Training in progress, step 15000, checkpoint
Browse files- last-checkpoint/global_step15000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt +3 -0
- last-checkpoint/global_step15000/mp_rank_00_model_states.pt +3 -0
- last-checkpoint/latest +1 -1
- last-checkpoint/model.safetensors +1 -1
- last-checkpoint/rng_state.pth +1 -1
- last-checkpoint/scheduler.pt +1 -1
- last-checkpoint/trainer_state.json +1453 -3
last-checkpoint/global_step15000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3334ef706136b1577dac6b2d789ff85be8781b111711dd2658201bac647a4ef0
|
| 3 |
+
size 5117197020
|
last-checkpoint/global_step15000/mp_rank_00_model_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1a4a5ec8e6dfded8b6bc8f4b18edbf43f408bd8958cf9a47572e22bc2c25d9ef
|
| 3 |
+
size 859127504
|
last-checkpoint/latest
CHANGED
|
@@ -1 +1 @@
|
|
| 1 |
-
|
|
|
|
| 1 |
+
global_step15000
|
last-checkpoint/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 962205216
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4b11d79fb3db23754b444fc4964b7982434af7a97aa600fcdcde4d9a8a9b3f21
|
| 3 |
size 962205216
|
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 sha256:63340953913b9f23b1567a89f7ad34314d1dcb47b9a43157322f84f789423e67
|
| 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 sha256:6b94442b72559c9262bb7b2684827bc59deb41027ddc14af8d9ffecb8119b4aa
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": 83.80009429514381,
|
| 3 |
"best_model_checkpoint": "./iteboshi_temp/checkpoint-10000",
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 1000,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -2907,6 +2907,1456 @@
|
|
| 2907 |
"eval_steps_per_second": 1.545,
|
| 2908 |
"eval_wer": 83.80009429514381,
|
| 2909 |
"step": 10000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2910 |
}
|
| 2911 |
],
|
| 2912 |
"logging_steps": 25,
|
|
@@ -2926,7 +4376,7 @@
|
|
| 2926 |
"attributes": {}
|
| 2927 |
}
|
| 2928 |
},
|
| 2929 |
-
"total_flos":
|
| 2930 |
"train_batch_size": 4,
|
| 2931 |
"trial_name": null,
|
| 2932 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": 83.80009429514381,
|
| 3 |
"best_model_checkpoint": "./iteboshi_temp/checkpoint-10000",
|
| 4 |
+
"epoch": 16.519823788546255,
|
| 5 |
"eval_steps": 1000,
|
| 6 |
+
"global_step": 15000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 2907 |
"eval_steps_per_second": 1.545,
|
| 2908 |
"eval_wer": 83.80009429514381,
|
| 2909 |
"step": 10000
|
| 2910 |
+
},
|
| 2911 |
+
{
|
| 2912 |
+
"epoch": 11.040748898678414,
|
| 2913 |
+
"grad_norm": 0.20153629779815674,
|
| 2914 |
+
"learning_rate": 1.0230769230769231e-05,
|
| 2915 |
+
"loss": 0.0168,
|
| 2916 |
+
"step": 10025
|
| 2917 |
+
},
|
| 2918 |
+
{
|
| 2919 |
+
"epoch": 11.068281938325992,
|
| 2920 |
+
"grad_norm": 0.10129767656326294,
|
| 2921 |
+
"learning_rate": 1.0205128205128205e-05,
|
| 2922 |
+
"loss": 0.0166,
|
| 2923 |
+
"step": 10050
|
| 2924 |
+
},
|
| 2925 |
+
{
|
| 2926 |
+
"epoch": 11.095814977973568,
|
| 2927 |
+
"grad_norm": 0.18236328661441803,
|
| 2928 |
+
"learning_rate": 1.017948717948718e-05,
|
| 2929 |
+
"loss": 0.0207,
|
| 2930 |
+
"step": 10075
|
| 2931 |
+
},
|
| 2932 |
+
{
|
| 2933 |
+
"epoch": 11.123348017621145,
|
| 2934 |
+
"grad_norm": 0.1023649126291275,
|
| 2935 |
+
"learning_rate": 1.0153846153846154e-05,
|
| 2936 |
+
"loss": 0.0151,
|
| 2937 |
+
"step": 10100
|
| 2938 |
+
},
|
| 2939 |
+
{
|
| 2940 |
+
"epoch": 11.150881057268723,
|
| 2941 |
+
"grad_norm": 0.2005663514137268,
|
| 2942 |
+
"learning_rate": 1.012820512820513e-05,
|
| 2943 |
+
"loss": 0.0139,
|
| 2944 |
+
"step": 10125
|
| 2945 |
+
},
|
| 2946 |
+
{
|
| 2947 |
+
"epoch": 11.178414096916299,
|
| 2948 |
+
"grad_norm": 0.5880993008613586,
|
| 2949 |
+
"learning_rate": 1.0102564102564103e-05,
|
| 2950 |
+
"loss": 0.0166,
|
| 2951 |
+
"step": 10150
|
| 2952 |
+
},
|
| 2953 |
+
{
|
| 2954 |
+
"epoch": 11.205947136563877,
|
| 2955 |
+
"grad_norm": 0.18318556249141693,
|
| 2956 |
+
"learning_rate": 1.0076923076923078e-05,
|
| 2957 |
+
"loss": 0.0147,
|
| 2958 |
+
"step": 10175
|
| 2959 |
+
},
|
| 2960 |
+
{
|
| 2961 |
+
"epoch": 11.233480176211454,
|
| 2962 |
+
"grad_norm": 0.10596877336502075,
|
| 2963 |
+
"learning_rate": 1.0051282051282052e-05,
|
| 2964 |
+
"loss": 0.0186,
|
| 2965 |
+
"step": 10200
|
| 2966 |
+
},
|
| 2967 |
+
{
|
| 2968 |
+
"epoch": 11.26101321585903,
|
| 2969 |
+
"grad_norm": 0.21908675134181976,
|
| 2970 |
+
"learning_rate": 1.0025641025641027e-05,
|
| 2971 |
+
"loss": 0.0167,
|
| 2972 |
+
"step": 10225
|
| 2973 |
+
},
|
| 2974 |
+
{
|
| 2975 |
+
"epoch": 11.288546255506608,
|
| 2976 |
+
"grad_norm": 0.3086978495121002,
|
| 2977 |
+
"learning_rate": 1e-05,
|
| 2978 |
+
"loss": 0.0155,
|
| 2979 |
+
"step": 10250
|
| 2980 |
+
},
|
| 2981 |
+
{
|
| 2982 |
+
"epoch": 11.316079295154186,
|
| 2983 |
+
"grad_norm": 0.09563290327787399,
|
| 2984 |
+
"learning_rate": 9.974358974358974e-06,
|
| 2985 |
+
"loss": 0.0201,
|
| 2986 |
+
"step": 10275
|
| 2987 |
+
},
|
| 2988 |
+
{
|
| 2989 |
+
"epoch": 11.343612334801762,
|
| 2990 |
+
"grad_norm": 0.44515764713287354,
|
| 2991 |
+
"learning_rate": 9.94871794871795e-06,
|
| 2992 |
+
"loss": 0.0168,
|
| 2993 |
+
"step": 10300
|
| 2994 |
+
},
|
| 2995 |
+
{
|
| 2996 |
+
"epoch": 11.37114537444934,
|
| 2997 |
+
"grad_norm": 0.3357783854007721,
|
| 2998 |
+
"learning_rate": 9.923076923076923e-06,
|
| 2999 |
+
"loss": 0.0144,
|
| 3000 |
+
"step": 10325
|
| 3001 |
+
},
|
| 3002 |
+
{
|
| 3003 |
+
"epoch": 11.398678414096917,
|
| 3004 |
+
"grad_norm": 0.10834582895040512,
|
| 3005 |
+
"learning_rate": 9.897435897435899e-06,
|
| 3006 |
+
"loss": 0.0207,
|
| 3007 |
+
"step": 10350
|
| 3008 |
+
},
|
| 3009 |
+
{
|
| 3010 |
+
"epoch": 11.426211453744493,
|
| 3011 |
+
"grad_norm": 0.36438989639282227,
|
| 3012 |
+
"learning_rate": 9.871794871794872e-06,
|
| 3013 |
+
"loss": 0.0159,
|
| 3014 |
+
"step": 10375
|
| 3015 |
+
},
|
| 3016 |
+
{
|
| 3017 |
+
"epoch": 11.45374449339207,
|
| 3018 |
+
"grad_norm": 0.2105102390050888,
|
| 3019 |
+
"learning_rate": 9.846153846153848e-06,
|
| 3020 |
+
"loss": 0.0174,
|
| 3021 |
+
"step": 10400
|
| 3022 |
+
},
|
| 3023 |
+
{
|
| 3024 |
+
"epoch": 11.481277533039648,
|
| 3025 |
+
"grad_norm": 0.29330697655677795,
|
| 3026 |
+
"learning_rate": 9.820512820512821e-06,
|
| 3027 |
