Training in progress, step 10000, checkpoint
Browse files- last-checkpoint/global_step10000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt +3 -0
- last-checkpoint/global_step10000/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 +2035 -5
last-checkpoint/global_step10000/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:adab3b00ea21e4478744ddcff0db1221fab1769952ecb555cb359cb26742d43b
|
| 3 |
+
size 5117197020
|
last-checkpoint/global_step10000/mp_rank_00_model_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3772db65d20e3a1c30f97b9cb12603abd67d2b285e8fee06bfb0b019c0543d25
|
| 3 |
+
size 859127504
|
last-checkpoint/latest
CHANGED
|
@@ -1 +1 @@
|
|
| 1 |
-
|
|
|
|
| 1 |
+
global_step10000
|
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:d2e18aa6175dc70c57ec1ac7704cdfa7862fdb3d72cd69f59bef3dd34ba288be
|
| 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:e451b480b233580c5c6b16cef40a41ba3c2d5da3b6ac8e0c30ef4d8ee0bfacf1
|
| 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 sha256:4bfabff5a7af861fca0d4328cbdeeac43000e633dd64effe96ee68fd4be8bd44
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric":
|
| 3 |
-
"best_model_checkpoint": "./iteboshi_temp/checkpoint-
|
| 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,
|
|
@@ -877,6 +877,2036 @@
|
|
| 877 |
"eval_steps_per_second": 1.513,
|
| 878 |
"eval_wer": 87.76991984912777,
|
| 879 |
"step": 3000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 880 |
}
|
| 881 |
],
|
| 882 |
"logging_steps": 25,
|
|
@@ -896,7 +2926,7 @@
|
|
| 896 |
"attributes": {}
|
| 897 |
}
|
| 898 |
},
|
| 899 |
-
"total_flos":
|
| 900 |
"train_batch_size": 4,
|
| 901 |
"trial_name": null,
|
| 902 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 83.80009429514381,
|
| 3 |
+
"best_model_checkpoint": "./iteboshi_temp/checkpoint-10000",
|
| 4 |
+
"epoch": 11.013215859030836,
|
| 5 |
"eval_steps": 1000,
|
| 6 |
+
"global_step": 10000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 877 |
"eval_steps_per_second": 1.513,
|
| 878 |
"eval_wer": 87.76991984912777,
|
| 879 |
"step": 3000
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 3.3314977973568283,
|
| 883 |
+
"grad_norm": 1.067468285560608,
|
| 884 |
+
"learning_rate": 1.741025641025641e-05,
|
| 885 |
+
"loss": 0.3583,
|
| 886 |
+
"step": 3025
|
| 887 |
+
},
|
| 888 |
+
{
|
| 889 |
+
"epoch": 3.359030837004405,
|
| 890 |
+
"grad_norm": 0.8070423603057861,
|
| 891 |
+
"learning_rate": 1.7384615384615385e-05,
|
| 892 |
+
"loss": 0.3962,
|
| 893 |
+
"step": 3050
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 3.3865638766519823,
|
| 897 |
+
"grad_norm": 0.9048041105270386,
|
| 898 |
+
"learning_rate": 1.7358974358974362e-05,
|
| 899 |
+
"loss": 0.3524,
|
| 900 |
+
"step": 3075
|
| 901 |
+
},
|
| 902 |
+
{
|
| 903 |
+
"epoch": 3.4140969162995596,
|
| 904 |
+
"grad_norm": 0.8543446063995361,
|
| 905 |
+
"learning_rate": 1.7333333333333336e-05,
|
| 906 |
+
"loss": 0.4073,
|
| 907 |
+
"step": 3100
|
| 908 |
+
},
|
| 909 |
+
{
|
| 910 |
+
"epoch": 3.4416299559471364,
|
| 911 |
+
"grad_norm": 0.9092735648155212,
|
| 912 |
+
"learning_rate": 1.730769230769231e-05,
|
| 913 |
+
"loss": 0.3792,
|
| 914 |
+
"step": 3125
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 3.4691629955947136,
|
| 918 |
+
"grad_norm": 0.8556696772575378,
|
| 919 |
+
"learning_rate": 1.7282051282051283e-05,
|
| 920 |
+
"loss": 0.3548,
|
| 921 |
+
"step": 3150
|
| 922 |
+
},
|
| 923 |
+
{
|
| 924 |
+
"epoch": 3.496696035242291,
|
| 925 |
+
"grad_norm": 0.8062282800674438,
|
| 926 |
+
"learning_rate": 1.725641025641026e-05,
|
| 927 |
+
"loss": 0.3695,
|
| 928 |
+
"step": 3175
|
| 929 |
+
},
|
| 930 |
+
{
|
| 931 |
+
"epoch": 3.5242290748898677,
|
| 932 |
+
"grad_norm": 0.8389378190040588,
|
| 933 |
+
"learning_rate": 1.7230769230769234e-05,
|
| 934 |
+
"loss": 0.3795,
|
| 935 |
+
"step": 3200
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 3.551762114537445,
|
| 939 |
+
"grad_norm": 0.866780698299408,
|
| 940 |
+
"learning_rate": 1.7205128205128207e-05,
|
| 941 |
+
"loss": 0.3535,
|
| 942 |
+
"step": 3225
|
| 943 |
+
},
|
| 944 |
+
{
|
| 945 |
+
"epoch": 3.579295154185022,
|
| 946 |
+
"grad_norm": 1.140093207359314,
|
| 947 |
+
"learning_rate": 1.717948717948718e-05,
|
| 948 |
+
"loss": 0.3453,
|
| 949 |
+
"step": 3250
|
| 950 |
+
},
|
| 951 |
+
{
|
| 952 |
+
"epoch": 3.606828193832599,
|
| 953 |
+
"grad_norm": 0.7346594929695129,
|
| 954 |
+
"learning_rate": 1.7153846153846155e-05,
|
| 955 |
+
"loss": 0.3768,
|
| 956 |
+
"step": 3275
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 3.6343612334801763,
|
| 960 |
+
"grad_norm": 0.9942306280136108,
|
| 961 |
+
"learning_rate": 1.7128205128205128e-05,
|
| 962 |
+
"loss": 0.3843,
|
| 963 |
+
"step": 3300
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 3.6618942731277535,
|
| 967 |
+
"grad_norm": 0.8022450804710388,
|
| 968 |
+
"learning_rate": 1.7102564102564102e-05,
|
| 969 |
+
"loss": 0.3319,
|
| 970 |
+
"step": 3325
|
| 971 |
+
},
|
| 972 |
+
{
|
| 973 |
+
"epoch": 3.6894273127753303,
|
| 974 |
+
"grad_norm": 0.7250021696090698,
|
| 975 |
+
"learning_rate": 1.707692307692308e-05,
|
| 976 |
+
"loss": 0.3664,
|
| 977 |
+
"step": 3350
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 3.7169603524229076,
|
| 981 |
+
"grad_norm": 0.8458446264266968,
|
| 982 |
+
"learning_rate": 1.7051282051282053e-05,
|
| 983 |
+
"loss": 0.32,
|
| 984 |
+
"step": 3375
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 3.744493392070485,
|
| 988 |
+
"grad_norm": 0.8499473333358765,
|
| 989 |
+
"learning_rate": 1.7025641025641026e-05,
|
| 990 |
+
"loss": 0.365,
|
| 991 |
+
"step": 3400
|
| 992 |
+
},
|
