Upload 12 files
Browse files- config.json +1 -1
- optimizer.pt +3 -0
- rng_state.pth +3 -0
- scheduler.pt +3 -0
- trainer_state.json +1013 -0
config.json
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
{
|
| 2 |
-
"_name_or_path": "/
|
| 3 |
"activation_dropout": 0.1,
|
| 4 |
"apply_spec_augment": true,
|
| 5 |
"architectures": [
|
|
|
|
| 1 |
{
|
| 2 |
+
"_name_or_path": "microsoft/speecht5_tts",
|
| 3 |
"activation_dropout": 0.1,
|
| 4 |
"apply_spec_augment": true,
|
| 5 |
"architectures": [
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:deb19bba32a1b2d1a56502bcc9dd4143436fd107087be9a75f7453e3da600960
|
| 3 |
+
size 1155772233
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1a38b5cc9b19a579f4dc951803470e10b4110c1c6038f3047a16d3d014b7550c
|
| 3 |
+
size 14244
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4783d1aad67332b882158dbf71f702c2c19d09ed6c5f6dbf2a69dcc433c36f68
|
| 3 |
+
size 1064
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1013 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.3645675480365753,
|
| 3 |
+
"best_model_checkpoint": "./speecht5_tts_lg/checkpoint-4000",
|
| 4 |
+
"epoch": 59.25925925925926,
|
| 5 |
+
"eval_steps": 1000,
|
| 6 |
+
"global_step": 4000,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.37,
|
| 13 |
+
"learning_rate": 4.800000000000001e-07,
|
| 14 |
+
"loss": 0.8112,
|
| 15 |
+
"step": 25
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.74,
|
| 19 |
+
"learning_rate": 9.800000000000001e-07,
|
| 20 |
+
"loss": 0.8001,
|
| 21 |
+
"step": 50
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 1.11,
|
| 25 |
+
"learning_rate": 1.48e-06,
|
| 26 |
+
"loss": 0.7423,
|
| 27 |
+
"step": 75
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 1.48,
|
| 31 |
+
"learning_rate": 1.98e-06,
|
| 32 |
+
"loss": 0.7274,
|
| 33 |
+
"step": 100
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 1.85,
|
| 37 |
+
"learning_rate": 2.4800000000000004e-06,
|
| 38 |
+
"loss": 0.687,
|
| 39 |
+
"step": 125
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 2.22,
|
| 43 |
+
"learning_rate": 2.9800000000000003e-06,
|
| 44 |
+
"loss": 0.6456,
|
| 45 |
+
"step": 150
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 2.59,
|
| 49 |
+
"learning_rate": 3.48e-06,
|
| 50 |
+
"loss": 0.656,
|
| 51 |
+
"step": 175
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 2.96,
|
| 55 |
+
"learning_rate": 3.980000000000001e-06,
|
| 56 |
+
"loss": 0.6285,
|
| 57 |
+
"step": 200
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 3.33,
|
| 61 |
+
"learning_rate": 4.48e-06,
|
| 62 |
+
"loss": 0.6037,
|
| 63 |
+
"step": 225
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 3.7,
|
| 67 |
+
"learning_rate": 4.980000000000001e-06,
|
| 68 |
+
"loss": 0.5912,
|
| 69 |
+
"step": 250
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 4.07,
|
| 73 |
+
"learning_rate": 5.480000000000001e-06,
|
| 74 |
+
"loss": 0.5756,
|
| 75 |
+
"step": 275
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 4.44,
|
| 79 |
+
"learning_rate": 5.98e-06,
|
| 80 |
+
"loss": 0.535,
|
| 81 |
+
"step": 300
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 4.81,
|
| 85 |
+
"learning_rate": 6.480000000000001e-06,
|
| 86 |
+
"loss": 0.5014,
|
| 87 |
+
"step": 325
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 5.19,
|
| 91 |
+
"learning_rate": 6.98e-06,
|
| 92 |
+
"loss": 0.4977,
|
| 93 |
+
"step": 350
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 5.56,
|
| 97 |
+
"learning_rate": 7.48e-06,
|
| 98 |
+
"loss": 0.4805,
|
| 99 |
+
"step": 375
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 5.93,
