 |
},
|
5045 |
"attributes": {}
|
5046 |
}
|
5047 |
},
|
5048 |
-
"total_flos":
|
5049 |
"train_batch_size": 4,
|
5050 |
"trial_name": null,
|
5051 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.9992002132764596,
|
5 |
"eval_steps": 500,
|
6 |
+
"global_step": 937,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
5026 |
"learning_rate": 1.6099602899673083e-05,
|
5027 |
"loss": 0.0007,
|
5028 |
"step": 717
|
5029 |
+
},
|
5030 |
+
{
|
5031 |
+
"epoch": 0.765662490002666,
|
5032 |
+
"grad_norm": 0.012030128389596939,
|
5033 |
+
"learning_rate": 1.5961894463176942e-05,
|
5034 |
+
"loss": 0.0003,
|
5035 |
+
"step": 718
|
5036 |
+
},
|
5037 |
+
{
|
5038 |
+
"epoch": 0.7667288723007198,
|
5039 |
+
"grad_norm": 0.19891726970672607,
|
5040 |
+
"learning_rate": 1.582466555438158e-05,
|
5041 |
+
"loss": 0.002,
|
5042 |
+
"step": 719
|
5043 |
+
},
|
5044 |
+
{
|
5045 |
+
"epoch": 0.7677952545987736,
|
5046 |
+
"grad_norm": 0.022642523050308228,
|
5047 |
+
"learning_rate": 1.5687918106563326e-05,
|
5048 |
+
"loss": 0.0004,
|
5049 |
+
"step": 720
|
5050 |
+
},
|
5051 |
+
{
|
5052 |
+
"epoch": 0.7688616368968275,
|
5053 |
+
"grad_norm": 0.03008243627846241,
|
5054 |
+
"learning_rate": 1.555165404621567e-05,
|
5055 |
+
"loss": 0.0007,
|
5056 |
+
"step": 721
|
5057 |
+
},
|
5058 |
+
{
|
5059 |
+
"epoch": 0.7699280191948814,
|
5060 |
+
"grad_norm": 0.018216097727417946,
|
5061 |
+
"learning_rate": 1.5415875293022182e-05,
|
5062 |
+
"loss": 0.0005,
|
5063 |
+
"step": 722
|
5064 |
+
},
|
5065 |
+
{
|
5066 |
+
"epoch": 0.7709944014929352,
|
5067 |
+
"grad_norm": 0.0316886305809021,
|
5068 |
+
"learning_rate": 1.5280583759829436e-05,
|
5069 |
+
"loss": 0.0005,
|
5070 |
+
"step": 723
|
5071 |
+
},
|
5072 |
+
{
|
5073 |
+
"epoch": 0.7720607837909891,
|
5074 |
+
"grad_norm": 0.006309879012405872,
|
5075 |
+
"learning_rate": 1.5145781352620053e-05,
|
5076 |
+
"loss": 0.0002,
|
5077 |
+
"step": 724
|
5078 |
+
},
|
5079 |
+
{
|
5080 |
+
"epoch": 0.773127166089043,
|
5081 |
+
"grad_norm": 0.024186374619603157,
|
5082 |
+
"learning_rate": 1.5011469970485853e-05,
|
5083 |
+
"loss": 0.0005,
|
5084 |
+
"step": 725
|
5085 |
+
},
|
5086 |
+
{
|
5087 |
+
"epoch": 0.7741935483870968,
|
5088 |
+
"grad_norm": 0.03155244141817093,
|
5089 |
+
"learning_rate": 1.4877651505601158e-05,
|
5090 |
+
"loss": 0.0007,
|
5091 |
+
"step": 726
|
5092 |
+
},
|
5093 |
+
{
|
5094 |
+
"epoch": 0.7752599306851506,
|
5095 |
+
"grad_norm": 0.005562330596148968,
|
5096 |
+
"learning_rate": 1.4744327843196043e-05,
|
5097 |
+
"loss": 0.0002,
|
5098 |
+
"step": 727
|
5099 |
+
},
|
5100 |
+
{
|
5101 |
+
"epoch": 0.7763263129832044,
|
5102 |
+
"grad_norm": 0.03465036302804947,
|
5103 |
+
"learning_rate": 1.4611500861529825e-05,
|
5104 |
+
"loss": 0.0006,
|
5105 |
+
"step": 728
|
5106 |
+
},
|
5107 |
+
{
|
5108 |
+
"epoch": 0.7773926952812583,
|
5109 |
+
"grad_norm": 0.009592439979314804,
|
5110 |
+
"learning_rate": 1.4479172431864646e-05,
|
5111 |
+
"loss": 0.0002,
|
5112 |
+
"step": 729
|
5113 |
+
},
|
5114 |
+
{
|
5115 |
+
"epoch": 0.7784590775793122,
|
5116 |
+
"grad_norm": 0.01366474200040102,
|
5117 |
+
"learning_rate": 1.434734441843899e-05,
|
5118 |
+
"loss": 0.0003,
|
5119 |
+
"step": 730
|
5120 |
+
},
|
5121 |
+
{
|
5122 |
+
"epoch": 0.779525459877366,
|
5123 |
+
"grad_norm": 0.019273320212960243,
|
5124 |
+
"learning_rate": 1.4216018678441556e-05,
|
5125 |
+
"loss": 0.0004,
|
5126 |
+
"step": 731
|
5127 |
+
},
|
5128 |
+
{
|
5129 |
+
"epoch": 0.7805918421754199,
|
5130 |
+
"grad_norm": 0.007322750985622406,
|
5131 |
+
"learning_rate": 1.4085197061985023e-05,
|
5132 |
+
"loss": 0.0002,
|
5133 |
+
"step": 732
|
5134 |
+
},
|
5135 |
+
{
|
5136 |
+
"epoch": 0.7816582244734738,
|
5137 |
+
"grad_norm": 0.06155271455645561,
|
5138 |
+
"learning_rate": 1.3954881412079945e-05,
|
5139 |
+
"loss": 0.0011,
|
5140 |
+
"step": 733
|
5141 |
+
},
|
5142 |
+
{
|
5143 |
+
"epoch": 0.7827246067715276,
|
5144 |
+
"grad_norm": 0.00942557118833065,
|
5145 |
+
"learning_rate": 1.3825073564608909e-05,
|
5146 |
+
"loss": 0.0002,
|
5147 |
+
"step": 734
|
5148 |
+
},
|
5149 |
+
{
|
5150 |
+
"epoch": 0.7837909890695814,
|
5151 |
+
"grad_norm": 0.01583426259458065,
|
5152 |
+
"learning_rate": 1.3695775348300533e-05,
|
5153 |
+
"loss": 0.0003,
|
5154 |
+
"step": 735
|
5155 |
+
},
|
5156 |
+
{
|
5157 |
+
"epoch": 0.7848573713676353,
|
5158 |
+
"grad_norm": 0.4624207615852356,
|
5159 |
+
"learning_rate": 1.3566988584703816e-05,
|
5160 |
+
"loss": 0.0126,
|
5161 |
+
"step": 736
|
5162 |
+
},
|
5163 |
+
{
|
5164 |
+
"epoch": 0.7859237536656891,
|
5165 |
+
"grad_norm": 0.37781426310539246,
|
5166 |
+
"learning_rate": 1.3438715088162401e-05,
|
5167 |
+
"loss": 0.0168,
|
5168 |
+
"step": 737
|
5169 |
+
},
|
5170 |
+
{
|
5171 |
+
"epoch": 0.786990135963743,
|
5172 |
+
"grad_norm": 0.010702108964323997,
|
5173 |
+
"learning_rate": 1.3310956665789026e-05,
|
5174 |
+
"loss": 0.0002,
|
5175 |
+
"step": 738
|
5176 |
+
},
|
5177 |
+
{
|
5178 |
+
"epoch": 0.7880565182617969,
|
5179 |
+
"grad_norm": 0.40751180052757263,
|
5180 |
+
"learning_rate": 1.3183715117440142e-05,
|
5181 |
+
"loss": 0.0021,
|
5182 |
+
"step": 739
|
5183 |
+
},
|
5184 |
+
{
|
5185 |
+
"epoch": 0.7891229005598507,
|
5186 |
+
"grad_norm": 0.09710689634084702,
|
5187 |
+
"learning_rate": 1.305699223569043e-05,
|
5188 |
+
"loss": 0.0015,
|
5189 |
+
"step": 740
|
5190 |
+
},
|
5191 |
+
{
|
5192 |
+
"epoch": 0.7901892828579046,
|
5193 |
+
"grad_norm": 0.004762916825711727,
|
5194 |
+
"learning_rate": 1.2930789805807659e-05,
|
5195 |
+
"loss": 0.0002,
|
5196 |
+
"step": 741
|
5197 |
+
},
|
5198 |
+
{
|
5199 |
+
"epoch": 0.7912556651559585,
|
5200 |
+
"grad_norm": 0.019869035109877586,
|
5201 |
+
"learning_rate": 1.280510960572745e-05,
|
5202 |
+
"loss": 0.0004,
|
5203 |
+
"step": 742
|
5204 |
+
},
|
5205 |
+
{
|
5206 |
+
"epoch": 0.7923220474540122,
|
5207 |
+
"grad_norm": 0.3784305155277252,
|
5208 |
+
"learning_rate": 1.2679953406028317e-05,
|
5209 |
+
"loss": 0.0357,
|
5210 |
+