| 993 |
+
{
|
| 994 |
+
"epoch": 3.7720264317180616,
|
| 995 |
+
"grad_norm": 1.18294095993042,
|
| 996 |
+
"learning_rate": 1.7e-05,
|
| 997 |
+
"loss": 0.3453,
|
| 998 |
+
"step": 3425
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 3.799559471365639,
|
| 1002 |
+
"grad_norm": 0.7673613429069519,
|
| 1003 |
+
"learning_rate": 1.6974358974358977e-05,
|
| 1004 |
+
"loss": 0.3406,
|
| 1005 |
+
"step": 3450
|
| 1006 |
+
},
|
| 1007 |
+
{
|
| 1008 |
+
"epoch": 3.827092511013216,
|
| 1009 |
+
"grad_norm": 0.8733392953872681,
|
| 1010 |
+
"learning_rate": 1.694871794871795e-05,
|
| 1011 |
+
"loss": 0.3406,
|
| 1012 |
+
"step": 3475
|
| 1013 |
+
},
|
| 1014 |
+
{
|
| 1015 |
+
"epoch": 3.854625550660793,
|
| 1016 |
+
"grad_norm": 0.9603780508041382,
|
| 1017 |
+
"learning_rate": 1.6923076923076924e-05,
|
| 1018 |
+
"loss": 0.4131,
|
| 1019 |
+
"step": 3500
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 3.88215859030837,
|
| 1023 |
+
"grad_norm": 1.0154445171356201,
|
| 1024 |
+
"learning_rate": 1.6897435897435898e-05,
|
| 1025 |
+
"loss": 0.3401,
|
| 1026 |
+
"step": 3525
|
| 1027 |
+
},
|
| 1028 |
+
{
|
| 1029 |
+
"epoch": 3.909691629955947,
|
| 1030 |
+
"grad_norm": 0.7912996411323547,
|
| 1031 |
+
"learning_rate": 1.687179487179487e-05,
|
| 1032 |
+
"loss": 0.3621,
|
| 1033 |
+
"step": 3550
|
| 1034 |
+
},
|
| 1035 |
+
{
|
| 1036 |
+
"epoch": 3.9372246696035242,
|
| 1037 |
+
"grad_norm": 0.7029661536216736,
|
| 1038 |
+
"learning_rate": 1.684615384615385e-05,
|
| 1039 |
+
"loss": 0.3848,
|
| 1040 |
+
"step": 3575
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 3.964757709251101,
|
| 1044 |
+
"grad_norm": 0.7775823473930359,
|
| 1045 |
+
"learning_rate": 1.6820512820512822e-05,
|
| 1046 |
+
"loss": 0.3753,
|
| 1047 |
+
"step": 3600
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"epoch": 3.9922907488986783,
|
| 1051 |
+
"grad_norm": 0.8819046020507812,
|
| 1052 |
+
"learning_rate": 1.6794871794871796e-05,
|
| 1053 |
+
"loss": 0.373,
|
| 1054 |
+
"step": 3625
|
| 1055 |
+
},
|
| 1056 |
+
{
|
| 1057 |
+
"epoch": 4.0198237885462555,
|
| 1058 |
+
"grad_norm": 0.6218374967575073,
|
| 1059 |
+
"learning_rate": 1.676923076923077e-05,
|
| 1060 |
+
"loss": 0.2521,
|
| 1061 |
+
"step": 3650
|
| 1062 |
+
},
|
| 1063 |
+
{
|
| 1064 |
+
"epoch": 4.047356828193832,
|
| 1065 |
+
"grad_norm": 0.9981600642204285,
|
| 1066 |
+
"learning_rate": 1.6743589743589747e-05,
|
| 1067 |
+
"loss": 0.238,
|
| 1068 |
+
"step": 3675
|
| 1069 |
+
},
|
| 1070 |
+
{
|
| 1071 |
+
"epoch": 4.07488986784141,
|
| 1072 |
+
"grad_norm": 0.6263682246208191,
|
| 1073 |
+
"learning_rate": 1.671794871794872e-05,
|
| 1074 |
+
"loss": 0.2172,
|
| 1075 |
+
"step": 3700
|
| 1076 |
+
},
|
| 1077 |
+
{
|
| 1078 |
+
"epoch": 4.102422907488987,
|
| 1079 |
+
"grad_norm": 0.5469574332237244,
|
| 1080 |
+
"learning_rate": 1.6692307692307694e-05,
|
| 1081 |
+
"loss": 0.2553,
|
| 1082 |
+
"step": 3725
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 4.129955947136564,
|
| 1086 |
+
"grad_norm": 0.9186758399009705,
|
| 1087 |
+
"learning_rate": 1.6666666666666667e-05,
|
| 1088 |
+
"loss": 0.2413,
|
| 1089 |
+
"step": 3750
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 4.157488986784141,
|
| 1093 |
+
"grad_norm": 0.5088458061218262,
|
| 1094 |
+
"learning_rate": 1.6641025641025645e-05,
|
| 1095 |
+
"loss": 0.2303,
|
| 1096 |
+
"step": 3775
|
| 1097 |
+
},
|
| 1098 |
+
{
|
| 1099 |
+
"epoch": 4.185022026431718,
|
| 1100 |
+
"grad_norm": 0.5243244767189026,
|
| 1101 |
+
"learning_rate": 1.6615384615384618e-05,
|
| 1102 |
+
"loss": 0.1918,
|
| 1103 |
+
"step": 3800
|
| 1104 |
+
},
|
| 1105 |
+
{
|
| 1106 |
+
"epoch": 4.212555066079295,
|
| 1107 |
+
"grad_norm": 0.8335323333740234,
|
| 1108 |
+
"learning_rate": 1.6589743589743592e-05,
|
| 1109 |
+
"loss": 0.2309,
|
| 1110 |
+
"step": 3825
|
| 1111 |
+
},
|
| 1112 |
+
{
|
| 1113 |
+
"epoch": 4.240088105726873,
|
| 1114 |
+
"grad_norm": 0.6145501136779785,
|
| 1115 |
+
"learning_rate": 1.6564102564102565e-05,
|
| 1116 |
+
"loss": 0.226,
|
| 1117 |
+
"step": 3850
|
| 1118 |
+
},
|
| 1119 |
+
{
|
| 1120 |
+
"epoch": 4.2676211453744495,
|
| 1121 |
+
"grad_norm": 0.9030354022979736,
|
| 1122 |
+
"learning_rate": 1.653846153846154e-05,
|
| 1123 |
+
"loss": 0.2569,
|
| 1124 |
+
"step": 3875
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 4.295154185022026,
|
| 1128 |
+
"grad_norm": 0.6755716800689697,
|
| 1129 |
+
"learning_rate": 1.6512820512820513e-05,
|
| 1130 |
+
"loss": 0.2424,
|
| 1131 |
+
"step": 3900
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 4.322687224669604,
|
| 1135 |
+
"grad_norm": 1.1529523134231567,
|
| 1136 |
+
"learning_rate": 1.6487179487179486e-05,
|
| 1137 |
+
"loss": 0.2539,
|
| 1138 |
+
"step": 3925
|
| 1139 |
+
},
|
| 1140 |
+
{
|
| 1141 |
+
"epoch": 4.350220264317181,
|
| 1142 |
+
"grad_norm": 0.689730167388916,
|
| 1143 |
+
"learning_rate": 1.6461538461538463e-05,
|
| 1144 |
+
"loss": 0.2366,
|
| 1145 |
+
"step": 3950
|
| 1146 |
+
},
|
| 1147 |
+
{
|
| 1148 |
+
"epoch": 4.377753303964758,
|
| 1149 |
+
"grad_norm": 0.6913994550704956,
|
| 1150 |
+
"learning_rate": 1.6435897435897437e-05,
|
| 1151 |
+
"loss": 0.2667,
|
| 1152 |
+
"step": 3975
|
| 1153 |
+
},
|
| 1154 |
+
{
|
| 1155 |
+
"epoch": 4.405286343612334,
|
| 1156 |
+
"grad_norm": 0.6518797278404236,
|
| 1157 |
+
"learning_rate": 1.641025641025641e-05,
|
| 1158 |
+
"loss": 0.2445,
|
| 1159 |
+
"step": 4000
|
| 1160 |
+
},
|
| 1161 |
+
{
|
| 1162 |
+
"epoch": 4.405286343612334,
|
| 1163 |
+
"eval_cer": 28.12639712035583,
|
| 1164 |
+
"eval_loss": 0.7661674618721008,
|
| 1165 |
+
"eval_runtime": 1727.6269,
|
| 1166 |
+
"eval_samples_per_second": 6.125,