|
| 103 |
+
"learning_rate": 7.980000000000002e-06,
|
| 104 |
+
"loss": 0.4732,
|
| 105 |
+
"step": 400
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 6.3,
|
| 109 |
+
"learning_rate": 8.48e-06,
|
| 110 |
+
"loss": 0.4809,
|
| 111 |
+
"step": 425
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 6.67,
|
| 115 |
+
"learning_rate": 8.98e-06,
|
| 116 |
+
"loss": 0.4678,
|
| 117 |
+
"step": 450
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 7.04,
|
| 121 |
+
"learning_rate": 9.48e-06,
|
| 122 |
+
"loss": 0.4584,
|
| 123 |
+
"step": 475
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 7.41,
|
| 127 |
+
"learning_rate": 9.980000000000001e-06,
|
| 128 |
+
"loss": 0.458,
|
| 129 |
+
"step": 500
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 7.78,
|
| 133 |
+
"learning_rate": 9.931428571428571e-06,
|
| 134 |
+
"loss": 0.4526,
|
| 135 |
+
"step": 525
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 8.15,
|
| 139 |
+
"learning_rate": 9.86e-06,
|
| 140 |
+
"loss": 0.4509,
|
| 141 |
+
"step": 550
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 8.52,
|
| 145 |
+
"learning_rate": 9.78857142857143e-06,
|
| 146 |
+
"loss": 0.4437,
|
| 147 |
+
"step": 575
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 8.89,
|
| 151 |
+
"learning_rate": 9.717142857142858e-06,
|
| 152 |
+
"loss": 0.4523,
|
| 153 |
+
"step": 600
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 9.26,
|
| 157 |
+
"learning_rate": 9.645714285714286e-06,
|
| 158 |
+
"loss": 0.4467,
|
| 159 |
+
"step": 625
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 9.63,
|
| 163 |
+
"learning_rate": 9.574285714285715e-06,
|
| 164 |
+
"loss": 0.4478,
|
| 165 |
+
"step": 650
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 10.0,
|
| 169 |
+
"learning_rate": 9.502857142857144e-06,
|
| 170 |
+
"loss": 0.4455,
|
| 171 |
+
"step": 675
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 10.37,
|
| 175 |
+
"learning_rate": 9.431428571428573e-06,
|
| 176 |
+
"loss": 0.4349,
|
| 177 |
+
"step": 700
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 10.74,
|
| 181 |
+
"learning_rate": 9.360000000000002e-06,
|
| 182 |
+
"loss": 0.4336,
|
| 183 |
+
"step": 725
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 11.11,
|
| 187 |
+
"learning_rate": 9.28857142857143e-06,
|
| 188 |
+
"loss": 0.4311,
|
| 189 |
+
"step": 750
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 11.48,
|
| 193 |
+
"learning_rate": 9.217142857142858e-06,
|
| 194 |
+
"loss": 0.4314,
|
| 195 |
+
"step": 775
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 11.85,
|
| 199 |
+
"learning_rate": 9.145714285714287e-06,
|
| 200 |
+
"loss": 0.4291,
|
| 201 |
+
"step": 800
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 12.22,
|
| 205 |
+
"learning_rate": 9.074285714285716e-06,
|
| 206 |
+
"loss": 0.4344,
|
| 207 |
+
"step": 825
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 12.59,
|
| 211 |
+
"learning_rate": 9.002857142857144e-06,
|
| 212 |
+
"loss": 0.4276,
|
| 213 |
+
"step": 850
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 12.96,
|
| 217 |
+
"learning_rate": 8.931428571428573e-06,
|
| 218 |
+
"loss": 0.4191,
|
| 219 |
+
"step": 875
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 13.33,
|
| 223 |
+
"learning_rate": 8.860000000000002e-06,
|
| 224 |
+
"loss": 0.4271,
|
| 225 |
+
"step": 900
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 13.7,
|
| 229 |
+
"learning_rate": 8.788571428571429e-06,
|
| 230 |
+
"loss": 0.4217,
|
| 231 |
+
"step": 925
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 14.07,
|
| 235 |
+
"learning_rate": 8.717142857142858e-06,