"step": 743
|
5211 |
+
},
|
5212 |
+
{
|
5213 |
+
"epoch": 0.7933884297520661,
|
5214 |
+
"grad_norm": 0.010028552263975143,
|
5215 |
+
"learning_rate": 1.2555322969906618e-05,
|
5216 |
+
"loss": 0.0003,
|
5217 |
+
"step": 744
|
5218 |
+
},
|
5219 |
+
{
|
5220 |
+
"epoch": 0.7944548120501199,
|
5221 |
+
"grad_norm": 0.14310139417648315,
|
5222 |
+
"learning_rate": 1.2431220053151833e-05,
|
5223 |
+
"loss": 0.0012,
|
5224 |
+
"step": 745
|
5225 |
+
},
|
5226 |
+
{
|
5227 |
+
"epoch": 0.7955211943481738,
|
5228 |
+
"grad_norm": 0.010279890149831772,
|
5229 |
+
"learning_rate": 1.2307646404121692e-05,
|
5230 |
+
"loss": 0.0003,
|
5231 |
+
"step": 746
|
5232 |
+
},
|
5233 |
+
{
|
5234 |
+
"epoch": 0.7965875766462277,
|
5235 |
+
"grad_norm": 0.010400409810245037,
|
5236 |
+
"learning_rate": 1.2184603763717684e-05,
|
5237 |
+
"loss": 0.0003,
|
5238 |
+
"step": 747
|
5239 |
+
},
|
5240 |
+
{
|
5241 |
+
"epoch": 0.7976539589442815,
|
5242 |
+
"grad_norm": 0.19656755030155182,
|
5243 |
+
"learning_rate": 1.2062093865360458e-05,
|
5244 |
+
"loss": 0.0027,
|
5245 |
+
"step": 748
|
5246 |
+
},
|
5247 |
+
{
|
5248 |
+
"epoch": 0.7987203412423354,
|
5249 |
+
"grad_norm": 0.025430811569094658,
|
5250 |
+
"learning_rate": 1.194011843496537e-05,
|
5251 |
+
"loss": 0.0006,
|
5252 |
+
"step": 749
|
5253 |
+
},
|
5254 |
+
{
|
5255 |
+
"epoch": 0.7997867235403893,
|
5256 |
+
"grad_norm": 0.018771683797240257,
|
5257 |
+
"learning_rate": 1.1818679190918247e-05,
|
5258 |
+
"loss": 0.0005,
|
5259 |
+
"step": 750
|
5260 |
+
},
|
5261 |
+
{
|
5262 |
+
"epoch": 0.800853105838443,
|
5263 |
+
"grad_norm": 0.036095619201660156,
|
5264 |
+
"learning_rate": 1.1697777844051105e-05,
|
5265 |
+
"loss": 0.0005,
|
5266 |
+
"step": 751
|
5267 |
+
},
|
5268 |
+
{
|
5269 |
+
"epoch": 0.8019194881364969,
|
5270 |
+
"grad_norm": 0.010772421956062317,
|
5271 |
+
"learning_rate": 1.1577416097618137e-05,
|
5272 |
+
"loss": 0.0003,
|
5273 |
+
"step": 752
|
5274 |
+
},
|
5275 |
+
{
|
5276 |
+
"epoch": 0.8029858704345508,
|
5277 |
+
"grad_norm": 0.0590975359082222,
|
5278 |
+
"learning_rate": 1.1457595647271602e-05,
|
5279 |
+
"loss": 0.0011,
|
5280 |
+
"step": 753
|
5281 |
+
},
|
5282 |
+
{
|
5283 |
+
"epoch": 0.8040522527326046,
|
5284 |
+
"grad_norm": 0.004293241072446108,
|
5285 |
+
"learning_rate": 1.1338318181038038e-05,
|
5286 |
+
"loss": 0.0001,
|
5287 |
+
"step": 754
|
5288 |
+
},
|
5289 |
+
{
|
5290 |
+
"epoch": 0.8051186350306585,
|
5291 |
+
"grad_norm": 0.021963993087410927,
|
5292 |
+
"learning_rate": 1.1219585379294444e-05,
|
5293 |
+
"loss": 0.0006,
|
5294 |
+
"step": 755
|
5295 |
+
},
|
5296 |
+
{
|
5297 |
+
"epoch": 0.8061850173287124,
|
5298 |
+
"grad_norm": 0.0076378388330340385,
|
5299 |
+
"learning_rate": 1.1101398914744566e-05,
|
5300 |
+
"loss": 0.0002,
|
5301 |
+
"step": 756
|
5302 |
+
},
|
5303 |
+
{
|
5304 |
+
"epoch": 0.8072513996267662,
|
5305 |
+
"grad_norm": 0.2593614459037781,
|
5306 |
+
"learning_rate": 1.0983760452395413e-05,
|
5307 |
+
"loss": 0.0164,
|
5308 |
+
"step": 757
|
5309 |
+
},
|
5310 |
+
{
|
5311 |
+
"epoch": 0.8083177819248201,
|
5312 |
+
"grad_norm": 0.15556533634662628,
|
5313 |
+
"learning_rate": 1.0866671649533722e-05,
|
5314 |
+
"loss": 0.0055,
|
5315 |
+
"step": 758
|
5316 |
+
},
|
5317 |
+
{
|
5318 |
+
"epoch": 0.8093841642228738,
|
5319 |
+
"grad_norm": 0.1670495867729187,
|
5320 |
+
"learning_rate": 1.0750134155702674e-05,
|
5321 |
+
"loss": 0.001,
|
5322 |
+
"step": 759
|
5323 |
+
},
|
5324 |
+
{
|
5325 |
+
"epoch": 0.8104505465209277,
|
5326 |
+
"grad_norm": 0.02173975668847561,
|
5327 |
+
"learning_rate": 1.063414961267859e-05,
|
5328 |
+
"loss": 0.0005,
|
5329 |
+
"step": 760
|
5330 |
+
},
|
5331 |
+
{
|
5332 |
+
"epoch": 0.8115169288189816,
|
5333 |
+
"grad_norm": 0.048297807574272156,
|
5334 |
+
"learning_rate": 1.0518719654447894e-05,
|
5335 |
+
"loss": 0.0006,
|
5336 |
+
"step": 761
|
5337 |
+
},
|
5338 |
+
{
|
5339 |
+
"epoch": 0.8125833111170354,
|
5340 |
+
"grad_norm": 0.031382959336042404,
|
5341 |
+
"learning_rate": 1.040384590718399e-05,
|
5342 |
+
"loss": 0.0004,
|
5343 |
+
"step": 762
|
5344 |
+
},
|
5345 |
+
{
|
5346 |
+
"epoch": 0.8136496934150893,
|
5347 |
+
"grad_norm": 0.006396553944796324,
|
5348 |
+
"learning_rate": 1.02895299892244e-05,
|
5349 |
+
"loss": 0.0002,
|
5350 |
+
"step": 763
|
5351 |
+
},
|
5352 |
+
{
|
5353 |
+
"epoch": 0.8147160757131432,
|
5354 |
+
"grad_norm": 0.007619079668074846,
|
5355 |
+
"learning_rate": 1.017577351104801e-05,
|
5356 |
+
"loss": 0.0002,
|
5357 |
+
"step": 764
|
5358 |
+
},
|
5359 |
+
{
|
5360 |
+
"epoch": 0.815782458011197,
|
5361 |
+
"grad_norm": 0.019957121461629868,
|
5362 |
+
"learning_rate": 1.0062578075252283e-05,
|
5363 |
+
"loss": 0.0005,
|
5364 |
+
"step": 765
|
5365 |
+
},
|
5366 |
+
{
|
5367 |
+
"epoch": 0.8168488403092509,
|
5368 |
+
"grad_norm": 0.011204512789845467,
|
5369 |
+
"learning_rate": 9.949945276530781e-06,
|
5370 |
+
"loss": 0.0002,
|
5371 |
+
"step": 766
|
5372 |
+
},
|
5373 |
+
{
|
5374 |
+
"epoch": 0.8179152226073048,
|
5375 |
+
"grad_norm": 0.008724085055291653,
|
5376 |
+
"learning_rate": 9.837876701650606e-06,
|
5377 |
+
"loss": 0.0002,
|
5378 |
+
"step": 767
|
5379 |
+
},
|
5380 |
+
{
|
5381 |
+
"epoch": 0.8189816049053585,
|
5382 |
+
"grad_norm": 0.08012081682682037,
|
5383 |
+
"learning_rate": 9.726373929430154e-06,
|
5384 |
+
"loss": 0.001,
|
5385 |
+
"step": 768
|
5386 |
+
},
|
5387 |
+
{
|
5388 |
+
"epoch": 0.8200479872034124,
|
5389 |
+
"grad_norm": 0.0606585331261158,
|
5390 |
+
"learning_rate": 9.615438530716753e-06,
|
5391 |
+
"loss": 0.0008,
|
5392 |
+
"step": 769
|
5393 |
+
},
|
5394 |
+
{
|
5395 |
+
"epoch": 0.8211143695014663,
|
5396 |
+
"grad_norm": 0.1962718665599823,
|
5397 |
+
"learning_rate": 9.505072068364624e-06,
|
5398 |
+
"loss": 0.0025,
|
5399 |
+
"step": 770
|
5400 |
+
},
|
5401 |
+
{
|
5402 |
+
"epoch": 0.8221807517995201,
|
5403 |
+
"grad_norm": 0.05189717561006546,
|
5404 |
+
"learning_rate": 9.39527609721284e-06,
|
5405 |
+
"loss": 0.0007,
|