|
| 1167 |
+
"eval_steps_per_second": 1.532,
|
| 1168 |
+
"eval_wer": 86.67609618104667,
|
| 1169 |
+
"step": 4000
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 4.432819383259912,
|
| 1173 |
+
"grad_norm": 0.7283459305763245,
|
| 1174 |
+
"learning_rate": 1.6384615384615384e-05,
|
| 1175 |
+
"loss": 0.262,
|
| 1176 |
+
"step": 4025
|
| 1177 |
+
},
|
| 1178 |
+
{
|
| 1179 |
+
"epoch": 4.460352422907489,
|
| 1180 |
+
"grad_norm": 0.8054242134094238,
|
| 1181 |
+
"learning_rate": 1.635897435897436e-05,
|
| 1182 |
+
"loss": 0.2685,
|
| 1183 |
+
"step": 4050
|
| 1184 |
+
},
|
| 1185 |
+
{
|
| 1186 |
+
"epoch": 4.487885462555066,
|
| 1187 |
+
"grad_norm": 0.7349271774291992,
|
| 1188 |
+
"learning_rate": 1.6333333333333335e-05,
|
| 1189 |
+
"loss": 0.2636,
|
| 1190 |
+
"step": 4075
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 4.515418502202643,
|
| 1194 |
+
"grad_norm": 0.7914747595787048,
|
| 1195 |
+
"learning_rate": 1.630769230769231e-05,
|
| 1196 |
+
"loss": 0.2241,
|
| 1197 |
+
"step": 4100
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 4.54295154185022,
|
| 1201 |
+
"grad_norm": 0.6789693832397461,
|
| 1202 |
+
"learning_rate": 1.6282051282051282e-05,
|
| 1203 |
+
"loss": 0.2491,
|
| 1204 |
+
"step": 4125
|
| 1205 |
+
},
|
| 1206 |
+
{
|
| 1207 |
+
"epoch": 4.570484581497797,
|
| 1208 |
+
"grad_norm": 0.7700952887535095,
|
| 1209 |
+
"learning_rate": 1.625641025641026e-05,
|
| 1210 |
+
"loss": 0.2608,
|
| 1211 |
+
"step": 4150
|
| 1212 |
+
},
|
| 1213 |
+
{
|
| 1214 |
+
"epoch": 4.598017621145375,
|
| 1215 |
+
"grad_norm": 0.6949133276939392,
|
| 1216 |
+
"learning_rate": 1.6230769230769233e-05,
|
| 1217 |
+
"loss": 0.2542,
|
| 1218 |
+
"step": 4175
|
| 1219 |
+
},
|
| 1220 |
+
{
|
| 1221 |
+
"epoch": 4.6255506607929515,
|
| 1222 |
+
"grad_norm": 0.7408822774887085,
|
| 1223 |
+
"learning_rate": 1.6205128205128207e-05,
|
| 1224 |
+
"loss": 0.2427,
|
| 1225 |
+
"step": 4200
|
| 1226 |
+
},
|
| 1227 |
+
{
|
| 1228 |
+
"epoch": 4.653083700440528,
|
| 1229 |
+
"grad_norm": 0.7268140912055969,
|
| 1230 |
+
"learning_rate": 1.617948717948718e-05,
|
| 1231 |
+
"loss": 0.2473,
|
| 1232 |
+
"step": 4225
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 4.680616740088106,
|
| 1236 |
+
"grad_norm": 1.2149430513381958,
|
| 1237 |
+
"learning_rate": 1.6153846153846154e-05,
|
| 1238 |
+
"loss": 0.2374,
|
| 1239 |
+
"step": 4250
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 4.708149779735683,
|
| 1243 |
+
"grad_norm": 0.8126139044761658,
|
| 1244 |
+
"learning_rate": 1.612820512820513e-05,
|
| 1245 |
+
"loss": 0.2428,
|
| 1246 |
+
"step": 4275
|
| 1247 |
+
},
|
| 1248 |
+
{
|
| 1249 |
+
"epoch": 4.73568281938326,
|
| 1250 |
+
"grad_norm": 0.7101436257362366,
|
| 1251 |
+
"learning_rate": 1.6102564102564105e-05,
|
| 1252 |
+
"loss": 0.2475,
|
| 1253 |
+
"step": 4300
|
| 1254 |
+
},
|
| 1255 |
+
{
|
| 1256 |
+
"epoch": 4.763215859030837,
|
| 1257 |
+
"grad_norm": 0.6345697641372681,
|
| 1258 |
+
"learning_rate": 1.607692307692308e-05,
|
| 1259 |
+
"loss": 0.2432,
|
| 1260 |
+
"step": 4325
|
| 1261 |
+
},
|
| 1262 |
+
{
|
| 1263 |
+
"epoch": 4.790748898678414,
|
| 1264 |
+
"grad_norm": 0.781745195388794,
|
| 1265 |
+
"learning_rate": 1.6051282051282052e-05,
|
| 1266 |
+
"loss": 0.23,
|
| 1267 |
+
"step": 4350
|
| 1268 |
+
},
|
| 1269 |
+
{
|
| 1270 |
+
"epoch": 4.818281938325991,
|
| 1271 |
+
"grad_norm": 1.2168214321136475,
|
| 1272 |
+
"learning_rate": 1.602564102564103e-05,
|
| 1273 |
+
"loss": 0.2563,
|
| 1274 |
+
"step": 4375
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 4.845814977973569,
|
| 1278 |
+
"grad_norm": 0.47594118118286133,
|
| 1279 |
+
"learning_rate": 1.6000000000000003e-05,
|
| 1280 |
+
"loss": 0.2294,
|
| 1281 |
+
"step": 4400
|
| 1282 |
+
},
|
| 1283 |
+
{
|
| 1284 |
+
"epoch": 4.8733480176211454,
|
| 1285 |
+
"grad_norm": 0.4806981384754181,
|
| 1286 |
+
"learning_rate": 1.5974358974358976e-05,
|
| 1287 |
+
"loss": 0.2321,
|
| 1288 |
+
"step": 4425
|
| 1289 |
+
},
|
| 1290 |
+
{
|
| 1291 |
+
"epoch": 4.900881057268722,
|
| 1292 |
+
"grad_norm": 0.7716917991638184,
|
| 1293 |
+
"learning_rate": 1.594871794871795e-05,
|
| 1294 |
+
"loss": 0.2394,
|
| 1295 |
+
"step": 4450
|
| 1296 |
+
},
|
| 1297 |
+
{
|
| 1298 |
+
"epoch": 4.9284140969163,
|
| 1299 |
+
"grad_norm": 0.579592764377594,
|
| 1300 |
+
"learning_rate": 1.5923076923076924e-05,
|
| 1301 |
+
"loss": 0.2245,
|
| 1302 |
+
"step": 4475
|
| 1303 |
+
},
|
| 1304 |
+
{
|
| 1305 |
+
"epoch": 4.955947136563877,
|
| 1306 |
+
"grad_norm": 0.5050383806228638,
|
| 1307 |
+
"learning_rate": 1.5897435897435897e-05,
|
| 1308 |
+
"loss": 0.2357,
|
| 1309 |
+
"step": 4500
|
| 1310 |
+
},
|
| 1311 |
+
{
|
| 1312 |
+
"epoch": 4.983480176211454,
|
| 1313 |
+
"grad_norm": 1.2022616863250732,
|
| 1314 |
+
"learning_rate": 1.587179487179487e-05,
|
| 1315 |
+
"loss": 0.246,
|
| 1316 |
+
"step": 4525
|
| 1317 |
+
},
|
| 1318 |
+
{
|
| 1319 |
+
"epoch": 5.011013215859031,
|
| 1320 |
+
"grad_norm": 0.6758959889411926,
|
| 1321 |
+
"learning_rate": 1.5846153846153848e-05,
|
| 1322 |
+
"loss": 0.1977,
|
| 1323 |
+
"step": 4550
|
| 1324 |
+
},
|
| 1325 |
+
{
|
| 1326 |
+
"epoch": 5.038546255506608,
|
| 1327 |
+
"grad_norm": 0.3403749465942383,
|
| 1328 |
+
"learning_rate": 1.582051282051282e-05,
|
| 1329 |
+
"loss": 0.1768,
|
| 1330 |
+
"step": 4575
|
| 1331 |
+
},
|
| 1332 |
+
{
|
| 1333 |
+
"epoch": 5.066079295154185,
|
| 1334 |
+
"grad_norm": 0.6206667423248291,
|
| 1335 |
+
"learning_rate": 1.5794871794871795e-05,
|
| 1336 |
+
"loss": 0.1403,
|
| 1337 |
+
"step": 4600
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 5.093612334801762,
|
| 1341 |
+