|
| 236 |
+
"loss": 0.4205,
|
| 237 |
+
"step": 950
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 14.44,
|
| 241 |
+
"learning_rate": 8.645714285714287e-06,
|
| 242 |
+
"loss": 0.4191,
|
| 243 |
+
"step": 975
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 14.81,
|
| 247 |
+
"learning_rate": 8.574285714285714e-06,
|
| 248 |
+
"loss": 0.4213,
|
| 249 |
+
"step": 1000
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 14.81,
|
| 253 |
+
"eval_loss": 0.38490030169487,
|
| 254 |
+
"eval_runtime": 5.2563,
|
| 255 |
+
"eval_samples_per_second": 45.66,
|
| 256 |
+
"eval_steps_per_second": 1.902,
|
| 257 |
+
"step": 1000
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 15.19,
|
| 261 |
+
"learning_rate": 8.502857142857143e-06,
|
| 262 |
+
"loss": 0.415,
|
| 263 |
+
"step": 1025
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 15.56,
|
| 267 |
+
"learning_rate": 8.431428571428572e-06,
|
| 268 |
+
"loss": 0.4259,
|
| 269 |
+
"step": 1050
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 15.93,
|
| 273 |
+
"learning_rate": 8.36e-06,
|
| 274 |
+
"loss": 0.4189,
|
| 275 |
+
"step": 1075
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 16.3,
|
| 279 |
+
"learning_rate": 8.288571428571429e-06,
|
| 280 |
+
"loss": 0.4171,
|
| 281 |
+
"step": 1100
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 16.67,
|
| 285 |
+
"learning_rate": 8.217142857142858e-06,
|
| 286 |
+
"loss": 0.4111,
|
| 287 |
+
"step": 1125
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 17.04,
|
| 291 |
+
"learning_rate": 8.145714285714287e-06,
|
| 292 |
+
"loss": 0.415,
|
| 293 |
+
"step": 1150
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 17.41,
|
| 297 |
+
"learning_rate": 8.074285714285714e-06,
|
| 298 |
+
"loss": 0.4078,
|
| 299 |
+
"step": 1175
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 17.78,
|
| 303 |
+
"learning_rate": 8.002857142857143e-06,
|
| 304 |
+
"loss": 0.4099,
|
| 305 |
+
"step": 1200
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 18.15,
|
| 309 |
+
"learning_rate": 7.931428571428572e-06,
|
| 310 |
+
"loss": 0.4137,
|
| 311 |
+
"step": 1225
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 18.52,
|
| 315 |
+
"learning_rate": 7.860000000000001e-06,
|
| 316 |
+
"loss": 0.4137,
|
| 317 |
+
"step": 1250
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 18.89,
|
| 321 |
+
"learning_rate": 7.788571428571428e-06,
|
| 322 |
+
"loss": 0.4147,
|
| 323 |
+
"step": 1275
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 19.26,
|
| 327 |
+
"learning_rate": 7.717142857142857e-06,
|
| 328 |
+
"loss": 0.4111,
|
| 329 |
+
"step": 1300
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 19.63,
|
| 333 |
+
"learning_rate": 7.645714285714286e-06,
|
| 334 |
+
"loss": 0.4139,
|
| 335 |
+
"step": 1325
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 20.0,
|
| 339 |
+
"learning_rate": 7.574285714285715e-06,
|
| 340 |
+
"loss": 0.4104,
|
| 341 |
+
"step": 1350
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 20.37,
|
| 345 |
+
"learning_rate": 7.502857142857144e-06,
|
| 346 |
+
"loss": 0.4131,
|
| 347 |
+
"step": 1375
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 20.74,
|
| 351 |
+
"learning_rate": 7.431428571428572e-06,
|
| 352 |
+
"loss": 0.4001,
|
| 353 |
+
"step": 1400
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 21.11,
|
| 357 |
+
"learning_rate": 7.360000000000001e-06,
|
| 358 |
+
"loss": 0.4021,
|
| 359 |
+
"step": 1425
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 21.48,
|
| 363 |
+
"learning_rate": 7.28857142857143e-06,
|
| 364 |
+
"loss": 0.4009,