5406 |
+
"step": 771
|
5407 |
+
},
|
5408 |
+
{
|
5409 |
+
"epoch": 0.823247134097574,
|
5410 |
+
"grad_norm": 0.013195768930017948,
|
5411 |
+
"learning_rate": 9.28605216406337e-06,
|
5412 |
+
"loss": 0.0003,
|
5413 |
+
"step": 772
|
5414 |
+
},
|
5415 |
+
{
|
5416 |
+
"epoch": 0.8243135163956279,
|
5417 |
+
"grad_norm": 0.015235373750329018,
|
5418 |
+
"learning_rate": 9.177401807659385e-06,
|
5419 |
+
"loss": 0.0003,
|
5420 |
+
"step": 773
|
5421 |
+
},
|
5422 |
+
{
|
5423 |
+
"epoch": 0.8253798986936817,
|
5424 |
+
"grad_norm": 0.005428910255432129,
|
5425 |
+
"learning_rate": 9.069326558663488e-06,
|
5426 |
+
"loss": 0.0002,
|
5427 |
+
"step": 774
|
5428 |
+
},
|
5429 |
+
{
|
5430 |
+
"epoch": 0.8264462809917356,
|
5431 |
+
"grad_norm": 0.022677181288599968,
|
5432 |
+
"learning_rate": 8.961827939636196e-06,
|
5433 |
+
"loss": 0.0004,
|
5434 |
+
"step": 775
|
5435 |
+
},
|
5436 |
+
{
|
5437 |
+
"epoch": 0.8275126632897893,
|
5438 |
+
"grad_norm": 0.004976468160748482,
|
5439 |
+
"learning_rate": 8.854907465014477e-06,
|
5440 |
+
"loss": 0.0001,
|
5441 |
+
"step": 776
|
5442 |
+
},
|
5443 |
+
{
|
5444 |
+
"epoch": 0.8285790455878432,
|
5445 |
+
"grad_norm": 0.08786796778440475,
|
5446 |
+
"learning_rate": 8.748566641090433e-06,
|
5447 |
+
"loss": 0.0013,
|
5448 |
+
"step": 777
|
5449 |
+
},
|
5450 |
+
{
|
5451 |
+
"epoch": 0.8296454278858971,
|
5452 |
+
"grad_norm": 0.40387263894081116,
|
5453 |
+
"learning_rate": 8.64280696599008e-06,
|
5454 |
+
"loss": 0.0094,
|
5455 |
+
"step": 778
|
5456 |
+
},
|
5457 |
+
{
|
5458 |
+
"epoch": 0.8307118101839509,
|
5459 |
+
"grad_norm": 0.011585093103349209,
|
5460 |
+
"learning_rate": 8.537629929652186e-06,
|
5461 |
+
"loss": 0.0003,
|
5462 |
+
"step": 779
|
5463 |
+
},
|
5464 |
+
{
|
5465 |
+
"epoch": 0.8317781924820048,
|
5466 |
+
"grad_norm": 0.00565929152071476,
|
5467 |
+
"learning_rate": 8.433037013807365e-06,
|
5468 |
+
"loss": 0.0001,
|
5469 |
+
"step": 780
|
5470 |
+
},
|
5471 |
+
{
|
5472 |
+
"epoch": 0.8328445747800587,
|
5473 |
+
"grad_norm": 0.012812900356948376,
|
5474 |
+
"learning_rate": 8.329029691957124e-06,
|
5475 |
+
"loss": 0.0003,
|
5476 |
+
"step": 781
|
5477 |
+
},
|
5478 |
+
{
|
5479 |
+
"epoch": 0.8339109570781125,
|
5480 |
+
"grad_norm": 0.0063051446340978146,
|
5481 |
+
"learning_rate": 8.225609429353187e-06,
|
5482 |
+
"loss": 0.0001,
|
5483 |
+
"step": 782
|
5484 |
+
},
|
5485 |
+
{
|
5486 |
+
"epoch": 0.8349773393761664,
|
5487 |
+
"grad_norm": 0.09649275243282318,
|
5488 |
+
"learning_rate": 8.122777682976745e-06,
|
5489 |
+
"loss": 0.002,
|
5490 |
+
"step": 783
|
5491 |
+
},
|
5492 |
+
{
|
5493 |
+
"epoch": 0.8360437216742203,
|
5494 |
+
"grad_norm": 0.03791379928588867,
|
5495 |
+
"learning_rate": 8.02053590151805e-06,
|
5496 |
+
"loss": 0.0008,
|
5497 |
+
"step": 784
|
5498 |
+
},
|
5499 |
+
{
|
5500 |
+
"epoch": 0.837110103972274,
|
5501 |
+
"grad_norm": 0.2502143681049347,
|
5502 |
+
"learning_rate": 7.918885525355935e-06,
|
5503 |
+
"loss": 0.001,
|
5504 |
+
"step": 785
|
5505 |
+
},
|
5506 |
+
{
|
5507 |
+
"epoch": 0.8381764862703279,
|
5508 |
+
"grad_norm": 0.05838427320122719,
|
5509 |
+
"learning_rate": 7.817827986537507e-06,
|
5510 |
+
"loss": 0.0006,
|
5511 |
+
"step": 786
|
5512 |
+
},
|
5513 |
+
{
|
5514 |
+
"epoch": 0.8392428685683818,
|
5515 |
+
"grad_norm": 0.010844585485756397,
|
5516 |
+
"learning_rate": 7.717364708758024e-06,
|
5517 |
+
"loss": 0.0003,
|
5518 |
+
"step": 787
|
5519 |
+
},
|
5520 |
+
{
|
5521 |
+
"epoch": 0.8403092508664356,
|
5522 |
+
"grad_norm": 0.24611002206802368,
|
5523 |
+
"learning_rate": 7.617497107340788e-06,
|
5524 |
+
"loss": 0.0038,
|
5525 |
+
"step": 788
|
5526 |
+
},
|
5527 |
+
{
|
5528 |
+
"epoch": 0.8413756331644895,
|
5529 |
+
"grad_norm": 0.004596321377903223,
|
5530 |
+
"learning_rate": 7.518226589217287e-06,
|
5531 |
+
"loss": 0.0001,
|
5532 |
+
"step": 789
|
5533 |
+
},
|
5534 |
+
{
|
5535 |
+
"epoch": 0.8424420154625433,
|
5536 |
+
"grad_norm": 0.003910887520760298,
|
5537 |
+
"learning_rate": 7.41955455290726e-06,
|
5538 |
+
"loss": 0.0001,
|
5539 |
+
"step": 790
|
5540 |
+
},
|
5541 |
+
{
|
5542 |
+
"epoch": 0.8435083977605972,
|
5543 |
+
"grad_norm": 0.011446366086602211,
|
5544 |
+
"learning_rate": 7.321482388499096e-06,
|
5545 |
+
"loss": 0.0002,
|
5546 |
+
"step": 791
|
5547 |
+
},
|
5548 |
+
{
|
5549 |
+
"epoch": 0.844574780058651,
|
5550 |
+
"grad_norm": 0.005088797304779291,
|
5551 |
+
"learning_rate": 7.224011477630166e-06,
|
5552 |
+
"loss": 0.0002,
|
5553 |
+
"step": 792
|
5554 |
+
},
|
5555 |
+
{
|
5556 |
+
"epoch": 0.8456411623567048,
|
5557 |
+
"grad_norm": 0.01832517422735691,
|
5558 |
+
"learning_rate": 7.127143193467445e-06,
|
5559 |
+
"loss": 0.0005,
|
5560 |
+
"step": 793
|
5561 |
+
},
|
5562 |
+
{
|
5563 |
+
"epoch": 0.8467075446547587,
|
5564 |
+
"grad_norm": 0.010378585197031498,
|
5565 |
+
"learning_rate": 7.030878900688115e-06,
|
5566 |
+
"loss": 0.0003,
|
5567 |
+
"step": 794
|
5568 |
+
},
|
5569 |
+
{
|
5570 |
+
"epoch": 0.8477739269528126,
|
5571 |
+
"grad_norm": 0.006652985233813524,
|
5572 |
+
"learning_rate": 6.93521995546031e-06,
|
5573 |
+
"loss": 0.0002,
|
5574 |
+
"step": 795
|
5575 |
+
},
|
5576 |
+
{
|
5577 |
+
"epoch": 0.8488403092508664,
|
5578 |
+
"grad_norm": 0.01605672389268875,
|
5579 |
+
"learning_rate": 6.840167705424105e-06,
|
5580 |
+
"loss": 0.0004,
|
5581 |
+
"step": 796
|
5582 |
+
},
|
5583 |
+
{
|
5584 |
+
"epoch": 0.8499066915489203,
|
5585 |
+
"grad_norm": 0.003394564613699913,
|
5586 |
+
"learning_rate": 6.745723489672412e-06,
|
5587 |
+
"loss": 0.0001,
|
5588 |
+
"step": 797
|
5589 |
+
},
|
5590 |
+
{
|
5591 |
+
"epoch": 0.8509730738469742,
|
5592 |
+
"grad_norm": 0.4180551767349243,
|
5593 |
+
"learning_rate": 6.651888638732229e-06,
|
5594 |
+
"loss": 0.0054,
|
5595 |
+
"step": 798
|
5596 |
+
},
|
5597 |
+
{
|
5598 |
+
"epoch": 0.852039456145028,
|
5599 |
+
"grad_norm": 0.019269771873950958,
|
5600 |
+
"learning_rate": 6.558664474545817e-06,
|
5601 |
+
"loss": 0.0004,
|
5602 |
+