+
"loss": 0.0139,
|
| 3028 |
+
"step": 10425
|
| 3029 |
+
},
|
| 3030 |
+
{
|
| 3031 |
+
"epoch": 11.508810572687224,
|
| 3032 |
+
"grad_norm": 0.14886349439620972,
|
| 3033 |
+
"learning_rate": 9.794871794871795e-06,
|
| 3034 |
+
"loss": 0.0168,
|
| 3035 |
+
"step": 10450
|
| 3036 |
+
},
|
| 3037 |
+
{
|
| 3038 |
+
"epoch": 11.536343612334802,
|
| 3039 |
+
"grad_norm": 0.2385999709367752,
|
| 3040 |
+
"learning_rate": 9.76923076923077e-06,
|
| 3041 |
+
"loss": 0.0153,
|
| 3042 |
+
"step": 10475
|
| 3043 |
+
},
|
| 3044 |
+
{
|
| 3045 |
+
"epoch": 11.56387665198238,
|
| 3046 |
+
"grad_norm": 0.12921461462974548,
|
| 3047 |
+
"learning_rate": 9.743589743589744e-06,
|
| 3048 |
+
"loss": 0.0211,
|
| 3049 |
+
"step": 10500
|
| 3050 |
+
},
|
| 3051 |
+
{
|
| 3052 |
+
"epoch": 11.591409691629956,
|
| 3053 |
+
"grad_norm": 0.18354228138923645,
|
| 3054 |
+
"learning_rate": 9.71794871794872e-06,
|
| 3055 |
+
"loss": 0.0182,
|
| 3056 |
+
"step": 10525
|
| 3057 |
+
},
|
| 3058 |
+
{
|
| 3059 |
+
"epoch": 11.618942731277533,
|
| 3060 |
+
"grad_norm": 0.31387805938720703,
|
| 3061 |
+
"learning_rate": 9.692307692307693e-06,
|
| 3062 |
+
"loss": 0.0149,
|
| 3063 |
+
"step": 10550
|
| 3064 |
+
},
|
| 3065 |
+
{
|
| 3066 |
+
"epoch": 11.646475770925111,
|
| 3067 |
+
"grad_norm": 0.2267555296421051,
|
| 3068 |
+
"learning_rate": 9.666666666666667e-06,
|
| 3069 |
+
"loss": 0.0158,
|
| 3070 |
+
"step": 10575
|
| 3071 |
+
},
|
| 3072 |
+
{
|
| 3073 |
+
"epoch": 11.674008810572687,
|
| 3074 |
+
"grad_norm": 0.20056696236133575,
|
| 3075 |
+
"learning_rate": 9.641025641025642e-06,
|
| 3076 |
+
"loss": 0.0185,
|
| 3077 |
+
"step": 10600
|
| 3078 |
+
},
|
| 3079 |
+
{
|
| 3080 |
+
"epoch": 11.701541850220265,
|
| 3081 |
+
"grad_norm": 0.33203795552253723,
|
| 3082 |
+
"learning_rate": 9.615384615384616e-06,
|
| 3083 |
+
"loss": 0.0162,
|
| 3084 |
+
"step": 10625
|
| 3085 |
+
},
|
| 3086 |
+
{
|
| 3087 |
+
"epoch": 11.729074889867842,
|
| 3088 |
+
"grad_norm": 0.20103588700294495,
|
| 3089 |
+
"learning_rate": 9.589743589743591e-06,
|
| 3090 |
+
"loss": 0.0168,
|
| 3091 |
+
"step": 10650
|
| 3092 |
+
},
|
| 3093 |
+
{
|
| 3094 |
+
"epoch": 11.756607929515418,
|
| 3095 |
+
"grad_norm": 0.2045770138502121,
|
| 3096 |
+
"learning_rate": 9.564102564102565e-06,
|
| 3097 |
+
"loss": 0.0151,
|
| 3098 |
+
"step": 10675
|
| 3099 |
+
},
|
| 3100 |
+
{
|
| 3101 |
+
"epoch": 11.784140969162996,
|
| 3102 |
+
"grad_norm": 0.35391268134117126,
|
| 3103 |
+
"learning_rate": 9.53846153846154e-06,
|
| 3104 |
+
"loss": 0.0159,
|
| 3105 |
+
"step": 10700
|
| 3106 |
+
},
|
| 3107 |
+
{
|
| 3108 |
+
"epoch": 11.811674008810574,
|
| 3109 |
+
"grad_norm": 0.29899677634239197,
|
| 3110 |
+
"learning_rate": 9.512820512820514e-06,
|
| 3111 |
+
"loss": 0.0161,
|
| 3112 |
+
"step": 10725
|
| 3113 |
+
},
|
| 3114 |
+
{
|
| 3115 |
+
"epoch": 11.83920704845815,
|
| 3116 |
+
"grad_norm": 0.15067006647586823,
|
| 3117 |
+
"learning_rate": 9.487179487179487e-06,
|
| 3118 |
+
"loss": 0.0213,
|
| 3119 |
+
"step": 10750
|
| 3120 |
+
},
|
| 3121 |
+
{
|
| 3122 |
+
"epoch": 11.866740088105727,
|
| 3123 |
+
"grad_norm": 0.09212782979011536,
|
| 3124 |
+
"learning_rate": 9.461538461538463e-06,
|
| 3125 |
+
"loss": 0.018,
|
| 3126 |
+
"step": 10775
|
| 3127 |
+
},
|
| 3128 |
+
{
|
| 3129 |
+
"epoch": 11.894273127753303,
|
| 3130 |
+
"grad_norm": 0.2554011642932892,
|
| 3131 |
+
"learning_rate": 9.435897435897436e-06,
|
| 3132 |
+
"loss": 0.0171,
|
| 3133 |
+
"step": 10800
|
| 3134 |
+
},
|
| 3135 |
+
{
|
| 3136 |
+
"epoch": 11.92180616740088,
|
| 3137 |
+
"grad_norm": 0.15091727674007416,
|
| 3138 |
+
"learning_rate": 9.410256410256412e-06,
|
| 3139 |
+
"loss": 0.0173,
|
| 3140 |
+
"step": 10825
|
| 3141 |
+
},
|
| 3142 |
+
{
|
| 3143 |
+
"epoch": 11.949339207048459,
|
| 3144 |
+
"grad_norm": 0.5245828628540039,
|
| 3145 |
+
"learning_rate": 9.384615384615385e-06,
|
| 3146 |
+
"loss": 0.0143,
|
| 3147 |
+
"step": 10850
|
| 3148 |
+
},
|
| 3149 |
+
{
|
| 3150 |
+
"epoch": 11.976872246696034,
|
| 3151 |
+
"grad_norm": 0.2978862524032593,
|
| 3152 |
+
"learning_rate": 9.358974358974359e-06,
|
| 3153 |
+
"loss": 0.018,
|
| 3154 |
+
"step": 10875
|
| 3155 |
+
},
|
| 3156 |
+
{
|
| 3157 |
+
"epoch": 12.004405286343612,
|
| 3158 |
+
"grad_norm": 0.07283396273851395,
|
| 3159 |
+
"learning_rate": 9.333333333333334e-06,
|
| 3160 |
+
"loss": 0.0182,
|
| 3161 |
+
"step": 10900
|
| 3162 |
+
},
|
| 3163 |
+
{
|
| 3164 |
+
"epoch": 12.03193832599119,
|
| 3165 |
+
"grad_norm": 0.13879446685314178,
|
| 3166 |
+
"learning_rate": 9.307692307692308e-06,
|
| 3167 |
+
"loss": 0.0127,
|
| 3168 |
+
"step": 10925
|
| 3169 |
+
},
|
| 3170 |
+
{
|
| 3171 |
+
"epoch": 12.059471365638766,
|
| 3172 |
+
"grad_norm": 0.16657480597496033,
|
| 3173 |
+
"learning_rate": 9.282051282051283e-06,
|
| 3174 |
+
"loss": 0.014,
|
| 3175 |
+
"step": 10950
|
| 3176 |
+
},
|
| 3177 |
+
{
|
| 3178 |
+
"epoch": 12.087004405286343,
|
| 3179 |
+
"grad_norm": 0.16087482869625092,
|
| 3180 |
+
"learning_rate": 9.256410256410257e-06,
|
| 3181 |
+
"loss": 0.0112,
|
| 3182 |
+
"step": 10975
|
| 3183 |
+
},
|
| 3184 |
+
{
|
| 3185 |
+
"epoch": 12.114537444933921,
|
| 3186 |
+
"grad_norm": 0.18895640969276428,
|
| 3187 |
+
"learning_rate": 9.230769230769232e-06,
|
| 3188 |
+
"loss": 0.0117,
|
| 3189 |
+
"step": 11000
|
| 3190 |
+
},
|
| 3191 |
+
{
|
| 3192 |
+
"epoch": 12.114537444933921,
|
| 3193 |
+
"eval_cer": 24.002957596322496,
|
| 3194 |
+
"eval_loss": 0.8591621518135071,
|
| 3195 |
+
"eval_runtime": 1710.0216,
|
| 3196 |
+
"eval_samples_per_second": 6.188,
|
| 3197 |
+
"eval_steps_per_second": 1.547,
|
| 3198 |
+
"eval_wer": 84.15841584158416,
|
| 3199 |
+
"step": 11000
|
| 3200 |
+
},
|
| 3201 |
+
{
|
| 3202 |
+
"epoch": 12.142070484581497,
|
| 3203 |
+
"grad_norm": 0.08321177214384079,
|
| 3204 |
+
"learning_rate": 9.205128205128206e-06,
|
| 3205 |
+
"loss": 0.0117,
|
| 3206 |
+
"step": 11025
|
| 3207 |
+
},
|
| 3208 |
+
{
|
| 3209 |
+
"epoch": 12.169603524229075,
|
| 3210 |
+
"grad_norm": 0.20632462203502655,
|
| 3211 |
+
"learning_rate": 9.17948717948718e-06,
|
| 3212 |
+
"loss": 0.0131,
|
| 3213 |
+
"step": 11050
|
| 3214 |
+
},
|
| 3215 |
+
{
|
| 3216 |
+
"epoch": 12.197136563876652,
|
| 3217 |
+
"grad_norm": 0.4616119861602783,
|
| 3218 |
+