"grad_norm": 0.4241011440753937,
|
| 1342 |
+
"learning_rate": 1.576923076923077e-05,
|
| 1343 |
+
"loss": 0.1857,
|
| 1344 |
+
"step": 4625
|
| 1345 |
+
},
|
| 1346 |
+
{
|
| 1347 |
+
"epoch": 5.121145374449339,
|
| 1348 |
+
"grad_norm": 0.5990819931030273,
|
| 1349 |
+
"learning_rate": 1.5743589743589746e-05,
|
| 1350 |
+
"loss": 0.1732,
|
| 1351 |
+
"step": 4650
|
| 1352 |
+
},
|
| 1353 |
+
{
|
| 1354 |
+
"epoch": 5.148678414096916,
|
| 1355 |
+
"grad_norm": 0.36775925755500793,
|
| 1356 |
+
"learning_rate": 1.571794871794872e-05,
|
| 1357 |
+
"loss": 0.1679,
|
| 1358 |
+
"step": 4675
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 5.176211453744493,
|
| 1362 |
+
"grad_norm": 0.4457094669342041,
|
| 1363 |
+
"learning_rate": 1.5692307692307693e-05,
|
| 1364 |
+
"loss": 0.1422,
|
| 1365 |
+
"step": 4700
|
| 1366 |
+
},
|
| 1367 |
+
{
|
| 1368 |
+
"epoch": 5.203744493392071,
|
| 1369 |
+
"grad_norm": 0.3794902563095093,
|
| 1370 |
+
"learning_rate": 1.5666666666666667e-05,
|
| 1371 |
+
"loss": 0.1602,
|
| 1372 |
+
"step": 4725
|
| 1373 |
+
},
|
| 1374 |
+
{
|
| 1375 |
+
"epoch": 5.2312775330396475,
|
| 1376 |
+
"grad_norm": 0.9086684584617615,
|
| 1377 |
+
"learning_rate": 1.5641025641025644e-05,
|
| 1378 |
+
"loss": 0.1401,
|
| 1379 |
+
"step": 4750
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 5.258810572687224,
|
| 1383 |
+
"grad_norm": 0.6391773223876953,
|
| 1384 |
+
"learning_rate": 1.5615384615384618e-05,
|
| 1385 |
+
"loss": 0.2021,
|
| 1386 |
+
"step": 4775
|
| 1387 |
+
},
|
| 1388 |
+
{
|
| 1389 |
+
"epoch": 5.286343612334802,
|
| 1390 |
+
"grad_norm": 0.36679503321647644,
|
| 1391 |
+
"learning_rate": 1.558974358974359e-05,
|
| 1392 |
+
"loss": 0.1486,
|
| 1393 |
+
"step": 4800
|
| 1394 |
+
},
|
| 1395 |
+
{
|
| 1396 |
+
"epoch": 5.313876651982379,
|
| 1397 |
+
"grad_norm": 0.5840328335762024,
|
| 1398 |
+
"learning_rate": 1.5564102564102565e-05,
|
| 1399 |
+
"loss": 0.1356,
|
| 1400 |
+
"step": 4825
|
| 1401 |
+
},
|
| 1402 |
+
{
|
| 1403 |
+
"epoch": 5.341409691629956,
|
| 1404 |
+
"grad_norm": 0.4502796232700348,
|
| 1405 |
+
"learning_rate": 1.553846153846154e-05,
|
| 1406 |
+
"loss": 0.1549,
|
| 1407 |
+
"step": 4850
|
| 1408 |
+
},
|
| 1409 |
+
{
|
| 1410 |
+
"epoch": 5.368942731277533,
|
| 1411 |
+
"grad_norm": 0.5962740778923035,
|
| 1412 |
+
"learning_rate": 1.5512820512820516e-05,
|
| 1413 |
+
"loss": 0.1529,
|
| 1414 |
+
"step": 4875
|
| 1415 |
+
},
|
| 1416 |
+
{
|
| 1417 |
+
"epoch": 5.39647577092511,
|
| 1418 |
+
"grad_norm": 0.695899486541748,
|
| 1419 |
+
"learning_rate": 1.548717948717949e-05,
|
| 1420 |
+
"loss": 0.1456,
|
| 1421 |
+
"step": 4900
|
| 1422 |
+
},
|
| 1423 |
+
{
|
| 1424 |
+
"epoch": 5.424008810572687,
|
| 1425 |
+
"grad_norm": 0.5500043630599976,
|
| 1426 |
+
"learning_rate": 1.5461538461538463e-05,
|
| 1427 |
+
"loss": 0.1991,
|
| 1428 |
+
"step": 4925
|
| 1429 |
+
},
|
| 1430 |
+
{
|
| 1431 |
+
"epoch": 5.451541850220265,
|
| 1432 |
+
"grad_norm": 0.4101676642894745,
|
| 1433 |
+
"learning_rate": 1.5435897435897436e-05,
|
| 1434 |
+
"loss": 0.1708,
|
| 1435 |
+
"step": 4950
|
| 1436 |
+
},
|
| 1437 |
+
{
|
| 1438 |
+
"epoch": 5.479074889867841,
|
| 1439 |
+
"grad_norm": 0.5245118141174316,
|
| 1440 |
+
"learning_rate": 1.5410256410256414e-05,
|
| 1441 |
+
"loss": 0.1495,
|
| 1442 |
+
"step": 4975
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 5.506607929515418,
|
| 1446 |
+
"grad_norm": 0.9378005266189575,
|
| 1447 |
+
"learning_rate": 1.5384615384615387e-05,
|
| 1448 |
+
"loss": 0.1548,
|
| 1449 |
+
"step": 5000
|
| 1450 |
+
},
|
| 1451 |
+
{
|
| 1452 |
+
"epoch": 5.506607929515418,
|
| 1453 |
+
"eval_cer": 27.874772161911203,
|
| 1454 |
+
"eval_loss": 0.7709037661552429,
|
| 1455 |
+
"eval_runtime": 1715.4949,
|
| 1456 |
+
"eval_samples_per_second": 6.168,
|
| 1457 |
+
"eval_steps_per_second": 1.542,
|
| 1458 |
+
"eval_wer": 86.6006600660066,
|
| 1459 |
+
"step": 5000
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 5.534140969162996,
|
| 1463 |
+
"grad_norm": 0.5195850133895874,
|
| 1464 |
+
"learning_rate": 1.535897435897436e-05,
|
| 1465 |
+
"loss": 0.159,
|
| 1466 |
+
"step": 5025
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 5.561674008810573,
|
| 1470 |
+
"grad_norm": 0.5712270140647888,
|
| 1471 |
+
"learning_rate": 1.5333333333333334e-05,
|
| 1472 |
+
"loss": 0.184,
|
| 1473 |
+
"step": 5050
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 5.5892070484581495,
|
| 1477 |
+
"grad_norm": 0.5652280449867249,
|
| 1478 |
+
"learning_rate": 1.5307692307692308e-05,
|
| 1479 |
+
"loss": 0.1645,
|
| 1480 |
+
"step": 5075
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 5.616740088105727,
|
| 1484 |
+
"grad_norm": 0.6352373957633972,
|
| 1485 |
+
"learning_rate": 1.5282051282051282e-05,
|
| 1486 |
+
"loss": 0.1561,
|
| 1487 |
+
"step": 5100
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 5.644273127753304,
|
| 1491 |
+
"grad_norm": 0.9260919690132141,
|
| 1492 |
+
"learning_rate": 1.5256410256410257e-05,
|
| 1493 |
+
"loss": 0.1653,
|
| 1494 |
+
"step": 5125
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 5.671806167400881,
|
| 1498 |
+
"grad_norm": 0.5775518417358398,
|
| 1499 |
+
"learning_rate": 1.523076923076923e-05,
|
| 1500 |
+
"loss": 0.1767,
|
| 1501 |
+
"step": 5150
|
| 1502 |
+
},
|
| 1503 |
+
{
|
| 1504 |
+
"epoch": 5.6993392070484585,
|
| 1505 |
+
"grad_norm": 0.4431408941745758,
|
| 1506 |
+
"learning_rate": 1.5205128205128206e-05,
|
| 1507 |
+
"loss": 0.1365,
|
| 1508 |
+
"step": 5175
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 5.726872246696035,
|
| 1512 |
+
"grad_norm": 0.4839853346347809,
|
| 1513 |
+