"step": 799
|
5603 |
+
},
|
5604 |
+
{
|
5605 |
+
"epoch": 0.8531058384430819,
|
5606 |
+
"grad_norm": 0.017314445227384567,
|
5607 |
+
"learning_rate": 6.466052310452092e-06,
|
5608 |
+
"loss": 0.0005,
|
5609 |
+
"step": 800
|
5610 |
+
},
|
5611 |
+
{
|
5612 |
+
"epoch": 0.8541722207411357,
|
5613 |
+
"grad_norm": 0.005645404104143381,
|
5614 |
+
"learning_rate": 6.374053451168166e-06,
|
5615 |
+
"loss": 0.0001,
|
5616 |
+
"step": 801
|
5617 |
+
},
|
5618 |
+
{
|
5619 |
+
"epoch": 0.8552386030391895,
|
5620 |
+
"grad_norm": 0.006349225994199514,
|
5621 |
+
"learning_rate": 6.282669192770896e-06,
|
5622 |
+
"loss": 0.0002,
|
5623 |
+
"step": 802
|
5624 |
+
},
|
5625 |
+
{
|
5626 |
+
"epoch": 0.8563049853372434,
|
5627 |
+
"grad_norm": 0.0033127523493021727,
|
5628 |
+
"learning_rate": 6.191900822678698e-06,
|
5629 |
+
"loss": 0.0001,
|
5630 |
+
"step": 803
|
5631 |
+
},
|
5632 |
+
{
|
5633 |
+
"epoch": 0.8573713676352972,
|
5634 |
+
"grad_norm": 0.5225967168807983,
|
5635 |
+
"learning_rate": 6.101749619633346e-06,
|
5636 |
+
"loss": 0.0082,
|
5637 |
+
"step": 804
|
5638 |
+
},
|
5639 |
+
{
|
5640 |
+
"epoch": 0.8584377499333511,
|
5641 |
+
"grad_norm": 0.007548854220658541,
|
5642 |
+
"learning_rate": 6.012216853682001e-06,
|
5643 |
+
"loss": 0.0002,
|
5644 |
+
"step": 805
|
5645 |
+
},
|
5646 |
+
{
|
5647 |
+
"epoch": 0.859504132231405,
|
5648 |
+
"grad_norm": 0.04941653832793236,
|
5649 |
+
"learning_rate": 5.92330378615929e-06,
|
5650 |
+
"loss": 0.0005,
|
5651 |
+
"step": 806
|
5652 |
+
},
|
5653 |
+
{
|
5654 |
+
"epoch": 0.8605705145294588,
|
5655 |
+
"grad_norm": 0.18256814777851105,
|
5656 |
+
"learning_rate": 5.83501166966956e-06,
|
5657 |
+
"loss": 0.0014,
|
5658 |
+
"step": 807
|
5659 |
+
},
|
5660 |
+
{
|
5661 |
+
"epoch": 0.8616368968275127,
|
5662 |
+
"grad_norm": 0.030704107135534286,
|
5663 |
+
"learning_rate": 5.747341748069229e-06,
|
5664 |
+
"loss": 0.0005,
|
5665 |
+
"step": 808
|
5666 |
+
},
|
5667 |
+
{
|
5668 |
+
"epoch": 0.8627032791255665,
|
5669 |
+
"grad_norm": 0.009448669850826263,
|
5670 |
+
"learning_rate": 5.660295256449233e-06,
|
5671 |
+
"loss": 0.0002,
|
5672 |
+
"step": 809
|
5673 |
+
},
|
5674 |
+
{
|
5675 |
+
"epoch": 0.8637696614236203,
|
5676 |
+
"grad_norm": 0.01559404656291008,
|
5677 |
+
"learning_rate": 5.573873421117642e-06,
|
5678 |
+
"loss": 0.0003,
|
5679 |
+
"step": 810
|
5680 |
+
},
|
5681 |
+
{
|
5682 |
+
"epoch": 0.8648360437216742,
|
5683 |
+
"grad_norm": 0.033919911831617355,
|
5684 |
+
"learning_rate": 5.488077459582425e-06,
|
5685 |
+
"loss": 0.0006,
|
5686 |
+
"step": 811
|
5687 |
+
},
|
5688 |
+
{
|
5689 |
+
"epoch": 0.8659024260197281,
|
5690 |
+
"grad_norm": 0.6849818229675293,
|
5691 |
+
"learning_rate": 5.402908580534233e-06,
|
5692 |
+
"loss": 0.0135,
|
5693 |
+
"step": 812
|
5694 |
+
},
|
5695 |
+
{
|
5696 |
+
"epoch": 0.8669688083177819,
|
5697 |
+
"grad_norm": 0.005436232313513756,
|
5698 |
+
"learning_rate": 5.318367983829392e-06,
|
5699 |
+
"loss": 0.0001,
|
5700 |
+
"step": 813
|
5701 |
+
},
|
5702 |
+
{
|
5703 |
+
"epoch": 0.8680351906158358,
|
5704 |
+
"grad_norm": 0.005175617057830095,
|
5705 |
+
"learning_rate": 5.234456860473042e-06,
|
5706 |
+
"loss": 0.0002,
|
5707 |
+
"step": 814
|
5708 |
+
},
|
5709 |
+
{
|
5710 |
+
"epoch": 0.8691015729138897,
|
5711 |
+
"grad_norm": 0.006666592322289944,
|
5712 |
+
"learning_rate": 5.151176392602292e-06,
|
5713 |
+
"loss": 0.0002,
|
5714 |
+
"step": 815
|
5715 |
+
},
|
5716 |
+
{
|
5717 |
+
"epoch": 0.8701679552119435,
|
5718 |
+
"grad_norm": 0.05537564679980278,
|
5719 |
+
"learning_rate": 5.068527753469604e-06,
|
5720 |
+
"loss": 0.0014,
|
5721 |
+
"step": 816
|
5722 |
+
},
|
5723 |
+
{
|
5724 |
+
"epoch": 0.8712343375099973,
|
5725 |
+
"grad_norm": 0.016136392951011658,
|
5726 |
+
"learning_rate": 4.986512107426283e-06,
|
5727 |
+
"loss": 0.0002,
|
5728 |
+
"step": 817
|
5729 |
+
},
|
5730 |
+
{
|
5731 |
+
"epoch": 0.8723007198080512,
|
5732 |
+
"grad_norm": 0.004150747321546078,
|
5733 |
+
"learning_rate": 4.905130609906005e-06,
|
5734 |
+
"loss": 0.0001,
|
5735 |
+
"step": 818
|
5736 |
+
},
|
5737 |
+
{
|
5738 |
+
"epoch": 0.873367102106105,
|
5739 |
+
"grad_norm": 0.0023105363361537457,
|
5740 |
+
"learning_rate": 4.824384407408622e-06,
|
5741 |
+
"loss": 0.0001,
|
5742 |
+
"step": 819
|
5743 |
+
},
|
5744 |
+
{
|
5745 |
+
"epoch": 0.8744334844041589,
|
5746 |
+
"grad_norm": 0.0036519451532512903,
|
5747 |
+
"learning_rate": 4.744274637483936e-06,
|
5748 |
+
"loss": 0.0001,
|
5749 |
+
"step": 820
|
5750 |
+
},
|
5751 |
+
{
|
5752 |
+
"epoch": 0.8754998667022127,
|
5753 |
+
"grad_norm": 0.4714241623878479,
|
5754 |
+
"learning_rate": 4.6648024287157536e-06,
|
5755 |
+
"loss": 0.0017,
|
5756 |
+
"step": 821
|
5757 |
+
},
|
5758 |
+
{
|
5759 |
+
"epoch": 0.8765662490002666,
|
5760 |
+
"grad_norm": 0.009708943776786327,
|
5761 |
+
"learning_rate": 4.585968900705889e-06,
|
5762 |
+
"loss": 0.0002,
|
5763 |
+
"step": 822
|
5764 |
+
},
|
5765 |
+
{
|
5766 |
+
"epoch": 0.8776326312983205,
|
5767 |
+
"grad_norm": 0.0021507502533495426,
|
5768 |
+
"learning_rate": 4.507775164058469e-06,
|
5769 |
+
"loss": 0.0001,
|
5770 |
+
"step": 823
|
5771 |
+
},
|
5772 |
+
{
|
5773 |
+
"epoch": 0.8786990135963743,
|
5774 |
+
"grad_norm": 0.005819080397486687,
|
5775 |
+
"learning_rate": 4.43022232036428e-06,
|
5776 |
+
"loss": 0.0001,
|
5777 |
+
"step": 824
|
5778 |
+
},
|
5779 |
+
{
|
5780 |
+
"epoch": 0.8797653958944281,
|
5781 |
+
"grad_norm": 0.003718499094247818,
|
5782 |
+
"learning_rate": 4.353311462185183e-06,
|
5783 |
+
"loss": 0.0001,
|
5784 |
+
"step": 825
|
5785 |
+
},
|
5786 |
+
{
|
5787 |
+
"epoch": 0.880831778192482,
|
5788 |
+
"grad_norm": 0.003342332784086466,
|
5789 |
+
"learning_rate": 4.277043673038816e-06,
|
5790 |
+
"loss": 0.0001,
|
5791 |
+
"step": 826
|
5792 |
+
},
|
5793 |
+
{
|
5794 |
+
"epoch": 0.8818981604905358,
|
5795 |
+
"grad_norm": 0.008588786236941814,
|
5796 |
+
"learning_rate": 4.20142002738324e-06,
|
5797 |
+
"loss": 0.0002,
|
5798 |