|
| 365 |
+
"step": 1450
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 21.85,
|
| 369 |
+
"learning_rate": 7.217142857142858e-06,
|
| 370 |
+
"loss": 0.4036,
|
| 371 |
+
"step": 1475
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 22.22,
|
| 375 |
+
"learning_rate": 7.145714285714286e-06,
|
| 376 |
+
"loss": 0.4068,
|
| 377 |
+
"step": 1500
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 22.59,
|
| 381 |
+
"learning_rate": 7.074285714285715e-06,
|
| 382 |
+
"loss": 0.4,
|
| 383 |
+
"step": 1525
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 22.96,
|
| 387 |
+
"learning_rate": 7.002857142857143e-06,
|
| 388 |
+
"loss": 0.3998,
|
| 389 |
+
"step": 1550
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 23.33,
|
| 393 |
+
"learning_rate": 6.931428571428572e-06,
|
| 394 |
+
"loss": 0.4017,
|
| 395 |
+
"step": 1575
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 23.7,
|
| 399 |
+
"learning_rate": 6.860000000000001e-06,
|
| 400 |
+
"loss": 0.3986,
|
| 401 |
+
"step": 1600
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 24.07,
|
| 405 |
+
"learning_rate": 6.7885714285714286e-06,
|
| 406 |
+
"loss": 0.404,
|
| 407 |
+
"step": 1625
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 24.44,
|
| 411 |
+
"learning_rate": 6.7171428571428576e-06,
|
| 412 |
+
"loss": 0.3927,
|
| 413 |
+
"step": 1650
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 24.81,
|
| 417 |
+
"learning_rate": 6.645714285714287e-06,
|
| 418 |
+
"loss": 0.4004,
|
| 419 |
+
"step": 1675
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 25.19,
|
| 423 |
+
"learning_rate": 6.574285714285716e-06,
|
| 424 |
+
"loss": 0.398,
|
| 425 |
+
"step": 1700
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 25.56,
|
| 429 |
+
"learning_rate": 6.502857142857143e-06,
|
| 430 |
+
"loss": 0.3982,
|
| 431 |
+
"step": 1725
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 25.93,
|
| 435 |
+
"learning_rate": 6.431428571428572e-06,
|
| 436 |
+
"loss": 0.3975,
|
| 437 |
+
"step": 1750
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 26.3,
|
| 441 |
+
"learning_rate": 6.360000000000001e-06,
|
| 442 |
+
"loss": 0.3952,
|
| 443 |
+
"step": 1775
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 26.67,
|
| 447 |
+
"learning_rate": 6.288571428571429e-06,
|
| 448 |
+
"loss": 0.3964,
|
| 449 |
+
"step": 1800
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 27.04,
|
| 453 |
+
"learning_rate": 6.217142857142857e-06,
|
| 454 |
+
"loss": 0.3983,
|
| 455 |
+
"step": 1825
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 27.41,
|
| 459 |
+
"learning_rate": 6.145714285714286e-06,
|
| 460 |
+
"loss": 0.3937,
|
| 461 |
+
"step": 1850
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 27.78,
|
| 465 |
+
"learning_rate": 6.0742857142857145e-06,
|
| 466 |
+
"loss": 0.398,
|
| 467 |
+
"step": 1875
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 28.15,
|
| 471 |
+
"learning_rate": 6.0028571428571435e-06,
|
| 472 |
+
"loss": 0.398,
|
| 473 |
+
"step": 1900
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 28.52,
|
| 477 |
+
"learning_rate": 5.9314285714285725e-06,
|
| 478 |
+
"loss": 0.3976,
|
| 479 |
+
"step": 1925
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 28.89,
|
| 483 |
+
"learning_rate": 5.86e-06,
|
| 484 |
+
"loss": 0.3945,
|
| 485 |
+
"step": 1950
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 29.26,
|
| 489 |
+
"learning_rate": 5.791428571428572e-06,
|
| 490 |
+
"loss": 0.3909,
|
| 491 |
+
"step": 1975
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 29.63,
|
| 495 |
+
"learning_rate": 5.72e-06,
|