"learning_rate": 1.517948717948718e-05,
|
| 1514 |
+
"loss": 0.1932,
|
| 1515 |
+
"step": 5200
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 5.754405286343612,
|
| 1519 |
+
"grad_norm": 0.4924067258834839,
|
| 1520 |
+
"learning_rate": 1.5153846153846155e-05,
|
| 1521 |
+
"loss": 0.1506,
|
| 1522 |
+
"step": 5225
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 5.78193832599119,
|
| 1526 |
+
"grad_norm": 0.7064109444618225,
|
| 1527 |
+
"learning_rate": 1.5128205128205129e-05,
|
| 1528 |
+
"loss": 0.1899,
|
| 1529 |
+
"step": 5250
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 5.809471365638767,
|
| 1533 |
+
"grad_norm": 0.4375017285346985,
|
| 1534 |
+
"learning_rate": 1.5102564102564104e-05,
|
| 1535 |
+
"loss": 0.1629,
|
| 1536 |
+
"step": 5275
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 5.8370044052863435,
|
| 1540 |
+
"grad_norm": 0.7590004205703735,
|
| 1541 |
+
"learning_rate": 1.5076923076923078e-05,
|
| 1542 |
+
"loss": 0.1555,
|
| 1543 |
+
"step": 5300
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 5.864537444933921,
|
| 1547 |
+
"grad_norm": 0.704337477684021,
|
| 1548 |
+
"learning_rate": 1.5051282051282053e-05,
|
| 1549 |
+
"loss": 0.1603,
|
| 1550 |
+
"step": 5325
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 5.892070484581498,
|
| 1554 |
+
"grad_norm": 0.6816414594650269,
|
| 1555 |
+
"learning_rate": 1.5025641025641027e-05,
|
| 1556 |
+
"loss": 0.1753,
|
| 1557 |
+
"step": 5350
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 5.919603524229075,
|
| 1561 |
+
"grad_norm": 0.49803662300109863,
|
| 1562 |
+
"learning_rate": 1.5000000000000002e-05,
|
| 1563 |
+
"loss": 0.1517,
|
| 1564 |
+
"step": 5375
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 5.9471365638766525,
|
| 1568 |
+
"grad_norm": 1.4683668613433838,
|
| 1569 |
+
"learning_rate": 1.4974358974358976e-05,
|
| 1570 |
+
"loss": 0.1656,
|
| 1571 |
+
"step": 5400
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 5.974669603524229,
|
| 1575 |
+
"grad_norm": 0.5251373648643494,
|
| 1576 |
+
"learning_rate": 1.494871794871795e-05,
|
| 1577 |
+
"loss": 0.1454,
|
| 1578 |
+
"step": 5425
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 6.002202643171806,
|
| 1582 |
+
"grad_norm": 0.39470919966697693,
|
| 1583 |
+
"learning_rate": 1.4923076923076925e-05,
|
| 1584 |
+
"loss": 0.1317,
|
| 1585 |
+
"step": 5450
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 6.029735682819383,
|
| 1589 |
+
"grad_norm": 0.4022665321826935,
|
| 1590 |
+
"learning_rate": 1.4897435897435898e-05,
|
| 1591 |
+
"loss": 0.1069,
|
| 1592 |
+
"step": 5475
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 6.057268722466961,
|
| 1596 |
+
"grad_norm": 0.462735116481781,
|
| 1597 |
+
"learning_rate": 1.4871794871794874e-05,
|
| 1598 |
+
"loss": 0.0833,
|
| 1599 |
+
"step": 5500
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 6.084801762114537,
|
| 1603 |
+
"grad_norm": 0.5390641689300537,
|
| 1604 |
+
"learning_rate": 1.4846153846153847e-05,
|
| 1605 |
+
"loss": 0.1123,
|
| 1606 |
+
"step": 5525
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 6.112334801762114,
|
| 1610 |
+
"grad_norm": 0.34844598174095154,
|
| 1611 |
+
"learning_rate": 1.4820512820512823e-05,
|
| 1612 |
+
"loss": 0.1198,
|
| 1613 |
+
"step": 5550
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 6.139867841409692,
|
| 1617 |
+
"grad_norm": 0.3289618492126465,
|
| 1618 |
+
"learning_rate": 1.4794871794871796e-05,
|
| 1619 |
+
"loss": 0.1119,
|
| 1620 |
+
"step": 5575
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 6.167400881057269,
|
| 1624 |
+
"grad_norm": 0.5920740962028503,
|
| 1625 |
+
"learning_rate": 1.4769230769230772e-05,
|
| 1626 |
+
"loss": 0.1185,
|
| 1627 |
+
"step": 5600
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 6.1949339207048455,
|
| 1631 |
+
"grad_norm": 0.3685765862464905,
|
| 1632 |
+
"learning_rate": 1.4743589743589745e-05,
|
| 1633 |
+
"loss": 0.0869,
|
| 1634 |
+
"step": 5625
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 6.222466960352423,
|
| 1638 |
+
"grad_norm": 0.6981714367866516,
|
| 1639 |
+
"learning_rate": 1.471794871794872e-05,
|
| 1640 |
+
"loss": 0.1187,
|
| 1641 |
+
"step": 5650
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 6.25,
|
| 1645 |
+
"grad_norm": 0.5181892514228821,
|
| 1646 |
+
"learning_rate": 1.4692307692307694e-05,
|
| 1647 |
+
"loss": 0.1144,
|
| 1648 |
+
"step": 5675
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 6.277533039647577,
|
| 1652 |
+
"grad_norm": 0.5749639272689819,
|
| 1653 |
+
"learning_rate": 1.4666666666666666e-05,
|
| 1654 |
+
"loss": 0.082,
|
| 1655 |
+
"step": 5700
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 6.3050660792951545,
|
| 1659 |
+
"grad_norm": 0.34826797246932983,
|
| 1660 |
+
"learning_rate": 1.4641025641025642e-05,
|
| 1661 |
+
"loss": 0.1196,
|
| 1662 |
+
"step": 5725
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 6.332599118942731,
|
| 1666 |
+
"grad_norm": 0.4354853332042694,
|
| 1667 |
+
"learning_rate": 1.4615384615384615e-05,
|
| 1668 |
+
"loss": 0.0996,
|
| 1669 |
+
"step": 5750
|
| 1670 |
+
},
|
| 1671 |
+
{
|
| 1672 |
+
"epoch": 6.360132158590308,
|
| 1673 |
+
"grad_norm": 0.4485156238079071,
|
| 1674 |
+
"learning_rate": 1.458974358974359e-05,
|
| 1675 |
+
"loss": 0.1012,
|
| 1676 |
+
"step": 5775
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 6.387665198237886,
|
| 1680 |
+
"grad_norm": 1.4070178270339966,
|
| 1681 |
+
"learning_rate": 1.4564102564102564e-05,
|
| 1682 |
+
"loss": 0.1116,
|
| 1683 |
+
"step": 5800
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 6.415198237885463,
|
| 1687 |
+
"grad_norm": 0.5709951519966125,