"learning_rate": 9.153846153846155e-06,
|
| 3219 |
+
"loss": 0.0184,
|
| 3220 |
+
"step": 11075
|
| 3221 |
+
},
|
| 3222 |
+
{
|
| 3223 |
+
"epoch": 12.224669603524228,
|
| 3224 |
+
"grad_norm": 0.89581298828125,
|
| 3225 |
+
"learning_rate": 9.128205128205129e-06,
|
| 3226 |
+
"loss": 0.0155,
|
| 3227 |
+
"step": 11100
|
| 3228 |
+
},
|
| 3229 |
+
{
|
| 3230 |
+
"epoch": 12.252202643171806,
|
| 3231 |
+
"grad_norm": 0.06262637674808502,
|
| 3232 |
+
"learning_rate": 9.102564102564104e-06,
|
| 3233 |
+
"loss": 0.0114,
|
| 3234 |
+
"step": 11125
|
| 3235 |
+
},
|
| 3236 |
+
{
|
| 3237 |
+
"epoch": 12.279735682819384,
|
| 3238 |
+
"grad_norm": 0.30928879976272583,
|
| 3239 |
+
"learning_rate": 9.076923076923078e-06,
|
| 3240 |
+
"loss": 0.0152,
|
| 3241 |
+
"step": 11150
|
| 3242 |
+
},
|
| 3243 |
+
{
|
| 3244 |
+
"epoch": 12.30726872246696,
|
| 3245 |
+
"grad_norm": 0.36566054821014404,
|
| 3246 |
+
"learning_rate": 9.051282051282051e-06,
|
| 3247 |
+
"loss": 0.011,
|
| 3248 |
+
"step": 11175
|
| 3249 |
+
},
|
| 3250 |
+
{
|
| 3251 |
+
"epoch": 12.334801762114537,
|
| 3252 |
+
"grad_norm": 0.21459051966667175,
|
| 3253 |
+
"learning_rate": 9.025641025641027e-06,
|
| 3254 |
+
"loss": 0.0127,
|
| 3255 |
+
"step": 11200
|
| 3256 |
+
},
|
| 3257 |
+
{
|
| 3258 |
+
"epoch": 12.362334801762115,
|
| 3259 |
+
"grad_norm": 0.35859718918800354,
|
| 3260 |
+
"learning_rate": 9e-06,
|
| 3261 |
+
"loss": 0.0125,
|
| 3262 |
+
"step": 11225
|
| 3263 |
+
},
|
| 3264 |
+
{
|
| 3265 |
+
"epoch": 12.389867841409691,
|
| 3266 |
+
"grad_norm": 0.5205895304679871,
|
| 3267 |
+
"learning_rate": 8.974358974358976e-06,
|
| 3268 |
+
"loss": 0.0135,
|
| 3269 |
+
"step": 11250
|
| 3270 |
+
},
|
| 3271 |
+
{
|
| 3272 |
+
"epoch": 12.417400881057269,
|
| 3273 |
+
"grad_norm": 0.09275946021080017,
|
| 3274 |
+
"learning_rate": 8.94871794871795e-06,
|
| 3275 |
+
"loss": 0.0124,
|
| 3276 |
+
"step": 11275
|
| 3277 |
+
},
|
| 3278 |
+
{
|
| 3279 |
+
"epoch": 12.444933920704846,
|
| 3280 |
+
"grad_norm": 0.4651426076889038,
|
| 3281 |
+
"learning_rate": 8.923076923076925e-06,
|
| 3282 |
+
"loss": 0.0118,
|
| 3283 |
+
"step": 11300
|
| 3284 |
+
},
|
| 3285 |
+
{
|
| 3286 |
+
"epoch": 12.472466960352422,
|
| 3287 |
+
"grad_norm": 0.1863466203212738,
|
| 3288 |
+
"learning_rate": 8.897435897435898e-06,
|
| 3289 |
+
"loss": 0.0116,
|
| 3290 |
+
"step": 11325
|
| 3291 |
+
},
|
| 3292 |
+
{
|
| 3293 |
+
"epoch": 12.5,
|
| 3294 |
+
"grad_norm": 0.467035174369812,
|
| 3295 |
+
"learning_rate": 8.871794871794872e-06,
|
| 3296 |
+
"loss": 0.0111,
|
| 3297 |
+
"step": 11350
|
| 3298 |
+
},
|
| 3299 |
+
{
|
| 3300 |
+
"epoch": 12.527533039647578,
|
| 3301 |
+
"grad_norm": 0.13402460515499115,
|
| 3302 |
+
"learning_rate": 8.846153846153847e-06,
|
| 3303 |
+
"loss": 0.0147,
|
| 3304 |
+
"step": 11375
|
| 3305 |
+
},
|
| 3306 |
+
{
|
| 3307 |
+
"epoch": 12.555066079295154,
|
| 3308 |
+
"grad_norm": 0.14492958784103394,
|
| 3309 |
+
"learning_rate": 8.820512820512821e-06,
|
| 3310 |
+
"loss": 0.0128,
|
| 3311 |
+
"step": 11400
|
| 3312 |
+
},
|
| 3313 |
+
{
|
| 3314 |
+
"epoch": 12.582599118942731,
|
| 3315 |
+
"grad_norm": 0.17025631666183472,
|
| 3316 |
+
"learning_rate": 8.794871794871796e-06,
|
| 3317 |
+
"loss": 0.0135,
|
| 3318 |
+
"step": 11425
|
| 3319 |
+
},
|
| 3320 |
+
{
|
| 3321 |
+
"epoch": 12.610132158590309,
|
| 3322 |
+
"grad_norm": 0.43454456329345703,
|
| 3323 |
+
"learning_rate": 8.76923076923077e-06,
|
| 3324 |
+
"loss": 0.0141,
|
| 3325 |
+
"step": 11450
|
| 3326 |
+
},
|
| 3327 |
+
{
|
| 3328 |
+
"epoch": 12.637665198237885,
|
| 3329 |
+
"grad_norm": 0.4362468421459198,
|
| 3330 |
+
"learning_rate": 8.743589743589743e-06,
|
| 3331 |
+
"loss": 0.0167,
|
| 3332 |
+
"step": 11475
|
| 3333 |
+
},
|
| 3334 |
+
{
|
| 3335 |
+
"epoch": 12.665198237885463,
|
| 3336 |
+
"grad_norm": 0.20894859731197357,
|
| 3337 |
+
"learning_rate": 8.717948717948719e-06,
|
| 3338 |
+
"loss": 0.019,
|
| 3339 |
+
"step": 11500
|
| 3340 |
+
},
|
| 3341 |
+
{
|
| 3342 |
+
"epoch": 12.69273127753304,
|
| 3343 |
+
"grad_norm": 0.37119993567466736,
|
| 3344 |
+
"learning_rate": 8.692307692307692e-06,
|
| 3345 |
+
"loss": 0.0136,
|
| 3346 |
+
"step": 11525
|
| 3347 |
+
},
|
| 3348 |
+
{
|
| 3349 |
+
"epoch": 12.720264317180616,
|
| 3350 |
+
"grad_norm": 0.06687525659799576,
|
| 3351 |
+
"learning_rate": 8.666666666666668e-06,
|
| 3352 |
+
"loss": 0.0133,
|
| 3353 |
+
"step": 11550
|
| 3354 |
+
},
|
| 3355 |
+
{
|
| 3356 |
+
"epoch": 12.747797356828194,
|
| 3357 |
+
"grad_norm": 0.1411520540714264,
|
| 3358 |
+
"learning_rate": 8.641025641025641e-06,
|
| 3359 |
+
"loss": 0.0103,
|
| 3360 |
+
"step": 11575
|
| 3361 |
+
},
|
| 3362 |
+
{
|
| 3363 |
+
"epoch": 12.775330396475772,
|
| 3364 |
+
"grad_norm": 0.5004802346229553,
|
| 3365 |
+
"learning_rate": 8.615384615384617e-06,
|
| 3366 |
+
"loss": 0.0139,
|
| 3367 |
+
"step": 11600
|
| 3368 |
+
},
|
| 3369 |
+
{
|
| 3370 |
+
"epoch": 12.802863436123348,
|
| 3371 |
+
"grad_norm": 0.12091943621635437,
|
| 3372 |
+
"learning_rate": 8.58974358974359e-06,
|
| 3373 |
+
"loss": 0.0151,
|
| 3374 |
+
"step": 11625
|
| 3375 |
+
},
|
| 3376 |
+
{
|
| 3377 |
+
"epoch": 12.830396475770925,
|
| 3378 |
+
"grad_norm": 0.39368936419487,
|
| 3379 |
+
"learning_rate": 8.564102564102564e-06,
|
| 3380 |
+
"loss": 0.0155,
|
| 3381 |
+
"step": 11650
|
| 3382 |
+
},
|
| 3383 |
+
{
|
| 3384 |
+
"epoch": 12.857929515418503,
|
| 3385 |
+
"grad_norm": 0.1569003313779831,
|
| 3386 |
+
"learning_rate": 8.53846153846154e-06,
|
| 3387 |
+
"loss": 0.0144,
|
| 3388 |
+
"step": 11675
|
| 3389 |
+
},
|
| 3390 |
+
{
|
| 3391 |
+
"epoch": 12.885462555066079,
|
| 3392 |
+
"grad_norm": 0.2494770586490631,
|
| 3393 |
+
"learning_rate": 8.512820512820513e-06,
|
| 3394 |
+
"loss": 0.0154,
|
| 3395 |
+
"step": 11700
|
| 3396 |
+
},
|
| 3397 |
+
{
|
| 3398 |
+
"epoch": 12.912995594713657,
|
| 3399 |
+
"grad_norm": 0.2350165992975235,
|
| 3400 |
+
"learning_rate": 8.487179487179488e-06,
|
| 3401 |
+
"loss": 0.0154,
|
| 3402 |
+
"step": 11725
|
| 3403 |
+
},
|
| 3404 |
+
{
|
| 3405 |
+
"epoch": 12.940528634361234,
|
| 3406 |
+
"grad_norm": 0.25457248091697693,
|
| 3407 |
+
"learning_rate": 8.461538461538462e-06,
|
| 3408 |
+
"loss": 0.0146,
|
| 3409 |
+
"step": 11750
|
| 3410 |
+
},
|
| 3411 |
+
{
|
| 3412 |
+
"epoch": 12.96806167400881,