+
"step": 827
|
5799 |
+
},
|
5800 |
+
{
|
5801 |
+
"epoch": 0.8829645427885897,
|
5802 |
+
"grad_norm": 0.671940267086029,
|
5803 |
+
"learning_rate": 4.126441590601893e-06,
|
5804 |
+
"loss": 0.0103,
|
5805 |
+
"step": 828
|
5806 |
+
},
|
5807 |
+
{
|
5808 |
+
"epoch": 0.8840309250866436,
|
5809 |
+
"grad_norm": 0.03621124476194382,
|
5810 |
+
"learning_rate": 4.05210941898847e-06,
|
5811 |
+
"loss": 0.0005,
|
5812 |
+
"step": 829
|
5813 |
+
},
|
5814 |
+
{
|
5815 |
+
"epoch": 0.8850973073846974,
|
5816 |
+
"grad_norm": 0.006909804884344339,
|
5817 |
+
"learning_rate": 3.978424559732158e-06,
|
5818 |
+
"loss": 0.0002,
|
5819 |
+
"step": 830
|
5820 |
+
},
|
5821 |
+
{
|
5822 |
+
"epoch": 0.8861636896827513,
|
5823 |
+
"grad_norm": 0.03682345896959305,
|
5824 |
+
"learning_rate": 3.905388050902809e-06,
|
5825 |
+
"loss": 0.0006,
|
5826 |
+
"step": 831
|
5827 |
+
},
|
5828 |
+
{
|
5829 |
+
"epoch": 0.8872300719808052,
|
5830 |
+
"grad_norm": 0.021985987201333046,
|
5831 |
+
"learning_rate": 3.833000921436319e-06,
|
5832 |
+
"loss": 0.0005,
|
5833 |
+
"step": 832
|
5834 |
+
},
|
5835 |
+
{
|
5836 |
+
"epoch": 0.888296454278859,
|
5837 |
+
"grad_norm": 0.005545215681195259,
|
5838 |
+
"learning_rate": 3.7612641911201894e-06,
|
5839 |
+
"loss": 0.0001,
|
5840 |
+
"step": 833
|
5841 |
+
},
|
5842 |
+
{
|
5843 |
+
"epoch": 0.8893628365769128,
|
5844 |
+
"grad_norm": 0.0029357897583395243,
|
5845 |
+
"learning_rate": 3.690178870579075e-06,
|
5846 |
+
"loss": 0.0001,
|
5847 |
+
"step": 834
|
5848 |
+
},
|
5849 |
+
{
|
5850 |
+
"epoch": 0.8904292188749666,
|
5851 |
+
"grad_norm": 0.44927045702934265,
|
5852 |
+
"learning_rate": 3.619745961260623e-06,
|
5853 |
+
"loss": 0.008,
|
5854 |
+
"step": 835
|
5855 |
+
},
|
5856 |
+
{
|
5857 |
+
"epoch": 0.8914956011730205,
|
5858 |
+
"grad_norm": 0.010482368990778923,
|
5859 |
+
"learning_rate": 3.549966455421305e-06,
|
5860 |
+
"loss": 0.0002,
|
5861 |
+
"step": 836
|
5862 |
+
},
|
5863 |
+
{
|
5864 |
+
"epoch": 0.8925619834710744,
|
5865 |
+
"grad_norm": 0.008478675968945026,
|
5866 |
+
"learning_rate": 3.4808413361125e-06,
|
5867 |
+
"loss": 0.0003,
|
5868 |
+
"step": 837
|
5869 |
+
},
|
5870 |
+
{
|
5871 |
+
"epoch": 0.8936283657691282,
|
5872 |
+
"grad_norm": 0.005715550389140844,
|
5873 |
+
"learning_rate": 3.4123715771665786e-06,
|
5874 |
+
"loss": 0.0001,
|
5875 |
+
"step": 838
|
5876 |
+
},
|
5877 |
+
{
|
5878 |
+
"epoch": 0.8946947480671821,
|
5879 |
+
"grad_norm": 0.00493262056261301,
|
5880 |
+
"learning_rate": 3.3445581431832463e-06,
|
5881 |
+
"loss": 0.0001,
|
5882 |
+
"step": 839
|
5883 |
+
},
|
5884 |
+
{
|
5885 |
+
"epoch": 0.895761130365236,
|
5886 |
+
"grad_norm": 0.07337388396263123,
|
5887 |
+
"learning_rate": 3.27740198951591e-06,
|
5888 |
+
"loss": 0.0012,
|
5889 |
+
"step": 840
|
5890 |
+
},
|
5891 |
+
{
|
5892 |
+
"epoch": 0.8968275126632897,
|
5893 |
+
"grad_norm": 0.03428462892770767,
|
5894 |
+
"learning_rate": 3.2109040622582186e-06,
|
5895 |
+
"loss": 0.0006,
|
5896 |
+
"step": 841
|
5897 |
+
},
|
5898 |
+
{
|
5899 |
+
"epoch": 0.8978938949613436,
|
5900 |
+
"grad_norm": 0.6033999919891357,
|
5901 |
+
"learning_rate": 3.1450652982307816e-06,
|
5902 |
+
"loss": 0.0076,
|
5903 |
+
"step": 842
|
5904 |
+
},
|
5905 |
+
{
|
5906 |
+
"epoch": 0.8989602772593975,
|
5907 |
+
"grad_norm": 0.005725290160626173,
|
5908 |
+
"learning_rate": 3.079886624967904e-06,
|
5909 |
+
"loss": 0.0002,
|
5910 |
+
"step": 843
|
5911 |
+
},
|
5912 |
+
{
|
5913 |
+
"epoch": 0.9000266595574513,
|
5914 |
+
"grad_norm": 0.5092418789863586,
|
5915 |
+
"learning_rate": 3.0153689607045845e-06,
|
5916 |
+
"loss": 0.0196,
|
5917 |
+
"step": 844
|
5918 |
+
},
|
5919 |
+
{
|
5920 |
+
"epoch": 0.9010930418555052,
|
5921 |
+
"grad_norm": 0.5069144368171692,
|
5922 |
+
"learning_rate": 2.9515132143635215e-06,
|
5923 |
+
"loss": 0.032,
|
5924 |
+
"step": 845
|
5925 |
+
},
|
5926 |
+
{
|
5927 |
+
"epoch": 0.9021594241535591,
|
5928 |
+
"grad_norm": 0.004980255849659443,
|
5929 |
+
"learning_rate": 2.8883202855423676e-06,
|
5930 |
+
"loss": 0.0001,
|
5931 |
+
"step": 846
|
5932 |
+
},
|
5933 |
+
{
|
5934 |
+
"epoch": 0.9032258064516129,
|
5935 |
+
"grad_norm": 0.012960120104253292,
|
5936 |
+
"learning_rate": 2.825791064500993e-06,
|
5937 |
+
"loss": 0.0002,
|
5938 |
+
"step": 847
|
5939 |
+
},
|
5940 |
+
{
|
5941 |
+
"epoch": 0.9042921887496668,
|
5942 |
+
"grad_norm": 0.017126217484474182,
|
5943 |
+
"learning_rate": 2.763926432148989e-06,
|
5944 |
+
"loss": 0.0004,
|
5945 |
+
"step": 848
|
5946 |
+
},
|
5947 |
+
{
|
5948 |
+
"epoch": 0.9053585710477207,
|
5949 |
+
"grad_norm": 0.21750324964523315,
|
5950 |
+
"learning_rate": 2.702727260033244e-06,
|
5951 |
+
"loss": 0.0033,
|
5952 |
+
"step": 849
|
5953 |
+
},
|
5954 |
+
{
|
5955 |
+
"epoch": 0.9064249533457744,
|
5956 |
+
"grad_norm": 0.033922020345926285,
|
5957 |
+
"learning_rate": 2.6421944103256657e-06,
|
5958 |
+
"loss": 0.0007,
|
5959 |
+
"step": 850
|
5960 |
+
},
|
5961 |
+
{
|
5962 |
+
"epoch": 0.9074913356438283,
|
5963 |
+
"grad_norm": 0.03163854032754898,
|
5964 |
+
"learning_rate": 2.5823287358110294e-06,
|
5965 |
+
"loss": 0.0005,
|
5966 |
+
"step": 851
|
5967 |
+
},
|
5968 |
+
{
|
5969 |
+
"epoch": 0.9085577179418821,
|
5970 |
+
"grad_norm": 0.024384107440710068,
|
5971 |
+
"learning_rate": 2.523131079874963e-06,
|
5972 |
+
"loss": 0.0003,
|
5973 |
+
"step": 852
|
5974 |
+
},
|
5975 |
+
{
|
5976 |
+
"epoch": 0.909624100239936,
|
5977 |
+
"grad_norm": 0.003758068894967437,
|
5978 |
+
"learning_rate": 2.464602276492084e-06,
|
5979 |
+
"loss": 0.0001,
|
5980 |
+
"step": 853
|
5981 |
+
},
|
5982 |
+
{
|
5983 |
+
"epoch": 0.9106904825379899,
|
5984 |
+
"grad_norm": 0.009570575319230556,
|
5985 |
+
"learning_rate": 2.406743150214241e-06,
|
5986 |
+
"loss": 0.0002,
|
5987 |
+
"step": 854
|
5988 |
+
},
|
5989 |
+
{
|
5990 |
+
"epoch": 0.9117568648360437,
|
5991 |
+
"grad_norm": 0.011624293401837349,
|
5992 |
+
"learning_rate": 2.3495545161588696e-06,
|
5993 |
+
"loss": 0.0003,
|
5994 |