|
| 1688 |
+
"learning_rate": 1.453846153846154e-05,
|
| 1689 |
+
"loss": 0.0992,
|
| 1690 |
+
"step": 5825
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"epoch": 6.442731277533039,
|
| 1694 |
+
"grad_norm": 0.7376241087913513,
|
| 1695 |
+
"learning_rate": 1.4512820512820513e-05,
|
| 1696 |
+
"loss": 0.1296,
|
| 1697 |
+
"step": 5850
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 6.470264317180617,
|
| 1701 |
+
"grad_norm": 0.5828295350074768,
|
| 1702 |
+
"learning_rate": 1.4487179487179489e-05,
|
| 1703 |
+
"loss": 0.1363,
|
| 1704 |
+
"step": 5875
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 6.497797356828194,
|
| 1708 |
+
"grad_norm": 0.44842031598091125,
|
| 1709 |
+
"learning_rate": 1.4461538461538462e-05,
|
| 1710 |
+
"loss": 0.1186,
|
| 1711 |
+
"step": 5900
|
| 1712 |
+
},
|
| 1713 |
+
{
|
| 1714 |
+
"epoch": 6.525330396475771,
|
| 1715 |
+
"grad_norm": 0.5105710625648499,
|
| 1716 |
+
"learning_rate": 1.4435897435897438e-05,
|
| 1717 |
+
"loss": 0.1319,
|
| 1718 |
+
"step": 5925
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 6.5528634361233475,
|
| 1722 |
+
"grad_norm": 0.45816680788993835,
|
| 1723 |
+
"learning_rate": 1.4410256410256411e-05,
|
| 1724 |
+
"loss": 0.1169,
|
| 1725 |
+
"step": 5950
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 6.580396475770925,
|
| 1729 |
+
"grad_norm": 0.5432708859443665,
|
| 1730 |
+
"learning_rate": 1.4384615384615387e-05,
|
| 1731 |
+
"loss": 0.1163,
|
| 1732 |
+
"step": 5975
|
| 1733 |
+
},
|
| 1734 |
+
{
|
| 1735 |
+
"epoch": 6.607929515418502,
|
| 1736 |
+
"grad_norm": 0.5402922034263611,
|
| 1737 |
+
"learning_rate": 1.435897435897436e-05,
|
| 1738 |
+
"loss": 0.1102,
|
| 1739 |
+
"step": 6000
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 6.607929515418502,
|
| 1743 |
+
"eval_cer": 26.293375213508654,
|
| 1744 |
+
"eval_loss": 0.7888639569282532,
|
| 1745 |
+
"eval_runtime": 1700.1004,
|
| 1746 |
+
"eval_samples_per_second": 6.224,
|
| 1747 |
+
"eval_steps_per_second": 1.556,
|
| 1748 |
+
"eval_wer": 86.31777463460631,
|
| 1749 |
+
"step": 6000
|
| 1750 |
+
},
|
| 1751 |
+
{
|
| 1752 |
+
"epoch": 6.635462555066079,
|
| 1753 |
+
"grad_norm": 0.8144319653511047,
|
| 1754 |
+
"learning_rate": 1.4333333333333334e-05,
|
| 1755 |
+
"loss": 0.0867,
|
| 1756 |
+
"step": 6025
|
| 1757 |
+
},
|
| 1758 |
+
{
|
| 1759 |
+
"epoch": 6.6629955947136565,
|
| 1760 |
+
"grad_norm": 0.3316783607006073,
|
| 1761 |
+
"learning_rate": 1.430769230769231e-05,
|
| 1762 |
+
"loss": 0.1343,
|
| 1763 |
+
"step": 6050
|
| 1764 |
+
},
|
| 1765 |
+
{
|
| 1766 |
+
"epoch": 6.690528634361233,
|
| 1767 |
+
"grad_norm": 0.6257611513137817,
|
| 1768 |
+
"learning_rate": 1.4282051282051283e-05,
|
| 1769 |
+
"loss": 0.141,
|
| 1770 |
+
"step": 6075
|
| 1771 |
+
},
|
| 1772 |
+
{
|
| 1773 |
+
"epoch": 6.71806167400881,
|
| 1774 |
+
"grad_norm": 0.6132906079292297,
|
| 1775 |
+
"learning_rate": 1.4256410256410258e-05,
|
| 1776 |
+
"loss": 0.1405,
|
| 1777 |
+
"step": 6100
|
| 1778 |
+
},
|
| 1779 |
+
{
|
| 1780 |
+
"epoch": 6.745594713656388,
|
| 1781 |
+
"grad_norm": 0.4126946032047272,
|
| 1782 |
+
"learning_rate": 1.4230769230769232e-05,
|
| 1783 |
+
"loss": 0.125,
|
| 1784 |
+
"step": 6125
|
| 1785 |
+
},
|
| 1786 |
+
{
|
| 1787 |
+
"epoch": 6.773127753303965,
|
| 1788 |
+
"grad_norm": 0.5601705312728882,
|
| 1789 |
+
"learning_rate": 1.4205128205128207e-05,
|
| 1790 |
+
"loss": 0.109,
|
| 1791 |
+
"step": 6150
|
| 1792 |
+
},
|
| 1793 |
+
{
|
| 1794 |
+
"epoch": 6.8006607929515415,
|
| 1795 |
+
"grad_norm": 0.51097571849823,
|
| 1796 |
+
"learning_rate": 1.4179487179487181e-05,
|
| 1797 |
+
"loss": 0.1227,
|
| 1798 |
+
"step": 6175
|
| 1799 |
+
},
|
| 1800 |
+
{
|
| 1801 |
+
"epoch": 6.828193832599119,
|
| 1802 |
+
"grad_norm": 0.557555615901947,
|
| 1803 |
+
"learning_rate": 1.4153846153846156e-05,
|
| 1804 |
+
"loss": 0.1051,
|
| 1805 |
+
"step": 6200
|
| 1806 |
+
},
|
| 1807 |
+
{
|
| 1808 |
+
"epoch": 6.855726872246696,
|
| 1809 |
+
"grad_norm": 0.602931797504425,
|
| 1810 |
+
"learning_rate": 1.412820512820513e-05,
|
| 1811 |
+
"loss": 0.1323,
|
| 1812 |
+
"step": 6225
|
| 1813 |
+
},
|
| 1814 |
+
{
|
| 1815 |
+
"epoch": 6.883259911894273,
|
| 1816 |
+
"grad_norm": 0.4400341212749481,
|
| 1817 |
+
"learning_rate": 1.4102564102564105e-05,
|
| 1818 |
+
"loss": 0.1118,
|
| 1819 |
+
"step": 6250
|
| 1820 |
+
},
|
| 1821 |
+
{
|
| 1822 |
+
"epoch": 6.9107929515418505,
|
| 1823 |
+
"grad_norm": 0.8456242084503174,
|
| 1824 |
+
"learning_rate": 1.4076923076923079e-05,
|
| 1825 |
+
"loss": 0.093,
|
| 1826 |
+
"step": 6275
|
| 1827 |
+
},
|
| 1828 |
+
{
|
| 1829 |
+
"epoch": 6.938325991189427,
|
| 1830 |
+
"grad_norm": 0.34682753682136536,
|
| 1831 |
+
"learning_rate": 1.405128205128205e-05,
|
| 1832 |
+
"loss": 0.1108,
|
| 1833 |
+
"step": 6300
|
| 1834 |
+
},
|
| 1835 |
+
{
|
| 1836 |
+
"epoch": 6.965859030837004,
|
| 1837 |
+
"grad_norm": 0.5863762497901917,
|
| 1838 |
+
"learning_rate": 1.4025641025641026e-05,
|
| 1839 |
+
"loss": 0.1206,
|
| 1840 |
+
"step": 6325
|
| 1841 |
+
},
|
| 1842 |
+
{
|
| 1843 |
+
"epoch": 6.993392070484582,
|
| 1844 |
+
"grad_norm": 0.4746488332748413,
|
| 1845 |
+
"learning_rate": 1.4e-05,
|
| 1846 |
+
"loss": 0.1128,
|
| 1847 |
+
"step": 6350
|
| 1848 |
+
},
|
| 1849 |
+
{
|
| 1850 |
+
"epoch": 7.020925110132159,
|
| 1851 |
+
"grad_norm": 0.19997574388980865,
|
| 1852 |
+
"learning_rate": 1.3974358974358975e-05,
|
| 1853 |
+
"loss": 0.0823,
|
| 1854 |
+
"step": 6375
|
| 1855 |
+
},
|
| 1856 |
+
{
|
| 1857 |
+
"epoch": 7.048458149779735,
|
| 1858 |
+
"grad_norm": 0.44562461972236633,
|
| 1859 |
+
"learning_rate": 1.3948717948717949e-05,
|
| 1860 |
+