|
| 3413 |
+
"grad_norm": 0.5396599769592285,
|
| 3414 |
+
"learning_rate": 8.435897435897436e-06,
|
| 3415 |
+
"loss": 0.0164,
|
| 3416 |
+
"step": 11775
|
| 3417 |
+
},
|
| 3418 |
+
{
|
| 3419 |
+
"epoch": 12.995594713656388,
|
| 3420 |
+
"grad_norm": 0.08093304932117462,
|
| 3421 |
+
"learning_rate": 8.410256410256411e-06,
|
| 3422 |
+
"loss": 0.0152,
|
| 3423 |
+
"step": 11800
|
| 3424 |
+
},
|
| 3425 |
+
{
|
| 3426 |
+
"epoch": 13.023127753303966,
|
| 3427 |
+
"grad_norm": 0.2992173433303833,
|
| 3428 |
+
"learning_rate": 8.384615384615385e-06,
|
| 3429 |
+
"loss": 0.0135,
|
| 3430 |
+
"step": 11825
|
| 3431 |
+
},
|
| 3432 |
+
{
|
| 3433 |
+
"epoch": 13.050660792951541,
|
| 3434 |
+
"grad_norm": 0.19341129064559937,
|
| 3435 |
+
"learning_rate": 8.35897435897436e-06,
|
| 3436 |
+
"loss": 0.0126,
|
| 3437 |
+
"step": 11850
|
| 3438 |
+
},
|
| 3439 |
+
{
|
| 3440 |
+
"epoch": 13.07819383259912,
|
| 3441 |
+
"grad_norm": 0.3333016633987427,
|
| 3442 |
+
"learning_rate": 8.333333333333334e-06,
|
| 3443 |
+
"loss": 0.0172,
|
| 3444 |
+
"step": 11875
|
| 3445 |
+
},
|
| 3446 |
+
{
|
| 3447 |
+
"epoch": 13.105726872246697,
|
| 3448 |
+
"grad_norm": 0.22165773808956146,
|
| 3449 |
+
"learning_rate": 8.307692307692309e-06,
|
| 3450 |
+
"loss": 0.0107,
|
| 3451 |
+
"step": 11900
|
| 3452 |
+
},
|
| 3453 |
+
{
|
| 3454 |
+
"epoch": 13.133259911894273,
|
| 3455 |
+
"grad_norm": 0.14357928931713104,
|
| 3456 |
+
"learning_rate": 8.282051282051283e-06,
|
| 3457 |
+
"loss": 0.0125,
|
| 3458 |
+
"step": 11925
|
| 3459 |
+
},
|
| 3460 |
+
{
|
| 3461 |
+
"epoch": 13.16079295154185,
|
| 3462 |
+
"grad_norm": 0.46441957354545593,
|
| 3463 |
+
"learning_rate": 8.256410256410256e-06,
|
| 3464 |
+
"loss": 0.0098,
|
| 3465 |
+
"step": 11950
|
| 3466 |
+
},
|
| 3467 |
+
{
|
| 3468 |
+
"epoch": 13.188325991189428,
|
| 3469 |
+
"grad_norm": 0.0940885990858078,
|
| 3470 |
+
"learning_rate": 8.230769230769232e-06,
|
| 3471 |
+
"loss": 0.0126,
|
| 3472 |
+
"step": 11975
|
| 3473 |
+
},
|
| 3474 |
+
{
|
| 3475 |
+
"epoch": 13.215859030837004,
|
| 3476 |
+
"grad_norm": 0.09288233518600464,
|
| 3477 |
+
"learning_rate": 8.205128205128205e-06,
|
| 3478 |
+
"loss": 0.0111,
|
| 3479 |
+
"step": 12000
|
| 3480 |
+
},
|
| 3481 |
+
{
|
| 3482 |
+
"epoch": 13.215859030837004,
|
| 3483 |
+
"eval_cer": 24.353742276432083,
|
| 3484 |
+
"eval_loss": 0.8609709739685059,
|
| 3485 |
+
"eval_runtime": 1703.4403,
|
| 3486 |
+
"eval_samples_per_second": 6.212,
|
| 3487 |
+
"eval_steps_per_second": 1.553,
|
| 3488 |
+
"eval_wer": 83.83781235266385,
|
| 3489 |
+
"step": 12000
|
| 3490 |
+
},
|
| 3491 |
+
{
|
| 3492 |
+
"epoch": 13.243392070484582,
|
| 3493 |
+
"grad_norm": 0.2091829776763916,
|
| 3494 |
+
"learning_rate": 8.17948717948718e-06,
|
| 3495 |
+
"loss": 0.0098,
|
| 3496 |
+
"step": 12025
|
| 3497 |
+
},
|
| 3498 |
+
{
|
| 3499 |
+
"epoch": 13.270925110132158,
|
| 3500 |
+
"grad_norm": 0.02503281459212303,
|
| 3501 |
+
"learning_rate": 8.153846153846154e-06,
|
| 3502 |
+
"loss": 0.0066,
|
| 3503 |
+
"step": 12050
|
| 3504 |
+
},
|
| 3505 |
+
{
|
| 3506 |
+
"epoch": 13.298458149779735,
|
| 3507 |
+
"grad_norm": 0.2379840463399887,
|
| 3508 |
+
"learning_rate": 8.12820512820513e-06,
|
| 3509 |
+
"loss": 0.0106,
|
| 3510 |
+
"step": 12075
|
| 3511 |
+
},
|
| 3512 |
+
{
|
| 3513 |
+
"epoch": 13.325991189427313,
|
| 3514 |
+
"grad_norm": 0.06553735584020615,
|
| 3515 |
+
"learning_rate": 8.102564102564103e-06,
|
| 3516 |
+
"loss": 0.0114,
|
| 3517 |
+
"step": 12100
|
| 3518 |
+
},
|
| 3519 |
+
{
|
| 3520 |
+
"epoch": 13.353524229074889,
|
| 3521 |
+
"grad_norm": 0.05876093730330467,
|
| 3522 |
+
"learning_rate": 8.076923076923077e-06,
|
| 3523 |
+
"loss": 0.0089,
|
| 3524 |
+
"step": 12125
|
| 3525 |
+
},
|
| 3526 |
+
{
|
| 3527 |
+
"epoch": 13.381057268722467,
|
| 3528 |
+
"grad_norm": 0.0722096711397171,
|
| 3529 |
+
"learning_rate": 8.051282051282052e-06,
|
| 3530 |
+
"loss": 0.01,
|
| 3531 |
+
"step": 12150
|
| 3532 |
+
},
|
| 3533 |
+
{
|
| 3534 |
+
"epoch": 13.408590308370044,
|
| 3535 |
+
"grad_norm": 0.16798360645771027,
|
| 3536 |
+
"learning_rate": 8.025641025641026e-06,
|
| 3537 |
+
"loss": 0.0112,
|
| 3538 |
+
"step": 12175
|
| 3539 |
+
},
|
| 3540 |
+
{
|
| 3541 |
+
"epoch": 13.43612334801762,
|
| 3542 |
+
"grad_norm": 0.08319924026727676,
|
| 3543 |
+
"learning_rate": 8.000000000000001e-06,
|
| 3544 |
+
"loss": 0.0164,
|
| 3545 |
+
"step": 12200
|
| 3546 |
+
},
|
| 3547 |
+
{
|
| 3548 |
+
"epoch": 13.463656387665198,
|
| 3549 |
+
"grad_norm": 0.03909989818930626,
|
| 3550 |
+
"learning_rate": 7.974358974358975e-06,
|
| 3551 |
+
"loss": 0.0124,
|
| 3552 |
+
"step": 12225
|
| 3553 |
+
},
|
| 3554 |
+
{
|
| 3555 |
+
"epoch": 13.491189427312776,
|
| 3556 |
+
"grad_norm": 0.3100120425224304,
|
| 3557 |
+
"learning_rate": 7.948717948717949e-06,
|
| 3558 |
+
"loss": 0.0103,
|
| 3559 |
+
"step": 12250
|
| 3560 |
+
},
|
| 3561 |
+
{
|
| 3562 |
+
"epoch": 13.518722466960352,
|
| 3563 |
+
"grad_norm": 0.20275218784809113,
|
| 3564 |
+
"learning_rate": 7.923076923076924e-06,
|
| 3565 |
+
"loss": 0.0124,
|
| 3566 |
+
"step": 12275
|
| 3567 |
+
},
|
| 3568 |
+
{
|
| 3569 |
+
"epoch": 13.54625550660793,
|
| 3570 |
+
"grad_norm": 0.43703603744506836,
|
| 3571 |
+
"learning_rate": 7.897435897435898e-06,
|
| 3572 |
+
"loss": 0.0126,
|
| 3573 |
+
"step": 12300
|
| 3574 |
+
},
|
| 3575 |
+
{
|
| 3576 |
+
"epoch": 13.573788546255507,
|
| 3577 |
+
"grad_norm": 0.17036165297031403,
|
| 3578 |
+
"learning_rate": 7.871794871794873e-06,
|
| 3579 |
+
"loss": 0.0123,
|
| 3580 |
+
"step": 12325
|
| 3581 |
+
},
|
| 3582 |
+
{
|
| 3583 |
+
"epoch": 13.601321585903083,
|
| 3584 |
+
"grad_norm": 0.2403183877468109,
|
| 3585 |
+
"learning_rate": 7.846153846153847e-06,
|
| 3586 |
+
"loss": 0.0124,
|
| 3587 |
+
"step": 12350
|
| 3588 |
+
},
|
| 3589 |
+
{
|
| 3590 |
+
"epoch": 13.62885462555066,
|
| 3591 |
+
"grad_norm": 0.14506351947784424,
|
| 3592 |
+
"learning_rate": 7.820512820512822e-06,
|
| 3593 |
+
"loss": 0.0113,
|
| 3594 |
+
"step": 12375
|
| 3595 |
+
},
|
| 3596 |
+
{
|
| 3597 |
+
"epoch": 13.656387665198238,
|
| 3598 |
+
"grad_norm": 0.17597095668315887,
|
| 3599 |
+
"learning_rate": 7.794871794871796e-06,
|
| 3600 |
+
"loss": 0.013,
|
| 3601 |
+
"step": 12400
|
| 3602 |
+
},
|
| 3603 |
+