+
"step": 855
|
5995 |
+
},
|
5996 |
+
{
|
5997 |
+
"epoch": 0.9128232471340976,
|
5998 |
+
"grad_norm": 0.003363889642059803,
|
5999 |
+
"learning_rate": 2.2930371799975594e-06,
|
6000 |
+
"loss": 0.0001,
|
6001 |
+
"step": 856
|
6002 |
+
},
|
6003 |
+
{
|
6004 |
+
"epoch": 0.9138896294321515,
|
6005 |
+
"grad_norm": 0.18208551406860352,
|
6006 |
+
"learning_rate": 2.2371919379446495e-06,
|
6007 |
+
"loss": 0.0016,
|
6008 |
+
"step": 857
|
6009 |
+
},
|
6010 |
+
{
|
6011 |
+
"epoch": 0.9149560117302052,
|
6012 |
+
"grad_norm": 0.004443952813744545,
|
6013 |
+
"learning_rate": 2.1820195767460805e-06,
|
6014 |
+
"loss": 0.0001,
|
6015 |
+
"step": 858
|
6016 |
+
},
|
6017 |
+
{
|
6018 |
+
"epoch": 0.9160223940282591,
|
6019 |
+
"grad_norm": 0.07205750048160553,
|
6020 |
+
"learning_rate": 2.127520873668226e-06,
|
6021 |
+
"loss": 0.0009,
|
6022 |
+
"step": 859
|
6023 |
+
},
|
6024 |
+
{
|
6025 |
+
"epoch": 0.917088776326313,
|
6026 |
+
"grad_norm": 0.004706911742687225,
|
6027 |
+
"learning_rate": 2.0736965964870127e-06,
|
6028 |
+
"loss": 0.0001,
|
6029 |
+
"step": 860
|
6030 |
+
},
|
6031 |
+
{
|
6032 |
+
"epoch": 0.9181551586243668,
|
6033 |
+
"grad_norm": 0.568541944026947,
|
6034 |
+
"learning_rate": 2.0205475034770604e-06,
|
6035 |
+
"loss": 0.041,
|
6036 |
+
"step": 861
|
6037 |
+
},
|
6038 |
+
{
|
6039 |
+
"epoch": 0.9192215409224207,
|
6040 |
+
"grad_norm": 0.01811330020427704,
|
6041 |
+
"learning_rate": 1.9680743434010384e-06,
|
6042 |
+
"loss": 0.0004,
|
6043 |
+
"step": 862
|
6044 |
+
},
|
6045 |
+
{
|
6046 |
+
"epoch": 0.9202879232204746,
|
6047 |
+
"grad_norm": 0.006233413703739643,
|
6048 |
+
"learning_rate": 1.9162778554990824e-06,
|
6049 |
+
"loss": 0.0001,
|
6050 |
+
"step": 863
|
6051 |
+
},
|
6052 |
+
{
|
6053 |
+
"epoch": 0.9213543055185284,
|
6054 |
+
"grad_norm": 0.015661677345633507,
|
6055 |
+
"learning_rate": 1.8651587694783923e-06,
|
6056 |
+
"loss": 0.0003,
|
6057 |
+
"step": 864
|
6058 |
+
},
|
6059 |
+
{
|
6060 |
+
"epoch": 0.9224206878165823,
|
6061 |
+
"grad_norm": 0.00473360437899828,
|
6062 |
+
"learning_rate": 1.8147178055029579e-06,
|
6063 |
+
"loss": 0.0001,
|
6064 |
+
"step": 865
|
6065 |
+
},
|
6066 |
+
{
|
6067 |
+
"epoch": 0.923487070114636,
|
6068 |
+
"grad_norm": 0.011882445774972439,
|
6069 |
+
"learning_rate": 1.7649556741833995e-06,
|
6070 |
+
"loss": 0.0003,
|
6071 |
+
"step": 866
|
6072 |
+
},
|
6073 |
+
{
|
6074 |
+
"epoch": 0.9245534524126899,
|
6075 |
+
"grad_norm": 0.18409405648708344,
|
6076 |
+
"learning_rate": 1.7158730765669817e-06,
|
6077 |
+
"loss": 0.0035,
|
6078 |
+
"step": 867
|
6079 |
+
},
|
6080 |
+
{
|
6081 |
+
"epoch": 0.9256198347107438,
|
6082 |
+
"grad_norm": 0.007312040776014328,
|
6083 |
+
"learning_rate": 1.6674707041276938e-06,
|
6084 |
+
"loss": 0.0002,
|
6085 |
+
"step": 868
|
6086 |
+
},
|
6087 |
+
{
|
6088 |
+
"epoch": 0.9266862170087976,
|
6089 |
+
"grad_norm": 0.008804773911833763,
|
6090 |
+
"learning_rate": 1.6197492387565627e-06,
|
6091 |
+
"loss": 0.0002,
|
6092 |
+
"step": 869
|
6093 |
+
},
|
6094 |
+
{
|
6095 |
+
"epoch": 0.9277525993068515,
|
6096 |
+
"grad_norm": 0.3166285753250122,
|
6097 |
+
"learning_rate": 1.572709352752011e-06,
|
6098 |
+
"loss": 0.0201,
|
6099 |
+
"step": 870
|
6100 |
+
},
|
6101 |
+
{
|
6102 |
+
"epoch": 0.9288189816049054,
|
6103 |
+
"grad_norm": 0.005019339732825756,
|
6104 |
+
"learning_rate": 1.5263517088103863e-06,
|
6105 |
+
"loss": 0.0001,
|
6106 |
+
"step": 871
|
6107 |
+
},
|
6108 |
+
{
|
6109 |
+
"epoch": 0.9298853639029592,
|
6110 |
+
"grad_norm": 0.003396758111193776,
|
6111 |
+
"learning_rate": 1.4806769600166359e-06,
|
6112 |
+
"loss": 0.0001,
|
6113 |
+
"step": 872
|
6114 |
+
},
|
6115 |
+
{
|
6116 |
+
"epoch": 0.9309517462010131,
|
6117 |
+
"grad_norm": 0.022154012694954872,
|
6118 |
+
"learning_rate": 1.4356857498351029e-06,
|
6119 |
+
"loss": 0.0003,
|
6120 |
+
"step": 873
|
6121 |
+
},
|
6122 |
+
{
|
6123 |
+
"epoch": 0.932018128499067,
|
6124 |
+
"grad_norm": 0.011645108461380005,
|
6125 |
+
"learning_rate": 1.3913787121004719e-06,
|
6126 |
+
"loss": 0.0003,
|
6127 |
+
"step": 874
|
6128 |
+
},
|
6129 |
+
{
|
6130 |
+
"epoch": 0.9330845107971207,
|
6131 |
+
"grad_norm": 0.004831390921026468,
|
6132 |
+
"learning_rate": 1.3477564710088098e-06,
|
6133 |
+
"loss": 0.0001,
|
6134 |
+
"step": 875
|
6135 |
+
},
|
6136 |
+
{
|
6137 |
+
"epoch": 0.9341508930951746,
|
6138 |
+
"grad_norm": 0.005583819001913071,
|
6139 |
+
"learning_rate": 1.304819641108801e-06,
|
6140 |
+
"loss": 0.0002,
|
6141 |
+
"step": 876
|
6142 |
+
},
|
6143 |
+
{
|
6144 |
+
"epoch": 0.9352172753932285,
|
6145 |
+
"grad_norm": 0.025920310989022255,
|
6146 |
+
"learning_rate": 1.2625688272930926e-06,
|
6147 |
+
"loss": 0.0002,
|
6148 |
+
"step": 877
|
6149 |
+
},
|
6150 |
+
{
|
6151 |
+
"epoch": 0.9362836576912823,
|
6152 |
+
"grad_norm": 0.010310935787856579,
|
6153 |
+
"learning_rate": 1.2210046247897344e-06,
|
6154 |
+
"loss": 0.0003,
|
6155 |
+
"step": 878
|
6156 |
+
},
|
6157 |
+
{
|
6158 |
+
"epoch": 0.9373500399893362,
|
6159 |
+
"grad_norm": 0.005592814180999994,
|
6160 |
+
"learning_rate": 1.180127619153837e-06,
|
6161 |
+
"loss": 0.0001,
|
6162 |
+
"step": 879
|
6163 |
+
},
|
6164 |
+
{
|
6165 |
+
"epoch": 0.9384164222873901,
|
6166 |
+
"grad_norm": 0.0035225884057581425,
|
6167 |
+
"learning_rate": 1.1399383862592927e-06,
|
6168 |
+
"loss": 0.0001,
|
6169 |
+
"step": 880
|
6170 |
+
},
|
6171 |
+
{
|
6172 |
+
"epoch": 0.9394828045854439,
|
6173 |
+
"grad_norm": 0.014407276175916195,
|
6174 |
+
"learning_rate": 1.1004374922906847e-06,
|
6175 |
+
"loss": 0.0002,
|
6176 |
+
"step": 881
|
6177 |
+
},
|
6178 |
+
{
|
6179 |
+
"epoch": 0.9405491868834978,
|
6180 |
+
"grad_norm": 0.01085433829575777,
|
6181 |
+
"learning_rate": 1.0616254937352966e-06,
|
6182 |
+
"loss": 0.0002,
|
6183 |
+
"step": 882
|
6184 |
+
},
|
6185 |
+
{
|
6186 |
+
"epoch": 0.9416155691815515,
|
6187 |
+
"grad_norm": 0.0071522849611938,
|
6188 |
+
"learning_rate": 1.0235029373752758e-06,
|
6189 |
+