| 496 |
+
"loss": 0.392,
|
| 497 |
+
"step": 2000
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 29.63,
|
| 501 |
+
"eval_loss": 0.3671075701713562,
|
| 502 |
+
"eval_runtime": 5.4439,
|
| 503 |
+
"eval_samples_per_second": 44.086,
|
| 504 |
+
"eval_steps_per_second": 1.837,
|
| 505 |
+
"step": 2000
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 30.0,
|
| 509 |
+
"learning_rate": 5.6485714285714285e-06,
|
| 510 |
+
"loss": 0.3933,
|
| 511 |
+
"step": 2025
|
| 512 |
+
},
|
| 513 |
+
{
|
| 514 |
+
"epoch": 30.37,
|
| 515 |
+
"learning_rate": 5.5771428571428575e-06,
|
| 516 |
+
"loss": 0.394,
|
| 517 |
+
"step": 2050
|
| 518 |
+
},
|
| 519 |
+
{
|
| 520 |
+
"epoch": 30.74,
|
| 521 |
+
"learning_rate": 5.5057142857142865e-06,
|
| 522 |
+
"loss": 0.3975,
|
| 523 |
+
"step": 2075
|
| 524 |
+
},
|
| 525 |
+
{
|
| 526 |
+
"epoch": 31.11,
|
| 527 |
+
"learning_rate": 5.4342857142857155e-06,
|
| 528 |
+
"loss": 0.3916,
|
| 529 |
+
"step": 2100
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 31.48,
|
| 533 |
+
"learning_rate": 5.362857142857143e-06,
|
| 534 |
+
"loss": 0.3991,
|
| 535 |
+
"step": 2125
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 31.85,
|
| 539 |
+
"learning_rate": 5.291428571428572e-06,
|
| 540 |
+
"loss": 0.3934,
|
| 541 |
+
"step": 2150
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 32.22,
|
| 545 |
+
"learning_rate": 5.220000000000001e-06,
|
| 546 |
+
"loss": 0.3944,
|
| 547 |
+
"step": 2175
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 32.59,
|
| 551 |
+
"learning_rate": 5.14857142857143e-06,
|
| 552 |
+
"loss": 0.3946,
|
| 553 |
+
"step": 2200
|
| 554 |
+
},
|
| 555 |
+
{
|
| 556 |
+
"epoch": 32.96,
|
| 557 |
+
"learning_rate": 5.077142857142857e-06,
|
| 558 |
+
"loss": 0.3907,
|
| 559 |
+
"step": 2225
|
| 560 |
+
},
|
| 561 |
+
{
|
| 562 |
+
"epoch": 33.33,
|
| 563 |
+
"learning_rate": 5.005714285714286e-06,
|
| 564 |
+
"loss": 0.3926,
|
| 565 |
+
"step": 2250
|
| 566 |
+
},
|
| 567 |
+
{
|
| 568 |
+
"epoch": 33.7,
|
| 569 |
+
"learning_rate": 4.934285714285715e-06,
|
| 570 |
+
"loss": 0.3954,
|
| 571 |
+
"step": 2275
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 34.07,
|
| 575 |
+
"learning_rate": 4.862857142857143e-06,
|
| 576 |
+
"loss": 0.3994,
|
| 577 |
+
"step": 2300
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 34.44,
|
| 581 |
+
"learning_rate": 4.7914285714285715e-06,
|
| 582 |
+
"loss": 0.3954,
|
| 583 |
+
"step": 2325
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 34.81,
|
| 587 |
+
"learning_rate": 4.7200000000000005e-06,
|
| 588 |
+
"loss": 0.3916,
|
| 589 |
+
"step": 2350
|
| 590 |
+
},
|
| 591 |
+
{
|
| 592 |
+
"epoch": 35.19,
|
| 593 |
+
"learning_rate": 4.648571428571429e-06,
|
| 594 |
+
"loss": 0.3904,
|
| 595 |
+
"step": 2375
|
| 596 |
+
},
|
| 597 |
+
{
|
| 598 |
+
"epoch": 35.56,
|
| 599 |
+
"learning_rate": 4.577142857142858e-06,
|
| 600 |
+
"loss": 0.3939,
|
| 601 |
+
"step": 2400
|
| 602 |
+
},
|
| 603 |
+
{
|
| 604 |
+
"epoch": 35.93,
|
| 605 |
+
"learning_rate": 4.505714285714286e-06,
|
| 606 |
+
"loss": 0.3867,
|
| 607 |
+
"step": 2425
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"epoch": 36.3,
|
| 611 |
+
"learning_rate": 4.434285714285715e-06,
|
| 612 |
+
"loss": 0.3843,
|
| 613 |
+
"step": 2450
|
| 614 |
+
},
|
| 615 |
+
{
|
| 616 |
+
"epoch": 36.67,
|
| 617 |
+
"learning_rate": 4.362857142857143e-06,
|
| 618 |
+
"loss": 0.389,
|
| 619 |
+
"step": 2475
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 37.04,
|
| 623 |
+
"learning_rate": 4.291428571428572e-06,
|
| 624 |
+
"loss": 0.3915,
|
| 625 |
+
"step": 2500