{
|
| 3604 |
+
"epoch": 13.683920704845814,
|
| 3605 |
+
"grad_norm": 0.10913864523172379,
|
| 3606 |
+
"learning_rate": 7.76923076923077e-06,
|
| 3607 |
+
"loss": 0.0111,
|
| 3608 |
+
"step": 12425
|
| 3609 |
+
},
|
| 3610 |
+
{
|
| 3611 |
+
"epoch": 13.711453744493392,
|
| 3612 |
+
"grad_norm": 0.09308384358882904,
|
| 3613 |
+
"learning_rate": 7.743589743589745e-06,
|
| 3614 |
+
"loss": 0.0107,
|
| 3615 |
+
"step": 12450
|
| 3616 |
+
},
|
| 3617 |
+
{
|
| 3618 |
+
"epoch": 13.73898678414097,
|
| 3619 |
+
"grad_norm": 0.15387412905693054,
|
| 3620 |
+
"learning_rate": 7.717948717948718e-06,
|
| 3621 |
+
"loss": 0.0081,
|
| 3622 |
+
"step": 12475
|
| 3623 |
+
},
|
| 3624 |
+
{
|
| 3625 |
+
"epoch": 13.766519823788546,
|
| 3626 |
+
"grad_norm": 0.1064794659614563,
|
| 3627 |
+
"learning_rate": 7.692307692307694e-06,
|
| 3628 |
+
"loss": 0.0103,
|
| 3629 |
+
"step": 12500
|
| 3630 |
+
},
|
| 3631 |
+
{
|
| 3632 |
+
"epoch": 13.794052863436123,
|
| 3633 |
+
"grad_norm": 0.461704283952713,
|
| 3634 |
+
"learning_rate": 7.666666666666667e-06,
|
| 3635 |
+
"loss": 0.0099,
|
| 3636 |
+
"step": 12525
|
| 3637 |
+
},
|
| 3638 |
+
{
|
| 3639 |
+
"epoch": 13.821585903083701,
|
| 3640 |
+
"grad_norm": 0.1432102471590042,
|
| 3641 |
+
"learning_rate": 7.641025641025641e-06,
|
| 3642 |
+
"loss": 0.012,
|
| 3643 |
+
"step": 12550
|
| 3644 |
+
},
|
| 3645 |
+
{
|
| 3646 |
+
"epoch": 13.849118942731277,
|
| 3647 |
+
"grad_norm": 0.07016371935606003,
|
| 3648 |
+
"learning_rate": 7.615384615384615e-06,
|
| 3649 |
+
"loss": 0.0102,
|
| 3650 |
+
"step": 12575
|
| 3651 |
+
},
|
| 3652 |
+
{
|
| 3653 |
+
"epoch": 13.876651982378855,
|
| 3654 |
+
"grad_norm": 0.07148536294698715,
|
| 3655 |
+
"learning_rate": 7.58974358974359e-06,
|
| 3656 |
+
"loss": 0.0135,
|
| 3657 |
+
"step": 12600
|
| 3658 |
+
},
|
| 3659 |
+
{
|
| 3660 |
+
"epoch": 13.904185022026432,
|
| 3661 |
+
"grad_norm": 0.11293036490678787,
|
| 3662 |
+
"learning_rate": 7.564102564102564e-06,
|
| 3663 |
+
"loss": 0.0119,
|
| 3664 |
+
"step": 12625
|
| 3665 |
+
},
|
| 3666 |
+
{
|
| 3667 |
+
"epoch": 13.931718061674008,
|
| 3668 |
+
"grad_norm": 0.12203430384397507,
|
| 3669 |
+
"learning_rate": 7.538461538461539e-06,
|
| 3670 |
+
"loss": 0.0133,
|
| 3671 |
+
"step": 12650
|
| 3672 |
+
},
|
| 3673 |
+
{
|
| 3674 |
+
"epoch": 13.959251101321586,
|
| 3675 |
+
"grad_norm": 0.18345963954925537,
|
| 3676 |
+
"learning_rate": 7.512820512820513e-06,
|
| 3677 |
+
"loss": 0.0131,
|
| 3678 |
+
"step": 12675
|
| 3679 |
+
},
|
| 3680 |
+
{
|
| 3681 |
+
"epoch": 13.986784140969164,
|
| 3682 |
+
"grad_norm": 0.135273739695549,
|
| 3683 |
+
"learning_rate": 7.487179487179488e-06,
|
| 3684 |
+
"loss": 0.0144,
|
| 3685 |
+
"step": 12700
|
| 3686 |
+
},
|
| 3687 |
+
{
|
| 3688 |
+
"epoch": 14.01431718061674,
|
| 3689 |
+
"grad_norm": 0.07620234042406082,
|
| 3690 |
+
"learning_rate": 7.461538461538462e-06,
|
| 3691 |
+
"loss": 0.0095,
|
| 3692 |
+
"step": 12725
|
| 3693 |
+
},
|
| 3694 |
+
{
|
| 3695 |
+
"epoch": 14.041850220264317,
|
| 3696 |
+
"grad_norm": 0.02833453379571438,
|
| 3697 |
+
"learning_rate": 7.435897435897437e-06,
|
| 3698 |
+
"loss": 0.0064,
|
| 3699 |
+
"step": 12750
|
| 3700 |
+
},
|
| 3701 |
+
{
|
| 3702 |
+
"epoch": 14.069383259911895,
|
| 3703 |
+
"grad_norm": 0.19002589583396912,
|
| 3704 |
+
"learning_rate": 7.410256410256411e-06,
|
| 3705 |
+
"loss": 0.0088,
|
| 3706 |
+
"step": 12775
|
| 3707 |
+
},
|
| 3708 |
+
{
|
| 3709 |
+
"epoch": 14.09691629955947,
|
| 3710 |
+
"grad_norm": 0.15754035115242004,
|
| 3711 |
+
"learning_rate": 7.384615384615386e-06,
|
| 3712 |
+
"loss": 0.0055,
|
| 3713 |
+
"step": 12800
|
| 3714 |
+
},
|
| 3715 |
+
{
|
| 3716 |
+
"epoch": 14.124449339207048,
|
| 3717 |
+
"grad_norm": 0.43471795320510864,
|
| 3718 |
+
"learning_rate": 7.35897435897436e-06,
|
| 3719 |
+
"loss": 0.0088,
|
| 3720 |
+
"step": 12825
|
| 3721 |
+
},
|
| 3722 |
+
{
|
| 3723 |
+
"epoch": 14.151982378854626,
|
| 3724 |
+
"grad_norm": 0.2886705696582794,
|
| 3725 |
+
"learning_rate": 7.333333333333333e-06,
|
| 3726 |
+
"loss": 0.0074,
|
| 3727 |
+
"step": 12850
|
| 3728 |
+
},
|
| 3729 |
+
{
|
| 3730 |
+
"epoch": 14.179515418502202,
|
| 3731 |
+
"grad_norm": 0.2692849040031433,
|
| 3732 |
+
"learning_rate": 7.307692307692308e-06,
|
| 3733 |
+
"loss": 0.0141,
|
| 3734 |
+
"step": 12875
|
| 3735 |
+
},
|
| 3736 |
+
{
|
| 3737 |
+
"epoch": 14.20704845814978,
|
| 3738 |
+
"grad_norm": 0.12292686104774475,
|
| 3739 |
+
"learning_rate": 7.282051282051282e-06,
|
| 3740 |
+
"loss": 0.0151,
|
| 3741 |
+
"step": 12900
|
| 3742 |
+
},
|
| 3743 |
+
{
|
| 3744 |
+
"epoch": 14.234581497797357,
|
| 3745 |
+
"grad_norm": 0.36948704719543457,
|
| 3746 |
+
"learning_rate": 7.256410256410257e-06,
|
| 3747 |
+
"loss": 0.0083,
|
| 3748 |
+
"step": 12925
|
| 3749 |
+
},
|
| 3750 |
+
{
|
| 3751 |
+
"epoch": 14.262114537444933,
|
| 3752 |
+
"grad_norm": 0.19203431904315948,
|
| 3753 |
+
"learning_rate": 7.230769230769231e-06,
|
| 3754 |
+
"loss": 0.0066,
|
| 3755 |
+
"step": 12950
|
| 3756 |
+
},
|
| 3757 |
+
{
|
| 3758 |
+
"epoch": 14.289647577092511,
|
| 3759 |
+
"grad_norm": 0.13032004237174988,
|
| 3760 |
+
"learning_rate": 7.205128205128206e-06,
|
| 3761 |
+
"loss": 0.0117,
|
| 3762 |
+
"step": 12975
|
| 3763 |
+
},
|
| 3764 |
+
{
|
| 3765 |
+
"epoch": 14.317180616740089,
|
| 3766 |
+
"grad_norm": 0.4582443833351135,
|
| 3767 |
+
"learning_rate": 7.17948717948718e-06,
|
| 3768 |
+
"loss": 0.0088,
|
| 3769 |
+
"step": 13000
|
| 3770 |
+
},
|
| 3771 |
+
{
|
| 3772 |
+
"epoch": 14.317180616740089,
|
| 3773 |
+
"eval_cer": 24.632306581224995,
|
| 3774 |
+
"eval_loss": 0.8743005394935608,
|
| 3775 |
+
"eval_runtime": 1717.1726,
|
| 3776 |
+
"eval_samples_per_second": 6.162,
|
| 3777 |
+
"eval_steps_per_second": 1.541,
|
| 3778 |
+
"eval_wer": 84.0924092409241,
|
| 3779 |
+
"step": 13000
|
| 3780 |
+
},
|
| 3781 |
+
{
|
| 3782 |
+
"epoch": 14.344713656387665,
|
| 3783 |
+
"grad_norm": 0.17969651520252228,
|
| 3784 |
+
"learning_rate": 7.153846153846155e-06,
|
| 3785 |
+
"loss": 0.0088,
|
| 3786 |
+
"step": 13025
|
| 3787 |
+
},
|
| 3788 |
+
{
|
| 3789 |
+
"epoch": 14.372246696035242,
|
| 3790 |
+
"grad_norm": 0.2400025725364685,
|
| 3791 |
+
"learning_rate": 7.128205128205129e-06,
|
| 3792 |
+
"loss": 0.0112,
|
| 3793 |
+
"step": 13050