"loss": 0.0001,
|
6190 |
+
"step": 883
|
6191 |
+
},
|
6192 |
+
{
|
6193 |
+
"epoch": 0.9426819514796054,
|
6194 |
+
"grad_norm": 0.005545539781451225,
|
6195 |
+
"learning_rate": 9.860703602799281e-07,
|
6196 |
+
"loss": 0.0002,
|
6197 |
+
"step": 884
|
6198 |
+
},
|
6199 |
+
{
|
6200 |
+
"epoch": 0.9437483337776593,
|
6201 |
+
"grad_norm": 0.01488356851041317,
|
6202 |
+
"learning_rate": 9.49328289798157e-07,
|
6203 |
+
"loss": 0.0003,
|
6204 |
+
"step": 885
|
6205 |
+
},
|
6206 |
+
{
|
6207 |
+
"epoch": 0.9448147160757131,
|
6208 |
+
"grad_norm": 0.39096885919570923,
|
6209 |
+
"learning_rate": 9.132772435510362e-07,
|
6210 |
+
"loss": 0.0359,
|
6211 |
+
"step": 886
|
6212 |
+
},
|
6213 |
+
{
|
6214 |
+
"epoch": 0.945881098373767,
|
6215 |
+
"grad_norm": 0.023358656093478203,
|
6216 |
+
"learning_rate": 8.779177294245045e-07,
|
6217 |
+
"loss": 0.0004,
|
6218 |
+
"step": 887
|
6219 |
+
},
|
6220 |
+
{
|
6221 |
+
"epoch": 0.9469474806718209,
|
6222 |
+
"grad_norm": 0.04002382233738899,
|
6223 |
+
"learning_rate": 8.432502455622216e-07,
|
6224 |
+
"loss": 0.0007,
|
6225 |
+
"step": 888
|
6226 |
+
},
|
6227 |
+
{
|
6228 |
+
"epoch": 0.9480138629698747,
|
6229 |
+
"grad_norm": 0.005228077061474323,
|
6230 |
+
"learning_rate": 8.092752803585513e-07,
|
6231 |
+
"loss": 0.0002,
|
6232 |
+
"step": 889
|
6233 |
+
},
|
6234 |
+
{
|
6235 |
+
"epoch": 0.9490802452679286,
|
6236 |
+
"grad_norm": 0.007714355830103159,
|
6237 |
+
"learning_rate": 7.759933124516727e-07,
|
6238 |
+
"loss": 0.0002,
|
6239 |
+
"step": 890
|
6240 |
+
},
|
6241 |
+
{
|
6242 |
+
"epoch": 0.9501466275659824,
|
6243 |
+
"grad_norm": 0.06364113837480545,
|
6244 |
+
"learning_rate": 7.434048107168523e-07,
|
6245 |
+
"loss": 0.0008,
|
6246 |
+
"step": 891
|
6247 |
+
},
|
6248 |
+
{
|
6249 |
+
"epoch": 0.9512130098640362,
|
6250 |
+
"grad_norm": 0.022954344749450684,
|
6251 |
+
"learning_rate": 7.115102342598101e-07,
|
6252 |
+
"loss": 0.0004,
|
6253 |
+
"step": 892
|
6254 |
+
},
|
6255 |
+
{
|
6256 |
+
"epoch": 0.9522793921620901,
|
6257 |
+
"grad_norm": 0.013913022354245186,
|
6258 |
+
"learning_rate": 6.803100324102918e-07,
|
6259 |
+
"loss": 0.0004,
|
6260 |
+
"step": 893
|
6261 |
+
},
|
6262 |
+
{
|
6263 |
+
"epoch": 0.953345774460144,
|
6264 |
+
"grad_norm": 0.27097034454345703,
|
6265 |
+
"learning_rate": 6.498046447156958e-07,
|
6266 |
+
"loss": 0.0041,
|
6267 |
+
"step": 894
|
6268 |
+
},
|
6269 |
+
{
|
6270 |
+
"epoch": 0.9544121567581978,
|
6271 |
+
"grad_norm": 0.2270175814628601,
|
6272 |
+
"learning_rate": 6.199945009349173e-07,
|
6273 |
+
"loss": 0.0028,
|
6274 |
+
"step": 895
|
6275 |
+
},
|
6276 |
+
{
|
6277 |
+
"epoch": 0.9554785390562517,
|
6278 |
+
"grad_norm": 0.010744959115982056,
|
6279 |
+
"learning_rate": 5.908800210322696e-07,
|
6280 |
+
"loss": 0.0002,
|
6281 |
+
"step": 896
|
6282 |
+
},
|
6283 |
+
{
|
6284 |
+
"epoch": 0.9565449213543055,
|
6285 |
+
"grad_norm": 0.027118468657135963,
|
6286 |
+
"learning_rate": 5.624616151715834e-07,
|
6287 |
+
"loss": 0.0005,
|
6288 |
+
"step": 897
|
6289 |
+
},
|
6290 |
+
{
|
6291 |
+
"epoch": 0.9576113036523594,
|
6292 |
+
"grad_norm": 0.008466150611639023,
|
6293 |
+
"learning_rate": 5.347396837104057e-07,
|
6294 |
+
"loss": 0.0003,
|
6295 |
+
"step": 898
|
6296 |
+
},
|
6297 |
+
{
|
6298 |
+
"epoch": 0.9586776859504132,
|
6299 |
+
"grad_norm": 0.5483177900314331,
|
6300 |
+
"learning_rate": 5.077146171943936e-07,
|
6301 |
+
"loss": 0.0265,
|
6302 |
+
"step": 899
|
6303 |
+
},
|
6304 |
+
{
|
6305 |
+
"epoch": 0.959744068248467,
|
6306 |
+
"grad_norm": 0.10898708552122116,
|
6307 |
+
"learning_rate": 4.813867963517904e-07,
|
6308 |
+
"loss": 0.0019,
|
6309 |
+
"step": 900
|
6310 |
+
},
|
6311 |
+
{
|
6312 |
+
"epoch": 0.9608104505465209,
|
6313 |
+
"grad_norm": 0.9734861254692078,
|
6314 |
+
"learning_rate": 4.55756592088058e-07,
|
6315 |
+
"loss": 0.0241,
|
6316 |
+
"step": 901
|
6317 |
+
},
|
6318 |
+
{
|
6319 |
+
"epoch": 0.9618768328445748,
|
6320 |
+
"grad_norm": 0.010221785865724087,
|
6321 |
+
"learning_rate": 4.308243654806643e-07,
|
6322 |
+
"loss": 0.0002,
|
6323 |
+
"step": 902
|
6324 |
+
},
|
6325 |
+
{
|
6326 |
+
"epoch": 0.9629432151426286,
|
6327 |
+
"grad_norm": 0.0044042919762432575,
|
6328 |
+
"learning_rate": 4.0659046777398736e-07,
|
6329 |
+
"loss": 0.0001,
|
6330 |
+
"step": 903
|
6331 |
+
},
|
6332 |
+
{
|
6333 |
+
"epoch": 0.9640095974406825,
|
6334 |
+
"grad_norm": 0.03026667982339859,
|
6335 |
+
"learning_rate": 3.8305524037438034e-07,
|
6336 |
+
"loss": 0.0004,
|
6337 |
+
"step": 904
|
6338 |
+
},
|
6339 |
+
{
|
6340 |
+
"epoch": 0.9650759797387364,
|
6341 |
+
"grad_norm": 0.01675579324364662,
|
6342 |
+
"learning_rate": 3.602190148453366e-07,
|
6343 |
+
"loss": 0.0003,
|
6344 |
+
"step": 905
|
6345 |
+
},
|
6346 |
+
{
|
6347 |
+
"epoch": 0.9661423620367902,
|
6348 |
+
"grad_norm": 0.025279685854911804,
|
6349 |
+
"learning_rate": 3.380821129028489e-07,
|
6350 |
+
"loss": 0.0006,
|
6351 |
+
"step": 906
|
6352 |
+
},
|
6353 |
+
{
|
6354 |
+
"epoch": 0.967208744334844,
|
6355 |
+
"grad_norm": 0.004679596517235041,
|
6356 |
+
"learning_rate": 3.1664484641086287e-07,
|
6357 |
+
"loss": 0.0001,
|
6358 |
+
"step": 907
|
6359 |
+
},
|
6360 |
+
{
|
6361 |
+
"epoch": 0.9682751266328979,
|
6362 |
+
"grad_norm": 0.015383715741336346,
|
6363 |
+
"learning_rate": 2.9590751737685883e-07,
|
6364 |
+
"loss": 0.0002,
|
6365 |
+
"step": 908
|
6366 |
+
},
|
6367 |
+
{
|
6368 |
+
"epoch": 0.9693415089309517,
|
6369 |
+
"grad_norm": 0.23335658013820648,
|
6370 |
+
"learning_rate": 2.758704179476601e-07,
|
6371 |
+
"loss": 0.0022,
|
6372 |
+
"step": 909
|
6373 |
+
},
|
6374 |
+
{
|
6375 |
+
"epoch": 0.9704078912290056,
|
6376 |
+
"grad_norm": 0.0032410675194114447,
|
6377 |
+
"learning_rate": 2.5653383040524227e-07,
|
6378 |
+
"loss": 0.0001,
|
6379 |
+
"step": 910
|
6380 |
+
},
|
6381 |
+
{
|
6382 |
+
"epoch": 0.9714742735270594,
|
6383 |
+
"grad_norm": 0.0194519255310297,
|
6384 |
+
"learning_rate": 2.378980271628195e-07,
|
6385 |
+