"loss": 0.0788,
|
| 1861 |
+
"step": 6400
|
| 1862 |
+
},
|
| 1863 |
+
{
|
| 1864 |
+
"epoch": 7.075991189427313,
|
| 1865 |
+
"grad_norm": 1.034386396408081,
|
| 1866 |
+
"learning_rate": 1.3923076923076924e-05,
|
| 1867 |
+
"loss": 0.0651,
|
| 1868 |
+
"step": 6425
|
| 1869 |
+
},
|
| 1870 |
+
{
|
| 1871 |
+
"epoch": 7.10352422907489,
|
| 1872 |
+
"grad_norm": 0.4024251103401184,
|
| 1873 |
+
"learning_rate": 1.3897435897435898e-05,
|
| 1874 |
+
"loss": 0.0603,
|
| 1875 |
+
"step": 6450
|
| 1876 |
+
},
|
| 1877 |
+
{
|
| 1878 |
+
"epoch": 7.131057268722467,
|
| 1879 |
+
"grad_norm": 0.2884047329425812,
|
| 1880 |
+
"learning_rate": 1.3871794871794873e-05,
|
| 1881 |
+
"loss": 0.0759,
|
| 1882 |
+
"step": 6475
|
| 1883 |
+
},
|
| 1884 |
+
{
|
| 1885 |
+
"epoch": 7.158590308370044,
|
| 1886 |
+
"grad_norm": 0.30277690291404724,
|
| 1887 |
+
"learning_rate": 1.3846153846153847e-05,
|
| 1888 |
+
"loss": 0.0789,
|
| 1889 |
+
"step": 6500
|
| 1890 |
+
},
|
| 1891 |
+
{
|
| 1892 |
+
"epoch": 7.186123348017621,
|
| 1893 |
+
"grad_norm": 0.5528630614280701,
|
| 1894 |
+
"learning_rate": 1.3820512820512822e-05,
|
| 1895 |
+
"loss": 0.0687,
|
| 1896 |
+
"step": 6525
|
| 1897 |
+
},
|
| 1898 |
+
{
|
| 1899 |
+
"epoch": 7.213656387665198,
|
| 1900 |
+
"grad_norm": 0.852532684803009,
|
| 1901 |
+
"learning_rate": 1.3794871794871796e-05,
|
| 1902 |
+
"loss": 0.0733,
|
| 1903 |
+
"step": 6550
|
| 1904 |
+
},
|
| 1905 |
+
{
|
| 1906 |
+
"epoch": 7.241189427312776,
|
| 1907 |
+
"grad_norm": 0.5282613039016724,
|
| 1908 |
+
"learning_rate": 1.3769230769230771e-05,
|
| 1909 |
+
"loss": 0.0793,
|
| 1910 |
+
"step": 6575
|
| 1911 |
+
},
|
| 1912 |
+
{
|
| 1913 |
+
"epoch": 7.2687224669603525,
|
| 1914 |
+
"grad_norm": 0.4679343104362488,
|
| 1915 |
+
"learning_rate": 1.3743589743589745e-05,
|
| 1916 |
+
"loss": 0.0734,
|
| 1917 |
+
"step": 6600
|
| 1918 |
+
},
|
| 1919 |
+
{
|
| 1920 |
+
"epoch": 7.296255506607929,
|
| 1921 |
+
"grad_norm": 0.30845746397972107,
|
| 1922 |
+
"learning_rate": 1.3717948717948718e-05,
|
| 1923 |
+
"loss": 0.0871,
|
| 1924 |
+
"step": 6625
|
| 1925 |
+
},
|
| 1926 |
+
{
|
| 1927 |
+
"epoch": 7.323788546255507,
|
| 1928 |
+
"grad_norm": 0.3263518810272217,
|
| 1929 |
+
"learning_rate": 1.3692307692307694e-05,
|
| 1930 |
+
"loss": 0.0774,
|
| 1931 |
+
"step": 6650
|
| 1932 |
+
},
|
| 1933 |
+
{
|
| 1934 |
+
"epoch": 7.351321585903084,
|
| 1935 |
+
"grad_norm": 0.6020768284797668,
|
| 1936 |
+
"learning_rate": 1.3666666666666667e-05,
|
| 1937 |
+
"loss": 0.0869,
|
| 1938 |
+
"step": 6675
|
| 1939 |
+
},
|
| 1940 |
+
{
|
| 1941 |
+
"epoch": 7.378854625550661,
|
| 1942 |
+
"grad_norm": 0.23268386721611023,
|
| 1943 |
+
"learning_rate": 1.3641025641025643e-05,
|
| 1944 |
+
"loss": 0.0673,
|
| 1945 |
+
"step": 6700
|
| 1946 |
+
},
|
| 1947 |
+
{
|
| 1948 |
+
"epoch": 7.406387665198238,
|
| 1949 |
+
"grad_norm": 0.34103459119796753,
|
| 1950 |
+
"learning_rate": 1.3615384615384616e-05,
|
| 1951 |
+
"loss": 0.0621,
|
| 1952 |
+
"step": 6725
|
| 1953 |
+
},
|
| 1954 |
+
{
|
| 1955 |
+
"epoch": 7.433920704845815,
|
| 1956 |
+
"grad_norm": 0.44131579995155334,
|
| 1957 |
+
"learning_rate": 1.3589743589743592e-05,
|
| 1958 |
+
"loss": 0.0752,
|
| 1959 |
+
"step": 6750
|
| 1960 |
+
},
|
| 1961 |
+
{
|
| 1962 |
+
"epoch": 7.461453744493392,
|
| 1963 |
+
"grad_norm": 0.41322773694992065,
|
| 1964 |
+
"learning_rate": 1.3564102564102565e-05,
|
| 1965 |
+
"loss": 0.0879,
|
| 1966 |
+
"step": 6775
|
| 1967 |
+
},
|
| 1968 |
+
{
|
| 1969 |
+
"epoch": 7.48898678414097,
|
| 1970 |
+
"grad_norm": 0.2960795760154724,
|
| 1971 |
+
"learning_rate": 1.353846153846154e-05,
|
| 1972 |
+
"loss": 0.0784,
|
| 1973 |
+
"step": 6800
|
| 1974 |
+
},
|
| 1975 |
+
{
|
| 1976 |
+
"epoch": 7.516519823788546,
|
| 1977 |
+
"grad_norm": 0.3674839735031128,
|
| 1978 |
+
"learning_rate": 1.3512820512820514e-05,
|
| 1979 |
+
"loss": 0.0745,
|
| 1980 |
+
"step": 6825
|
| 1981 |
+
},
|
| 1982 |
+
{
|
| 1983 |
+
"epoch": 7.544052863436123,
|
| 1984 |
+
"grad_norm": 0.16305261850357056,
|
| 1985 |
+
"learning_rate": 1.348717948717949e-05,
|
| 1986 |
+
"loss": 0.0794,
|
| 1987 |
+
"step": 6850
|
| 1988 |
+
},
|
| 1989 |
+
{
|
| 1990 |
+
"epoch": 7.5715859030837,
|
| 1991 |
+
"grad_norm": 0.4451794922351837,
|
| 1992 |
+
"learning_rate": 1.3461538461538463e-05,
|
| 1993 |
+
"loss": 0.0631,
|
| 1994 |
+
"step": 6875
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 7.599118942731278,
|
| 1998 |
+
"grad_norm": 0.507660448551178,
|
| 1999 |
+
"learning_rate": 1.3435897435897435e-05,
|
| 2000 |
+
"loss": 0.0895,
|
| 2001 |
+
"step": 6900
|
| 2002 |
+
},
|
| 2003 |
+
{
|
| 2004 |
+
"epoch": 7.6266519823788546,
|
| 2005 |
+
"grad_norm": 0.2863459587097168,
|
| 2006 |
+
"learning_rate": 1.341025641025641e-05,
|
| 2007 |
+
"loss": 0.0729,
|
| 2008 |
+
"step": 6925
|
| 2009 |
+
},
|
| 2010 |
+
{
|
| 2011 |
+
"epoch": 7.654185022026431,
|
| 2012 |
+
"grad_norm": 0.5745854377746582,
|
| 2013 |
+
"learning_rate": 1.3384615384615384e-05,
|
| 2014 |
+
"loss": 0.0995,
|
| 2015 |
+
"step": 6950
|
| 2016 |
+
},
|
| 2017 |
+
{
|
| 2018 |
+
"epoch": 7.681718061674009,
|
| 2019 |
+
"grad_norm": 0.42818760871887207,
|
| 2020 |
+
"learning_rate": 1.335897435897436e-05,
|
| 2021 |
+
"loss": 0.0932,
|
| 2022 |
+
"step": 6975
|
| 2023 |
+
},
|
| 2024 |
+
{
|
| 2025 |
+
"epoch": 7.709251101321586,
|
| 2026 |
+
"grad_norm": 0.6487565636634827,
|
| 2027 |
+
"learning_rate": 1.3333333333333333e-05,
|
| 2028 |
+
"loss": 0.0682,
|
| 2029 |
+
"step": 7000
|
| 2030 |
+
},
|
| 2031 |
+
{
|
| 2032 |
+
"epoch": 7.709251101321586,
|
| 2033 |
+
"eval_cer": 27.35776598305687,
|
| 2034 |
+
"eval_loss": 0.7991169095039368,