|
| 3794 |
+
},
|
| 3795 |
+
{
|
| 3796 |
+
"epoch": 14.39977973568282,
|
| 3797 |
+
"grad_norm": 0.1150280013680458,
|
| 3798 |
+
"learning_rate": 7.102564102564104e-06,
|
| 3799 |
+
"loss": 0.0079,
|
| 3800 |
+
"step": 13075
|
| 3801 |
+
},
|
| 3802 |
+
{
|
| 3803 |
+
"epoch": 14.427312775330396,
|
| 3804 |
+
"grad_norm": 0.676898717880249,
|
| 3805 |
+
"learning_rate": 7.076923076923078e-06,
|
| 3806 |
+
"loss": 0.0097,
|
| 3807 |
+
"step": 13100
|
| 3808 |
+
},
|
| 3809 |
+
{
|
| 3810 |
+
"epoch": 14.454845814977974,
|
| 3811 |
+
"grad_norm": 0.22521457076072693,
|
| 3812 |
+
"learning_rate": 7.051282051282053e-06,
|
| 3813 |
+
"loss": 0.0067,
|
| 3814 |
+
"step": 13125
|
| 3815 |
+
},
|
| 3816 |
+
{
|
| 3817 |
+
"epoch": 14.482378854625551,
|
| 3818 |
+
"grad_norm": 0.05301696062088013,
|
| 3819 |
+
"learning_rate": 7.025641025641025e-06,
|
| 3820 |
+
"loss": 0.0082,
|
| 3821 |
+
"step": 13150
|
| 3822 |
+
},
|
| 3823 |
+
{
|
| 3824 |
+
"epoch": 14.509911894273127,
|
| 3825 |
+
"grad_norm": 0.056506820023059845,
|
| 3826 |
+
"learning_rate": 7e-06,
|
| 3827 |
+
"loss": 0.0068,
|
| 3828 |
+
"step": 13175
|
| 3829 |
+
},
|
| 3830 |
+
{
|
| 3831 |
+
"epoch": 14.537444933920705,
|
| 3832 |
+
"grad_norm": 0.31176385283470154,
|
| 3833 |
+
"learning_rate": 6.974358974358974e-06,
|
| 3834 |
+
"loss": 0.0076,
|
| 3835 |
+
"step": 13200
|
| 3836 |
+
},
|
| 3837 |
+
{
|
| 3838 |
+
"epoch": 14.564977973568283,
|
| 3839 |
+
"grad_norm": 0.27583009004592896,
|
| 3840 |
+
"learning_rate": 6.948717948717949e-06,
|
| 3841 |
+
"loss": 0.0069,
|
| 3842 |
+
"step": 13225
|
| 3843 |
+
},
|
| 3844 |
+
{
|
| 3845 |
+
"epoch": 14.592511013215859,
|
| 3846 |
+
"grad_norm": 0.04734191671013832,
|
| 3847 |
+
"learning_rate": 6.923076923076923e-06,
|
| 3848 |
+
"loss": 0.0102,
|
| 3849 |
+
"step": 13250
|
| 3850 |
+
},
|
| 3851 |
+
{
|
| 3852 |
+
"epoch": 14.620044052863436,
|
| 3853 |
+
"grad_norm": 0.26911285519599915,
|
| 3854 |
+
"learning_rate": 6.897435897435898e-06,
|
| 3855 |
+
"loss": 0.0101,
|
| 3856 |
+
"step": 13275
|
| 3857 |
+
},
|
| 3858 |
+
{
|
| 3859 |
+
"epoch": 14.647577092511014,
|
| 3860 |
+
"grad_norm": 0.07513406872749329,
|
| 3861 |
+
"learning_rate": 6.871794871794872e-06,
|
| 3862 |
+
"loss": 0.008,
|
| 3863 |
+
"step": 13300
|
| 3864 |
+
},
|
| 3865 |
+
{
|
| 3866 |
+
"epoch": 14.67511013215859,
|
| 3867 |
+
"grad_norm": 0.2528563141822815,
|
| 3868 |
+
"learning_rate": 6.846153846153847e-06,
|
| 3869 |
+
"loss": 0.0065,
|
| 3870 |
+
"step": 13325
|
| 3871 |
+
},
|
| 3872 |
+
{
|
| 3873 |
+
"epoch": 14.702643171806168,
|
| 3874 |
+
"grad_norm": 0.052279576659202576,
|
| 3875 |
+
"learning_rate": 6.820512820512821e-06,
|
| 3876 |
+
"loss": 0.0092,
|
| 3877 |
+
"step": 13350
|
| 3878 |
+
},
|
| 3879 |
+
{
|
| 3880 |
+
"epoch": 14.730176211453745,
|
| 3881 |
+
"grad_norm": 0.16867485642433167,
|
| 3882 |
+
"learning_rate": 6.794871794871796e-06,
|
| 3883 |
+
"loss": 0.0063,
|
| 3884 |
+
"step": 13375
|
| 3885 |
+
},
|
| 3886 |
+
{
|
| 3887 |
+
"epoch": 14.757709251101321,
|
| 3888 |
+
"grad_norm": 0.09278307855129242,
|
| 3889 |
+
"learning_rate": 6.76923076923077e-06,
|
| 3890 |
+
"loss": 0.007,
|
| 3891 |
+
"step": 13400
|
| 3892 |
+
},
|
| 3893 |
+
{
|
| 3894 |
+
"epoch": 14.785242290748899,
|
| 3895 |
+
"grad_norm": 0.03741838410496712,
|
| 3896 |
+
"learning_rate": 6.743589743589745e-06,
|
| 3897 |
+
"loss": 0.0049,
|
| 3898 |
+
"step": 13425
|
| 3899 |
+
},
|
| 3900 |
+
{
|
| 3901 |
+
"epoch": 14.812775330396477,
|
| 3902 |
+
"grad_norm": 0.07582259178161621,
|
| 3903 |
+
"learning_rate": 6.717948717948718e-06,
|
| 3904 |
+
"loss": 0.0089,
|
| 3905 |
+
"step": 13450
|
| 3906 |
+
},
|
| 3907 |
+
{
|
| 3908 |
+
"epoch": 14.840308370044053,
|
| 3909 |
+
"grad_norm": 0.22878766059875488,
|
| 3910 |
+
"learning_rate": 6.692307692307692e-06,
|
| 3911 |
+
"loss": 0.006,
|
| 3912 |
+
"step": 13475
|
| 3913 |
+
},
|
| 3914 |
+
{
|
| 3915 |
+
"epoch": 14.86784140969163,
|
| 3916 |
+
"grad_norm": 0.19926956295967102,
|
| 3917 |
+
"learning_rate": 6.666666666666667e-06,
|
| 3918 |
+
"loss": 0.0079,
|
| 3919 |
+
"step": 13500
|
| 3920 |
+
},
|
| 3921 |
+
{
|
| 3922 |
+
"epoch": 14.895374449339208,
|
| 3923 |
+
"grad_norm": 0.08697038888931274,
|
| 3924 |
+
"learning_rate": 6.641025641025641e-06,
|
| 3925 |
+
"loss": 0.0092,
|
| 3926 |
+
"step": 13525
|
| 3927 |
+
},
|
| 3928 |
+
{
|
| 3929 |
+
"epoch": 14.922907488986784,
|
| 3930 |
+
"grad_norm": 0.05428479611873627,
|
| 3931 |
+
"learning_rate": 6.615384615384616e-06,
|
| 3932 |
+
"loss": 0.0055,
|
| 3933 |
+
"step": 13550
|
| 3934 |
+
},
|
| 3935 |
+
{
|
| 3936 |
+
"epoch": 14.950440528634362,
|
| 3937 |
+
"grad_norm": 0.06628289073705673,
|
| 3938 |
+
"learning_rate": 6.58974358974359e-06,
|
| 3939 |
+
"loss": 0.0068,
|
| 3940 |
+
"step": 13575
|
| 3941 |
+
},
|
| 3942 |
+
{
|
| 3943 |
+
"epoch": 14.97797356828194,
|
| 3944 |
+
"grad_norm": 0.07876273989677429,
|
| 3945 |
+
"learning_rate": 6.564102564102565e-06,
|
| 3946 |
+
"loss": 0.0073,
|
| 3947 |
+
"step": 13600
|
| 3948 |
+
},
|
| 3949 |
+
{
|
| 3950 |
+
"epoch": 15.005506607929515,
|
| 3951 |
+
"grad_norm": 0.047706760466098785,
|
| 3952 |
+
"learning_rate": 6.538461538461539e-06,
|
| 3953 |
+
"loss": 0.0087,
|
| 3954 |
+
"step": 13625
|
| 3955 |
+
},
|
| 3956 |
+
{
|
| 3957 |
+
"epoch": 15.033039647577093,
|
| 3958 |
+
"grad_norm": 0.06642945855855942,
|
| 3959 |
+
"learning_rate": 6.512820512820514e-06,
|
| 3960 |
+
"loss": 0.0043,
|
| 3961 |
+
"step": 13650
|
| 3962 |
+
},
|
| 3963 |
+
{
|
| 3964 |
+
"epoch": 15.060572687224669,
|
| 3965 |
+
"grad_norm": 0.028193380683660507,
|
| 3966 |
+
"learning_rate": 6.487179487179488e-06,
|
| 3967 |
+
"loss": 0.0038,
|
| 3968 |
+
"step": 13675
|
| 3969 |
+
},
|
| 3970 |
+
{
|
| 3971 |
+
"epoch": 15.088105726872246,
|
| 3972 |
+
"grad_norm": 0.11379621922969818,
|
| 3973 |
+
"learning_rate": 6.461538461538463e-06,
|
| 3974 |
+
"loss": 0.0049,
|
| 3975 |
+
"step": 13700
|
| 3976 |
+
},
|
| 3977 |
+
{
|
| 3978 |
+
"epoch": 15.115638766519824,
|
| 3979 |
+
"grad_norm": 0.09332601726055145,
|
| 3980 |
+
"learning_rate": 6.435897435897437e-06,
|
| 3981 |
+
"loss": 0.0082,
|
| 3982 |
+
"step": 13725
|
| 3983 |
+
},
|
| 3984 |
+
{
|
| 3985 |
+
"epoch": 15.1431718061674,
|
| 3986 |
+
"grad_norm": 0.029045449569821358,