"loss": 0.0004,
|
6386 |
+
"step": 911
|
6387 |
+
},
|
6388 |
+
{
|
6389 |
+
"epoch": 0.9725406558251133,
|
6390 |
+
"grad_norm": 0.004819793161004782,
|
6391 |
+
"learning_rate": 2.1996327076096447e-07,
|
6392 |
+
"loss": 0.0001,
|
6393 |
+
"step": 912
|
6394 |
+
},
|
6395 |
+
{
|
6396 |
+
"epoch": 0.9736070381231672,
|
6397 |
+
"grad_norm": 0.010149155743420124,
|
6398 |
+
"learning_rate": 2.027298138639333e-07,
|
6399 |
+
"loss": 0.0002,
|
6400 |
+
"step": 913
|
6401 |
+
},
|
6402 |
+
{
|
6403 |
+
"epoch": 0.974673420421221,
|
6404 |
+
"grad_norm": 0.004880509339272976,
|
6405 |
+
"learning_rate": 1.8619789925608534e-07,
|
6406 |
+
"loss": 0.0001,
|
6407 |
+
"step": 914
|
6408 |
+
},
|
6409 |
+
{
|
6410 |
+
"epoch": 0.9757398027192749,
|
6411 |
+
"grad_norm": 0.12890613079071045,
|
6412 |
+
"learning_rate": 1.7036775983849674e-07,
|
6413 |
+
"loss": 0.0025,
|
6414 |
+
"step": 915
|
6415 |
+
},
|
6416 |
+
{
|
6417 |
+
"epoch": 0.9768061850173287,
|
6418 |
+
"grad_norm": 0.038326557725667953,
|
6419 |
+
"learning_rate": 1.5523961862564106e-07,
|
6420 |
+
"loss": 0.0004,
|
6421 |
+
"step": 916
|
6422 |
+
},
|
6423 |
+
{
|
6424 |
+
"epoch": 0.9778725673153825,
|
6425 |
+
"grad_norm": 0.010187475942075253,
|
6426 |
+
"learning_rate": 1.4081368874226398e-07,
|
6427 |
+
"loss": 0.0002,
|
6428 |
+
"step": 917
|
6429 |
+
},
|
6430 |
+
{
|
6431 |
+
"epoch": 0.9789389496134364,
|
6432 |
+
"grad_norm": 0.01759922318160534,
|
6433 |
+
"learning_rate": 1.2709017342039664e-07,
|
6434 |
+
"loss": 0.0003,
|
6435 |
+
"step": 918
|
6436 |
+
},
|
6437 |
+
{
|
6438 |
+
"epoch": 0.9800053319114903,
|
6439 |
+
"grad_norm": 0.008808497339487076,
|
6440 |
+
"learning_rate": 1.1406926599646373e-07,
|
6441 |
+
"loss": 0.0002,
|
6442 |
+
"step": 919
|
6443 |
+
},
|
6444 |
+
{
|
6445 |
+
"epoch": 0.9810717142095441,
|
6446 |
+
"grad_norm": 0.0033097113482654095,
|
6447 |
+
"learning_rate": 1.0175114990857437e-07,
|
6448 |
+
"loss": 0.0001,
|
6449 |
+
"step": 920
|
6450 |
+
},
|
6451 |
+
{
|
6452 |
+
"epoch": 0.982138096507598,
|
6453 |
+
"grad_norm": 0.003725285641849041,
|
6454 |
+
"learning_rate": 9.013599869394095e-08,
|
6455 |
+
"loss": 0.0001,
|
6456 |
+
"step": 921
|
6457 |
+
},
|
6458 |
+
{
|
6459 |
+
"epoch": 0.9832044788056519,
|
6460 |
+
"grad_norm": 0.0028154279571026564,
|
6461 |
+
"learning_rate": 7.92239759864255e-08,
|
6462 |
+
"loss": 0.0001,
|
6463 |
+
"step": 922
|
6464 |
+
},
|
6465 |
+
{
|
6466 |
+
"epoch": 0.9842708611037057,
|
6467 |
+
"grad_norm": 0.009455343708395958,
|
6468 |
+
"learning_rate": 6.901523551424705e-08,
|
6469 |
+
"loss": 0.0002,
|
6470 |
+
"step": 923
|
6471 |
+
},
|
6472 |
+
{
|
6473 |
+
"epoch": 0.9853372434017595,
|
6474 |
+
"grad_norm": 0.0037830194924026728,
|
6475 |
+
"learning_rate": 5.950992109779452e-08,
|
6476 |
+
"loss": 0.0001,
|
6477 |
+
"step": 924
|
6478 |
+
},
|
6479 |
+
{
|
6480 |
+
"epoch": 0.9864036256998134,
|
6481 |
+
"grad_norm": 0.003709913929924369,
|
6482 |
+
"learning_rate": 5.070816664762834e-08,
|
6483 |
+
"loss": 0.0001,
|
6484 |
+
"step": 925
|
6485 |
+
},
|
6486 |
+
{
|
6487 |
+
"epoch": 0.9874700079978672,
|
6488 |
+
"grad_norm": 0.012310587801039219,
|
6489 |
+
"learning_rate": 4.261009616257639e-08,
|
6490 |
+
"loss": 0.0002,
|
6491 |
+
"step": 926
|
6492 |
+
},
|
6493 |
+
{
|
6494 |
+
"epoch": 0.9885363902959211,
|
6495 |
+
"grad_norm": 0.3587518632411957,
|
6496 |
+
"learning_rate": 3.521582372797427e-08,
|
6497 |
+
"loss": 0.0017,
|
6498 |
+
"step": 927
|
6499 |
+
},
|
6500 |
+
{
|
6501 |
+
"epoch": 0.9896027725939749,
|
6502 |
+
"grad_norm": 0.10922612249851227,
|
6503 |
+
"learning_rate": 2.8525453514099965e-08,
|
6504 |
+
"loss": 0.0012,
|
6505 |
+
"step": 928
|
6506 |
+
},
|
6507 |
+
{
|
6508 |
+
"epoch": 0.9906691548920288,
|
6509 |
+
"grad_norm": 0.027902130037546158,
|
6510 |
+
"learning_rate": 2.25390797746583e-08,
|
6511 |
+
"loss": 0.0006,
|
6512 |
+
"step": 929
|
6513 |
+
},
|
6514 |
+
{
|
6515 |
+
"epoch": 0.9917355371900827,
|
6516 |
+
"grad_norm": 0.06684616953134537,
|
6517 |
+
"learning_rate": 1.725678684548204e-08,
|
6518 |
+
"loss": 0.0013,
|
6519 |
+
"step": 930
|
6520 |
+
},
|
6521 |
+
{
|
6522 |
+
"epoch": 0.9928019194881365,
|
6523 |
+
"grad_norm": 0.05900757759809494,
|
6524 |
+
"learning_rate": 1.2678649143349486e-08,
|
6525 |
+
"loss": 0.0002,
|
6526 |
+
"step": 931
|
6527 |
+
},
|
6528 |
+
{
|
6529 |
+
"epoch": 0.9938683017861903,
|
6530 |
+
"grad_norm": 0.0055292327888309956,
|
6531 |
+
"learning_rate": 8.804731164901991e-09,
|
6532 |
+
"loss": 0.0002,
|
6533 |
+
"step": 932
|
6534 |
+
},
|
6535 |
+
{
|
6536 |
+
"epoch": 0.9949346840842442,
|
6537 |
+
"grad_norm": 0.02153046801686287,
|
6538 |
+
"learning_rate": 5.6350874857724615e-09,
|
6539 |
+
"loss": 0.0005,
|
6540 |
+
"step": 933
|
6541 |
+
},
|
6542 |
+
{
|
6543 |
+
"epoch": 0.996001066382298,
|
6544 |
+
"grad_norm": 0.30450108647346497,
|
6545 |
+
"learning_rate": 3.1697627597970793e-09,
|
6546 |
+
"loss": 0.0061,
|
6547 |
+
"step": 934
|
6548 |
+
},
|
6549 |
+
{
|
6550 |
+
"epoch": 0.9970674486803519,
|
6551 |
+
"grad_norm": 0.0034953670110553503,
|
6552 |
+
"learning_rate": 1.408791718382485e-09,
|
6553 |
+
"loss": 0.0001,
|
6554 |
+
"step": 935
|
6555 |
+
},
|
6556 |
+
{
|
6557 |
+
"epoch": 0.9981338309784058,
|
6558 |
+
"grad_norm": 0.007382003124803305,
|
6559 |
+
"learning_rate": 3.5219917003948e-10,
|
6560 |
+
"loss": 0.0002,
|
6561 |
+
"step": 936
|
6562 |
+
},
|
6563 |
+
{
|
6564 |
+
"epoch": 0.9992002132764596,
|
6565 |
+
"grad_norm": 0.01729605346918106,
|
6566 |
+
"learning_rate": 0.0,
|
6567 |
+
"loss": 0.0004,
|
6568 |
+
"step": 937
|
6569 |
}
|
6570 |
],
|
6571 |
"logging_steps": 1,
|
|
|
6580 |
"should_evaluate": false,
|
6581 |
"should_log": false,
|
6582 |
"should_save": true,
|
6583 |
+
"should_training_stop": true
|
6584 |
},
|
6585 |
"attributes": {}
|
6586 |
}
|
6587 |
},
|
6588 |
+
"total_flos": 4.306063357850419e+17,
|
6589 |
"train_batch_size": 4,
|
6590 |
"trial_name": null,
|
6591 |
"trial_params": null
|