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 37.41,
|
| 629 |
+
"learning_rate": 4.22e-06,
|
| 630 |
+
"loss": 0.3883,
|
| 631 |
+
"step": 2525
|
| 632 |
+
},
|
| 633 |
+
{
|
| 634 |
+
"epoch": 37.78,
|
| 635 |
+
"learning_rate": 4.148571428571429e-06,
|
| 636 |
+
"loss": 0.3886,
|
| 637 |
+
"step": 2550
|
| 638 |
+
},
|
| 639 |
+
{
|
| 640 |
+
"epoch": 38.15,
|
| 641 |
+
"learning_rate": 4.0771428571428574e-06,
|
| 642 |
+
"loss": 0.3916,
|
| 643 |
+
"step": 2575
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 38.52,
|
| 647 |
+
"learning_rate": 4.0057142857142864e-06,
|
| 648 |
+
"loss": 0.386,
|
| 649 |
+
"step": 2600
|
| 650 |
+
},
|
| 651 |
+
{
|
| 652 |
+
"epoch": 38.89,
|
| 653 |
+
"learning_rate": 3.934285714285715e-06,
|
| 654 |
+
"loss": 0.3877,
|
| 655 |
+
"step": 2625
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 39.26,
|
| 659 |
+
"learning_rate": 3.862857142857143e-06,
|
| 660 |
+
"loss": 0.388,
|
| 661 |
+
"step": 2650
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 39.63,
|
| 665 |
+
"learning_rate": 3.7914285714285722e-06,
|
| 666 |
+
"loss": 0.3855,
|
| 667 |
+
"step": 2675
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 40.0,
|
| 671 |
+
"learning_rate": 3.7200000000000004e-06,
|
| 672 |
+
"loss": 0.3846,
|
| 673 |
+
"step": 2700
|
| 674 |
+
},
|
| 675 |
+
{
|
| 676 |
+
"epoch": 40.37,
|
| 677 |
+
"learning_rate": 3.648571428571429e-06,
|
| 678 |
+
"loss": 0.3861,
|
| 679 |
+
"step": 2725
|
| 680 |
+
},
|
| 681 |
+
{
|
| 682 |
+
"epoch": 40.74,
|
| 683 |
+
"learning_rate": 3.5771428571428576e-06,
|
| 684 |
+
"loss": 0.3914,
|
| 685 |
+
"step": 2750
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 41.11,
|
| 689 |
+
"learning_rate": 3.505714285714286e-06,
|
| 690 |
+
"loss": 0.3856,
|
| 691 |
+
"step": 2775
|
| 692 |
+
},
|
| 693 |
+
{
|
| 694 |
+
"epoch": 41.48,
|
| 695 |
+
"learning_rate": 3.4342857142857143e-06,
|
| 696 |
+
"loss": 0.3893,
|
| 697 |
+
"step": 2800
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"epoch": 41.85,
|
| 701 |
+
"learning_rate": 3.3628571428571433e-06,
|
| 702 |
+
"loss": 0.3883,
|
| 703 |
+
"step": 2825
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 42.22,
|
| 707 |
+
"learning_rate": 3.2914285714285715e-06,
|
| 708 |
+
"loss": 0.3854,
|
| 709 |
+
"step": 2850
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 42.59,
|
| 713 |
+
"learning_rate": 3.2200000000000005e-06,
|
| 714 |
+
"loss": 0.3873,
|
| 715 |
+
"step": 2875
|
| 716 |
+
},
|
| 717 |
+
{
|
| 718 |
+
"epoch": 42.96,
|
| 719 |
+
"learning_rate": 3.1485714285714287e-06,
|
| 720 |
+
"loss": 0.3871,
|
| 721 |
+
"step": 2900
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 43.33,
|
| 725 |
+
"learning_rate": 3.0771428571428573e-06,
|
| 726 |
+
"loss": 0.3859,
|
| 727 |
+
"step": 2925
|
| 728 |
+
},
|
| 729 |
+
{
|
| 730 |
+
"epoch": 43.7,
|
| 731 |
+
"learning_rate": 3.005714285714286e-06,
|
| 732 |
+
"loss": 0.3857,
|
| 733 |
+
"step": 2950
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 44.07,
|
| 737 |
+
"learning_rate": 2.9342857142857144e-06,
|
| 738 |
+
"loss": 0.3868,
|
| 739 |
+
"step": 2975
|
| 740 |
+
},
|
| 741 |
+
{
|
| 742 |
+
"epoch": 44.44,
|
| 743 |
+
"learning_rate": 2.8628571428571435e-06,
|
| 744 |
+
"loss": 0.3884,
|
| 745 |
+
"step": 3000
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 44.44,
|
| 749 |
+
"eval_loss": 0.3664553463459015,
|
| 750 |
+
"eval_runtime": 5.1439,
|
| 751 |
+
"eval_samples_per_second": 46.657,
|
| 752 |
+
"eval_steps_per_second": 1.944,
|
| 753 |
+
"step": 3000
|
| 754 |
+
},
|
| 755 |
+
{
|
| 756 |
+
"epoch": 44.81,
|