|
| 3987 |
+
"learning_rate": 6.410256410256412e-06,
|
| 3988 |
+
"loss": 0.0054,
|
| 3989 |
+
"step": 13750
|
| 3990 |
+
},
|
| 3991 |
+
{
|
| 3992 |
+
"epoch": 15.170704845814978,
|
| 3993 |
+
"grad_norm": 0.025950396433472633,
|
| 3994 |
+
"learning_rate": 6.384615384615384e-06,
|
| 3995 |
+
"loss": 0.0061,
|
| 3996 |
+
"step": 13775
|
| 3997 |
+
},
|
| 3998 |
+
{
|
| 3999 |
+
"epoch": 15.198237885462555,
|
| 4000 |
+
"grad_norm": 0.09153343737125397,
|
| 4001 |
+
"learning_rate": 6.358974358974359e-06,
|
| 4002 |
+
"loss": 0.0064,
|
| 4003 |
+
"step": 13800
|
| 4004 |
+
},
|
| 4005 |
+
{
|
| 4006 |
+
"epoch": 15.225770925110131,
|
| 4007 |
+
"grad_norm": 0.02982979267835617,
|
| 4008 |
+
"learning_rate": 6.333333333333333e-06,
|
| 4009 |
+
"loss": 0.0079,
|
| 4010 |
+
"step": 13825
|
| 4011 |
+
},
|
| 4012 |
+
{
|
| 4013 |
+
"epoch": 15.253303964757709,
|
| 4014 |
+
"grad_norm": 0.3530775010585785,
|
| 4015 |
+
"learning_rate": 6.307692307692308e-06,
|
| 4016 |
+
"loss": 0.0075,
|
| 4017 |
+
"step": 13850
|
| 4018 |
+
},
|
| 4019 |
+
{
|
| 4020 |
+
"epoch": 15.280837004405287,
|
| 4021 |
+
"grad_norm": 0.7259305715560913,
|
| 4022 |
+
"learning_rate": 6.282051282051282e-06,
|
| 4023 |
+
"loss": 0.0057,
|
| 4024 |
+
"step": 13875
|
| 4025 |
+
},
|
| 4026 |
+
{
|
| 4027 |
+
"epoch": 15.308370044052863,
|
| 4028 |
+
"grad_norm": 0.04625716805458069,
|
| 4029 |
+
"learning_rate": 6.256410256410257e-06,
|
| 4030 |
+
"loss": 0.0051,
|
| 4031 |
+
"step": 13900
|
| 4032 |
+
},
|
| 4033 |
+
{
|
| 4034 |
+
"epoch": 15.33590308370044,
|
| 4035 |
+
"grad_norm": 0.04628950357437134,
|
| 4036 |
+
"learning_rate": 6.230769230769231e-06,
|
| 4037 |
+
"loss": 0.0059,
|
| 4038 |
+
"step": 13925
|
| 4039 |
+
},
|
| 4040 |
+
{
|
| 4041 |
+
"epoch": 15.363436123348018,
|
| 4042 |
+
"grad_norm": 0.020349033176898956,
|
| 4043 |
+
"learning_rate": 6.205128205128206e-06,
|
| 4044 |
+
"loss": 0.0088,
|
| 4045 |
+
"step": 13950
|
| 4046 |
+
},
|
| 4047 |
+
{
|
| 4048 |
+
"epoch": 15.390969162995594,
|
| 4049 |
+
"grad_norm": 0.9620158672332764,
|
| 4050 |
+
"learning_rate": 6.17948717948718e-06,
|
| 4051 |
+
"loss": 0.006,
|
| 4052 |
+
"step": 13975
|
| 4053 |
+
},
|
| 4054 |
+
{
|
| 4055 |
+
"epoch": 15.418502202643172,
|
| 4056 |
+
"grad_norm": 0.12168940901756287,
|
| 4057 |
+
"learning_rate": 6.153846153846155e-06,
|
| 4058 |
+
"loss": 0.0112,
|
| 4059 |
+
"step": 14000
|
| 4060 |
+
},
|
| 4061 |
+
{
|
| 4062 |
+
"epoch": 15.418502202643172,
|
| 4063 |
+
"eval_cer": 24.93437116687492,
|
| 4064 |
+
"eval_loss": 0.8769130110740662,
|
| 4065 |
+
"eval_runtime": 1701.8659,
|
| 4066 |
+
"eval_samples_per_second": 6.217,
|
| 4067 |
+
"eval_steps_per_second": 1.555,
|
| 4068 |
+
"eval_wer": 84.18670438472418,
|
| 4069 |
+
"step": 14000
|
| 4070 |
+
},
|
| 4071 |
+
{
|
| 4072 |
+
"epoch": 15.44603524229075,
|
| 4073 |
+
"grad_norm": 0.0730656310915947,
|
| 4074 |
+
"learning_rate": 6.128205128205129e-06,
|
| 4075 |
+
"loss": 0.0051,
|
| 4076 |
+
"step": 14025
|
| 4077 |
+
},
|
| 4078 |
+
{
|
| 4079 |
+
"epoch": 15.473568281938325,
|
| 4080 |
+
"grad_norm": 0.38274314999580383,
|
| 4081 |
+
"learning_rate": 6.102564102564104e-06,
|
| 4082 |
+
"loss": 0.0118,
|
| 4083 |
+
"step": 14050
|
| 4084 |
+
},
|
| 4085 |
+
{
|
| 4086 |
+
"epoch": 15.501101321585903,
|
| 4087 |
+
"grad_norm": 0.6983866691589355,
|
| 4088 |
+
"learning_rate": 6.076923076923077e-06,
|
| 4089 |
+
"loss": 0.0126,
|
| 4090 |
+
"step": 14075
|
| 4091 |
+
},
|
| 4092 |
+
{
|
| 4093 |
+
"epoch": 15.52863436123348,
|
| 4094 |
+
"grad_norm": 0.14446529746055603,
|
| 4095 |
+
"learning_rate": 6.051282051282051e-06,
|
| 4096 |
+
"loss": 0.0063,
|
| 4097 |
+
"step": 14100
|
| 4098 |
+
},
|
| 4099 |
+
{
|
| 4100 |
+
"epoch": 15.556167400881057,
|
| 4101 |
+
"grad_norm": 0.03284487873315811,
|
| 4102 |
+
"learning_rate": 6.025641025641026e-06,
|
| 4103 |
+
"loss": 0.0072,
|
| 4104 |
+
"step": 14125
|
| 4105 |
+
},
|
| 4106 |
+
{
|
| 4107 |
+
"epoch": 15.583700440528634,
|
| 4108 |
+
"grad_norm": 0.14825402200222015,
|
| 4109 |
+
"learning_rate": 6e-06,
|
| 4110 |
+
"loss": 0.0074,
|
| 4111 |
+
"step": 14150
|
| 4112 |
+
},
|
| 4113 |
+
{
|
| 4114 |
+
"epoch": 15.611233480176212,
|
| 4115 |
+
"grad_norm": 0.08878491818904877,
|
| 4116 |
+
"learning_rate": 5.974358974358975e-06,
|
| 4117 |
+
"loss": 0.0067,
|
| 4118 |
+
"step": 14175
|
| 4119 |
+
},
|
| 4120 |
+
{
|
| 4121 |
+
"epoch": 15.638766519823788,
|
| 4122 |
+
"grad_norm": 0.36057594418525696,
|
| 4123 |
+
"learning_rate": 5.948717948717949e-06,
|
| 4124 |
+
"loss": 0.0112,
|
| 4125 |
+
"step": 14200
|
| 4126 |
+
},
|
| 4127 |
+
{
|
| 4128 |
+
"epoch": 15.666299559471366,
|
| 4129 |
+
"grad_norm": 0.09728468954563141,
|
| 4130 |
+
"learning_rate": 5.923076923076924e-06,
|
| 4131 |
+
"loss": 0.007,
|
| 4132 |
+
"step": 14225
|
| 4133 |
+
},
|
| 4134 |
+
{
|
| 4135 |
+
"epoch": 15.693832599118943,
|
| 4136 |
+
"grad_norm": 0.2802368402481079,
|
| 4137 |
+
"learning_rate": 5.897435897435898e-06,
|
| 4138 |
+
"loss": 0.0065,
|
| 4139 |
+
"step": 14250
|
| 4140 |
+
},
|
| 4141 |
+
{
|
| 4142 |
+
"epoch": 15.72136563876652,
|
| 4143 |
+
"grad_norm": 0.05950731784105301,
|
| 4144 |
+
"learning_rate": 5.871794871794873e-06,
|
| 4145 |
+
"loss": 0.0086,
|
| 4146 |
+
"step": 14275
|
| 4147 |
+
},
|
| 4148 |
+
{
|
| 4149 |
+
"epoch": 15.748898678414097,
|
| 4150 |
+
"grad_norm": 0.025011401623487473,
|
| 4151 |
+
"learning_rate": 5.846153846153847e-06,
|
| 4152 |
+
"loss": 0.0042,
|
| 4153 |
+
"step": 14300
|
| 4154 |
+
},
|
| 4155 |
+
{
|
| 4156 |
+
"epoch": 15.776431718061675,
|
| 4157 |
+
"grad_norm": 0.037724148482084274,
|
| 4158 |
+
"learning_rate": 5.820512820512822e-06,
|
| 4159 |
+
"loss": 0.0043,
|
| 4160 |
+
"step": 14325
|
| 4161 |
+
},
|
| 4162 |
+
{
|
| 4163 |
+
"epoch": 15.80396475770925,
|
| 4164 |
+
"grad_norm": 0.05708827078342438,
|
| 4165 |
+
"learning_rate": 5.794871794871796e-06,
|
| 4166 |
+
"loss": 0.0051,
|
| 4167 |
+
"step": 14350
|
| 4168 |
+
},
|
| 4169 |
+
{
|
| 4170 |
+
"epoch": 15.831497797356828,
|
| 4171 |
+
"grad_norm": 0.1876123994588852,
|
| 4172 |
+
"learning_rate": 5.769230769230769e-06,
|
| 4173 |
+
"loss": 0.0044,
|
| 4174 |
+
"step": 14375
|
| 4175 |
+
},
|
| 4176 |
+
{
|
| 4177 |
+
"epoch": 15.859030837004406,
|
| 4178 |
+
"grad_norm": 0.059942033141851425,
|
| 4179 |
+
"learning_rate": 5.743589743589743e-06,
|
| 4180 |
+
"loss": 0.0049,
|
| 4181 |
+
"step": 14400
|
| 4182 |
+
},
|
| 4183 |
+
{
|
| 4184 |
+
"epoch": 15.886563876651982,
|
| 4185 |
+
"grad_norm": 0.10881105065345764,
|
| 4186 |
+
"learning_rate": 5.717948717948718e-06,
|
| 4187 |
+
"loss": 0.0053,
|
| 4188 |
+
"step": 14425
|
| 4189 |
+
},
|
| 4190 |
+
{
|
| 4191 |
+
"epoch": 15.91409691629956,
|
| 4192 |
+
"grad_norm": 0.22075070440769196,
|
| 4193 |
+
"learning_rate": 5.692307692307692e-06,
|
| 4194 |
+
"loss": 0.0055,
|
| 4195 |
+
"step": 14450
|
| 4196 |
+
},
|
| 4197 |
+
{
|
| 4198 |
+
"epoch": 15.941629955947137,
|
| 4199 |
+
"grad_norm": 0.2150040864944458,
|
| 4200 |
+
"learning_rate": 5.666666666666667e-06,
|
| 4201 |
+
"loss": 0.0077,
|
| 4202 |
+
"step": 14475
|
| 4203 |
+
},
|
| 4204 |
+
{
|
| 4205 |
+
"epoch": 15.969162995594713,
|
| 4206 |
+
"grad_norm": 0.259221613407135,
|
| 4207 |
+
"learning_rate": 5.641025641025641e-06,
|
| 4208 |
+
"loss": 0.0044,
|
| 4209 |
+
"step": 14500
|
| 4210 |
+
},
|
| 4211 |
+
{
|
| 4212 |
+
"epoch": 15.996696035242291,
|
| 4213 |
+
"grad_norm": 0.08321662992238998,
|
| 4214 |
+
"learning_rate": 5.615384615384616e-06,
|
| 4215 |
+
"loss": 0.0052,
|
| 4216 |
+
"step": 14525
|
| 4217 |
+
},
|
| 4218 |
+
{
|
| 4219 |
+
"epoch": 16.024229074889867,
|
| 4220 |
+
"grad_norm": 0.07159877568483353,
|
| 4221 |
+
"learning_rate": 5.58974358974359e-06,
|
| 4222 |
+
"loss": 0.0043,
|
| 4223 |
+
"step": 14550
|
| 4224 |
+
},
|
| 4225 |
+
{
|
| 4226 |
+
"epoch": 16.051762114537446,
|
| 4227 |
+
"grad_norm": 0.04299210384488106,
|
| 4228 |
+
"learning_rate": 5.564102564102565e-06,
|
| 4229 |
+
"loss": 0.0061,
|
| 4230 |
+
"step": 14575
|
| 4231 |
+
},
|
| 4232 |
+
{
|
| 4233 |
+
"epoch": 16.079295154185022,
|
| 4234 |
+
"grad_norm": 0.013192574493587017,
|
| 4235 |
+
"learning_rate": 5.538461538461539e-06,
|
| 4236 |
+
"loss": 0.0069,
|
| 4237 |
+
"step": 14600
|
| 4238 |
+
},
|
| 4239 |
+
{
|
| 4240 |
+
"epoch": 16.106828193832598,
|
| 4241 |
+
"grad_norm": 0.027391914278268814,
|
| 4242 |
+
"learning_rate": 5.512820512820514e-06,
|
| 4243 |
+
"loss": 0.0038,
|
| 4244 |
+
"step": 14625
|
| 4245 |
+
},
|
| 4246 |
+
{
|
| 4247 |
+
"epoch": 16.134361233480178,
|
| 4248 |
+
"grad_norm": 0.04832287132740021,
|
| 4249 |
+
"learning_rate": 5.487179487179488e-06,
|
| 4250 |
+
"loss": 0.0048,
|
| 4251 |
+
"step": 14650
|
| 4252 |
+
},
|
| 4253 |
+
{
|
| 4254 |
+
"epoch": 16.161894273127754,
|
| 4255 |
+
"grad_norm": 0.18682819604873657,
|
| 4256 |
+
"learning_rate": 5.461538461538461e-06,
|
| 4257 |
+
"loss": 0.0039,
|
| 4258 |
+
"step": 14675
|
| 4259 |
+
},
|
| 4260 |
+
{
|
| 4261 |
+
"epoch": 16.18942731277533,
|
| 4262 |
+
"grad_norm": 0.0538317896425724,
|
| 4263 |
+
"learning_rate": 5.435897435897436e-06,
|
| 4264 |
+
"loss": 0.0061,
|
| 4265 |
+
"step": 14700
|
| 4266 |
+
},
|
| 4267 |
+
{
|
| 4268 |
+
"epoch": 16.21696035242291,
|
| 4269 |
+
"grad_norm": 0.024896269664168358,
|
| 4270 |
+
"learning_rate": 5.41025641025641e-06,
|
| 4271 |
+
"loss": 0.0063,
|
| 4272 |
+
"step": 14725
|
| 4273 |
+
},
|
| 4274 |
+
{
|
| 4275 |
+
"epoch": 16.244493392070485,
|
| 4276 |
+
"grad_norm": 0.0841188058257103,
|
| 4277 |
+
"learning_rate": 5.384615384615385e-06,
|
| 4278 |
+
"loss": 0.0033,
|
| 4279 |
+
"step": 14750
|
| 4280 |
+
},
|
| 4281 |
+
{
|
| 4282 |
+
"epoch": 16.27202643171806,
|
| 4283 |
+
"grad_norm": 0.04172555357217789,
|
| 4284 |
+
"learning_rate": 5.358974358974359e-06,
|
| 4285 |
+
"loss": 0.0057,
|
| 4286 |
+
"step": 14775
|
| 4287 |
+
},
|
| 4288 |
+
{
|
| 4289 |
+
"epoch": 16.29955947136564,
|
| 4290 |
+
"grad_norm": 0.026171807199716568,
|
| 4291 |
+
"learning_rate": 5.333333333333334e-06,
|
| 4292 |
+
"loss": 0.0056,
|
| 4293 |
+
"step": 14800
|
| 4294 |
+
},
|
| 4295 |
+
{
|
| 4296 |
+
"epoch": 16.327092511013216,
|
| 4297 |
+
"grad_norm": 0.3256167471408844,
|
| 4298 |
+
"learning_rate": 5.307692307692308e-06,
|
| 4299 |
+
"loss": 0.0065,
|
| 4300 |
+
"step": 14825
|
| 4301 |
+
},
|
| 4302 |
+
{
|
| 4303 |
+
"epoch": 16.354625550660792,
|
| 4304 |
+
"grad_norm": 0.08626765012741089,
|
| 4305 |
+
"learning_rate": 5.282051282051283e-06,
|
| 4306 |
+
"loss": 0.0031,
|
| 4307 |
+
"step": 14850
|
| 4308 |
+
},
|
| 4309 |
+
{
|
| 4310 |
+
"epoch": 16.38215859030837,
|
| 4311 |
+
"grad_norm": 0.02025369182229042,
|
| 4312 |
+
"learning_rate": 5.256410256410257e-06,
|
| 4313 |
+
"loss": 0.0052,
|
| 4314 |
+
"step": 14875
|
| 4315 |
+
},
|
| 4316 |
+
{
|
| 4317 |
+
"epoch": 16.409691629955947,
|
| 4318 |
+
"grad_norm": 0.030549725517630577,
|
| 4319 |
+
"learning_rate": 5.230769230769232e-06,
|
| 4320 |
+
"loss": 0.0045,
|
| 4321 |
+
"step": 14900
|
| 4322 |
+
},
|
| 4323 |
+
{
|
| 4324 |
+
"epoch": 16.437224669603523,
|
| 4325 |
+
"grad_norm": 0.03368716686964035,
|
| 4326 |
+
"learning_rate": 5.205128205128206e-06,
|
| 4327 |
+
"loss": 0.0043,
|
| 4328 |
+
"step": 14925
|
| 4329 |
+
},
|
| 4330 |
+
{
|
| 4331 |
+
"epoch": 16.464757709251103,
|
| 4332 |
+
"grad_norm": 0.35517045855522156,
|
| 4333 |
+
"learning_rate": 5.179487179487181e-06,
|
| 4334 |
+
"loss": 0.0065,
|
| 4335 |
+
"step": 14950
|
| 4336 |
+
},
|
| 4337 |
+
{
|
| 4338 |
+
"epoch": 16.49229074889868,
|
| 4339 |
+
"grad_norm": 0.01714472658932209,
|
| 4340 |
+
"learning_rate": 5.1538461538461534e-06,
|
| 4341 |
+
"loss": 0.0063,
|
| 4342 |
+
"step": 14975
|
| 4343 |
+
},
|
| 4344 |
+
{
|
| 4345 |
+
"epoch": 16.519823788546255,
|
| 4346 |
+
"grad_norm": 1.1677449941635132,
|
| 4347 |
+
"learning_rate": 5.128205128205128e-06,
|
| 4348 |
+
"loss": 0.0109,
|
| 4349 |
+
"step": 15000
|
| 4350 |
+
},
|
| 4351 |
+
{
|
| 4352 |
+
"epoch": 16.519823788546255,
|
| 4353 |
+
"eval_cer": 24.621416207169304,
|
| 4354 |
+
"eval_loss": 0.8773795962333679,
|
| 4355 |
+
"eval_runtime": 1704.8877,
|
| 4356 |
+
"eval_samples_per_second": 6.206,
|
| 4357 |
+
"eval_steps_per_second": 1.552,
|
| 4358 |
+
"eval_wer": 84.67703913248468,
|
| 4359 |
+
"step": 15000
|
| 4360 |
}
|
| 4361 |
],
|
| 4362 |
"logging_steps": 25,
|
|
|
|
| 4376 |
"attributes": {}
|
| 4377 |
}
|
| 4378 |
},
|
| 4379 |
+
"total_flos": 2.5772147797351465e+20,
|
| 4380 |
"train_batch_size": 4,
|
| 4381 |
"trial_name": null,
|
| 4382 |
"trial_params": null
|