| 757 |
+
"learning_rate": 2.7914285714285716e-06,
|
| 758 |
+
"loss": 0.3842,
|
| 759 |
+
"step": 3025
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 45.19,
|
| 763 |
+
"learning_rate": 2.7200000000000002e-06,
|
| 764 |
+
"loss": 0.3859,
|
| 765 |
+
"step": 3050
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 45.56,
|
| 769 |
+
"learning_rate": 2.648571428571429e-06,
|
| 770 |
+
"loss": 0.3932,
|
| 771 |
+
"step": 3075
|
| 772 |
+
},
|
| 773 |
+
{
|
| 774 |
+
"epoch": 45.93,
|
| 775 |
+
"learning_rate": 2.5771428571428574e-06,
|
| 776 |
+
"loss": 0.3852,
|
| 777 |
+
"step": 3100
|
| 778 |
+
},
|
| 779 |
+
{
|
| 780 |
+
"epoch": 46.3,
|
| 781 |
+
"learning_rate": 2.5057142857142856e-06,
|
| 782 |
+
"loss": 0.3872,
|
| 783 |
+
"step": 3125
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 46.67,
|
| 787 |
+
"learning_rate": 2.4342857142857146e-06,
|
| 788 |
+
"loss": 0.3843,
|
| 789 |
+
"step": 3150
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 47.04,
|
| 793 |
+
"learning_rate": 2.362857142857143e-06,
|
| 794 |
+
"loss": 0.3927,
|
| 795 |
+
"step": 3175
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 47.41,
|
| 799 |
+
"learning_rate": 2.2914285714285718e-06,
|
| 800 |
+
"loss": 0.3871,
|
| 801 |
+
"step": 3200
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 47.78,
|
| 805 |
+
"learning_rate": 2.2200000000000003e-06,
|
| 806 |
+
"loss": 0.3819,
|
| 807 |
+
"step": 3225
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 48.15,
|
| 811 |
+
"learning_rate": 2.148571428571429e-06,
|
| 812 |
+
"loss": 0.3871,
|
| 813 |
+
"step": 3250
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"epoch": 48.52,
|
| 817 |
+
"learning_rate": 2.077142857142857e-06,
|
| 818 |
+
"loss": 0.3851,
|
| 819 |
+
"step": 3275
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 48.89,
|
| 823 |
+
"learning_rate": 2.0057142857142857e-06,
|
| 824 |
+
"loss": 0.3841,
|
| 825 |
+
"step": 3300
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 49.26,
|
| 829 |
+
"learning_rate": 1.9342857142857143e-06,
|
| 830 |
+
"loss": 0.3814,
|
| 831 |
+
"step": 3325
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 49.63,
|
| 835 |
+
"learning_rate": 1.8628571428571429e-06,
|
| 836 |
+
"loss": 0.3846,
|
| 837 |
+
"step": 3350
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 50.0,
|
| 841 |
+
"learning_rate": 1.7914285714285715e-06,
|
| 842 |
+
"loss": 0.3811,
|
| 843 |
+
"step": 3375
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 50.37,
|
| 847 |
+
"learning_rate": 1.72e-06,
|
| 848 |
+
"loss": 0.3854,
|
| 849 |
+
"step": 3400
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 50.74,
|
| 853 |
+
"learning_rate": 1.6485714285714289e-06,
|
| 854 |
+
"loss": 0.3821,
|
| 855 |
+
"step": 3425
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 51.11,
|
| 859 |
+
"learning_rate": 1.5771428571428574e-06,
|
| 860 |
+
"loss": 0.3818,
|
| 861 |
+
"step": 3450
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"epoch": 51.48,
|
| 865 |
+
"learning_rate": 1.5057142857142858e-06,
|
| 866 |
+
"loss": 0.3895,
|
| 867 |
+
"step": 3475
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 51.85,
|
| 871 |
+
"learning_rate": 1.4342857142857144e-06,
|
| 872 |
+
"loss": 0.3829,
|
| 873 |
+
"step": 3500
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 52.22,
|
| 877 |
+
"learning_rate": 1.362857142857143e-06,
|
| 878 |
+
"loss": 0.385,
|
| 879 |
+
"step": 3525
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 52.59,
|
| 883 |
+
"learning_rate": 1.2914285714285716e-06,
|
| 884 |
+
"loss": 0.3834,
|
| 885 |
+
"step": 3550
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 52.96,
|
| 889 |
+
"learning_rate": 1.2200000000000002e-06,
|
| 890 |
+
"loss": 0.3833,
|
| 891 |
+
"step": 3575
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 53.33,
|
| 895 |
+
"learning_rate": 1.1485714285714286e-06,
|
| 896 |
+
"loss": 0.3797,
|
| 897 |
+
"step": 3600
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 53.7,
|
| 901 |
+
"learning_rate": 1.0771428571428574e-06,
|
| 902 |
+
"loss": 0.3874,
|
| 903 |
+
"step": 3625
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 54.07,
|
| 907 |
+
"learning_rate": 1.0057142857142857e-06,
|
| 908 |
+
"loss": 0.3785,
|
| 909 |
+
"step": 3650
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 54.44,
|
| 913 |
+
"learning_rate": 9.342857142857144e-07,
|
| 914 |
+
"loss": 0.3792,
|
| 915 |
+
"step": 3675
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"epoch": 54.81,
|
| 919 |
+
"learning_rate": 8.628571428571429e-07,
|
| 920 |
+
"loss": 0.3814,
|
| 921 |
+
"step": 3700
|
| 922 |
+
},
|
| 923 |
+
{
|
| 924 |
+
"epoch": 55.19,
|
| 925 |
+
"learning_rate": 7.914285714285715e-07,
|
| 926 |
+
"loss": 0.3815,
|
| 927 |
+
"step": 3725
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 55.56,
|
| 931 |
+
"learning_rate": 7.2e-07,
|
| 932 |
+
"loss": 0.3828,
|
| 933 |
+
"step": 3750
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 55.93,
|
| 937 |
+
"learning_rate": 6.485714285714287e-07,
|
| 938 |
+
"loss": 0.386,
|
| 939 |
+
"step": 3775
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 56.3,
|
| 943 |
+
"learning_rate": 5.771428571428572e-07,
|
| 944 |
+
"loss": 0.3846,
|
| 945 |
+
"step": 3800
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 56.67,
|
| 949 |
+
"learning_rate": 5.057142857142858e-07,
|
| 950 |
+
"loss": 0.3793,
|
| 951 |
+
"step": 3825
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 57.04,
|
| 955 |
+
"learning_rate": 4.342857142857143e-07,
|
| 956 |
+
"loss": 0.3814,
|
| 957 |
+
"step": 3850
|
| 958 |
+
},
|
| 959 |
+
{
|
| 960 |
+
"epoch": 57.41,
|
| 961 |
+
"learning_rate": 3.6285714285714283e-07,
|
| 962 |
+
"loss": 0.3808,
|
| 963 |
+
"step": 3875
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 57.78,
|
| 967 |
+
"learning_rate": 2.914285714285715e-07,
|
| 968 |
+
"loss": 0.3798,
|
| 969 |
+
"step": 3900
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 58.15,
|
| 973 |
+
"learning_rate": 2.2e-07,
|
| 974 |
+
"loss": 0.3852,
|
| 975 |
+
"step": 3925
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 58.52,
|
| 979 |
+
"learning_rate": 1.4857142857142857e-07,
|
| 980 |
+
"loss": 0.3812,
|
| 981 |
+
"step": 3950
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 58.89,
|
| 985 |
+
"learning_rate": 7.714285714285715e-08,
|
| 986 |
+
"loss": 0.3793,
|
| 987 |
+
"step": 3975
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"epoch": 59.26,
|
| 991 |
+
"learning_rate": 5.714285714285715e-09,
|
| 992 |
+
"loss": 0.3879,
|
| 993 |
+
"step": 4000
|
| 994 |
+
},
|
| 995 |
+
{
|
| 996 |
+
"epoch": 59.26,
|
| 997 |
+
"eval_loss": 0.3645675480365753,
|
| 998 |
+
"eval_runtime": 6.1965,
|
| 999 |
+
"eval_samples_per_second": 38.732,
|
| 1000 |
+
"eval_steps_per_second": 1.614,
|
| 1001 |
+
"step": 4000
|
| 1002 |
+
}
|
| 1003 |
+
],
|
| 1004 |
+
"logging_steps": 25,
|
| 1005 |
+
"max_steps": 4000,
|
| 1006 |
+
"num_input_tokens_seen": 0,
|
| 1007 |
+
"num_train_epochs": 60,
|
| 1008 |
+
"save_steps": 1000,
|
| 1009 |
+
"total_flos": 1.0116465250588488e+16,
|
| 1010 |
+
"train_batch_size": 8,
|
| 1011 |
+
"trial_name": null,
|
| 1012 |
+
"trial_params": null
|
| 1013 |
+
}
|