|
| 2035 |
+
"eval_runtime": 1721.4291,
|
| 2036 |
+
"eval_samples_per_second": 6.147,
|
| 2037 |
+
"eval_steps_per_second": 1.537,
|
| 2038 |
+
"eval_wer": 84.45073078736445,
|
| 2039 |
+
"step": 7000
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 7.736784140969163,
|
| 2043 |
+
"grad_norm": 0.7766274809837341,
|
| 2044 |
+
"learning_rate": 1.3307692307692309e-05,
|
| 2045 |
+
"loss": 0.0903,
|
| 2046 |
+
"step": 7025
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"epoch": 7.76431718061674,
|
| 2050 |
+
"grad_norm": 0.41921526193618774,
|
| 2051 |
+
"learning_rate": 1.3282051282051282e-05,
|
| 2052 |
+
"loss": 0.087,
|
| 2053 |
+
"step": 7050
|
| 2054 |
+
},
|
| 2055 |
+
{
|
| 2056 |
+
"epoch": 7.791850220264317,
|
| 2057 |
+
"grad_norm": 0.5744315385818481,
|
| 2058 |
+
"learning_rate": 1.3256410256410258e-05,
|
| 2059 |
+
"loss": 0.0823,
|
| 2060 |
+
"step": 7075
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 7.819383259911894,
|
| 2064 |
+
"grad_norm": 0.7453629970550537,
|
| 2065 |
+
"learning_rate": 1.3230769230769231e-05,
|
| 2066 |
+
"loss": 0.0752,
|
| 2067 |
+
"step": 7100
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 7.846916299559472,
|
| 2071 |
+
"grad_norm": 0.4299808442592621,
|
| 2072 |
+
"learning_rate": 1.3205128205128207e-05,
|
| 2073 |
+
"loss": 0.0725,
|
| 2074 |
+
"step": 7125
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 7.8744493392070485,
|
| 2078 |
+
"grad_norm": 0.6843218207359314,
|
| 2079 |
+
"learning_rate": 1.317948717948718e-05,
|
| 2080 |
+
"loss": 0.0956,
|
| 2081 |
+
"step": 7150
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 7.901982378854625,
|
| 2085 |
+
"grad_norm": 0.7576065063476562,
|
| 2086 |
+
"learning_rate": 1.3153846153846156e-05,
|
| 2087 |
+
"loss": 0.0878,
|
| 2088 |
+
"step": 7175
|
| 2089 |
+
},
|
| 2090 |
+
{
|
| 2091 |
+
"epoch": 7.929515418502203,
|
| 2092 |
+
"grad_norm": 0.5355419516563416,
|
| 2093 |
+
"learning_rate": 1.312820512820513e-05,
|
| 2094 |
+
"loss": 0.1018,
|
| 2095 |
+
"step": 7200
|
| 2096 |
+
},
|
| 2097 |
+
{
|
| 2098 |
+
"epoch": 7.95704845814978,
|
| 2099 |
+
"grad_norm": 0.5288059711456299,
|
| 2100 |
+
"learning_rate": 1.3102564102564103e-05,
|
| 2101 |
+
"loss": 0.0858,
|
| 2102 |
+
"step": 7225
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 7.984581497797357,
|
| 2106 |
+
"grad_norm": 0.4639795124530792,
|
| 2107 |
+
"learning_rate": 1.3076923076923078e-05,
|
| 2108 |
+
"loss": 0.0867,
|
| 2109 |
+
"step": 7250
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 8.012114537444933,
|
| 2113 |
+
"grad_norm": 0.4118720591068268,
|
| 2114 |
+
"learning_rate": 1.3051282051282052e-05,
|
| 2115 |
+
"loss": 0.0966,
|
| 2116 |
+
"step": 7275
|
| 2117 |
+
},
|
| 2118 |
+
{
|
| 2119 |
+
"epoch": 8.039647577092511,
|
| 2120 |
+
"grad_norm": 0.2552703320980072,
|
| 2121 |
+
"learning_rate": 1.3025641025641027e-05,
|
| 2122 |
+
"loss": 0.0644,
|
| 2123 |
+
"step": 7300
|
| 2124 |
+
},
|
| 2125 |
+
{
|
| 2126 |
+
"epoch": 8.067180616740089,
|
| 2127 |
+
"grad_norm": 0.4209102690219879,
|
| 2128 |
+
"learning_rate": 1.3000000000000001e-05,
|
| 2129 |
+
"loss": 0.0534,
|
| 2130 |
+
"step": 7325
|
| 2131 |
+
},
|
| 2132 |
+
{
|
| 2133 |
+
"epoch": 8.094713656387665,
|
| 2134 |
+
"grad_norm": 0.4842354655265808,
|
| 2135 |
+
"learning_rate": 1.2974358974358976e-05,
|
| 2136 |
+
"loss": 0.0567,
|
| 2137 |
+
"step": 7350
|
| 2138 |
+
},
|
| 2139 |
+
{
|
| 2140 |
+
"epoch": 8.122246696035242,
|
| 2141 |
+
"grad_norm": 0.5336529612541199,
|
| 2142 |
+
"learning_rate": 1.294871794871795e-05,
|
| 2143 |
+
"loss": 0.0581,
|
| 2144 |
+
"step": 7375
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 8.14977973568282,
|
| 2148 |
+
"grad_norm": 0.3102397918701172,
|
| 2149 |
+
"learning_rate": 1.2923076923076925e-05,
|
| 2150 |
+
"loss": 0.0502,
|
| 2151 |
+
"step": 7400
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 8.177312775330396,
|
| 2155 |
+
"grad_norm": 0.29148268699645996,
|
| 2156 |
+
"learning_rate": 1.2897435897435899e-05,
|
| 2157 |
+
"loss": 0.0494,
|
| 2158 |
+
"step": 7425
|
| 2159 |
+
},
|
| 2160 |
+
{
|
| 2161 |
+
"epoch": 8.204845814977974,
|
| 2162 |
+
"grad_norm": 0.11677376180887222,
|
| 2163 |
+
"learning_rate": 1.2871794871794874e-05,
|
| 2164 |
+
"loss": 0.0428,
|
| 2165 |
+
"step": 7450
|
| 2166 |
+
},
|
| 2167 |
+
{
|
| 2168 |
+
"epoch": 8.232378854625551,
|
| 2169 |
+
"grad_norm": 0.5054081082344055,
|
| 2170 |
+
"learning_rate": 1.2846153846153848e-05,
|
| 2171 |
+
"loss": 0.0613,
|
| 2172 |
+
"step": 7475
|
| 2173 |
+
},
|
| 2174 |
+
{
|
| 2175 |
+
"epoch": 8.259911894273127,
|
| 2176 |
+
"grad_norm": 0.16621895134449005,
|
| 2177 |
+
"learning_rate": 1.2820512820512823e-05,
|
| 2178 |
+
"loss": 0.0465,
|
| 2179 |
+
"step": 7500
|
| 2180 |
+
},
|
| 2181 |
+
{
|
| 2182 |
+
"epoch": 8.287444933920705,
|
| 2183 |
+
"grad_norm": 0.6071426272392273,
|
| 2184 |
+
"learning_rate": 1.2794871794871795e-05,
|
| 2185 |
+
"loss": 0.0466,
|
| 2186 |
+
"step": 7525
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 8.314977973568283,
|
| 2190 |
+
"grad_norm": 0.4491996765136719,
|
| 2191 |
+
"learning_rate": 1.2769230769230769e-05,
|
| 2192 |
+
"loss": 0.0504,
|
| 2193 |
+
"step": 7550
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 8.342511013215859,
|
| 2197 |
+
"grad_norm": 0.5606942176818848,
|
| 2198 |
+
"learning_rate": 1.2743589743589744e-05,
|
| 2199 |
+
"loss": 0.0493,
|
| 2200 |
+
"step": 7575
|
| 2201 |
+
},
|
| 2202 |
+
{
|
| 2203 |
+
"epoch": 8.370044052863436,
|
| 2204 |
+
"grad_norm": 0.6338093280792236,
|
| 2205 |
+
"learning_rate": 1.2717948717948718e-05,
|
| 2206 |
+
"loss": 0.0437,
|
| 2207 |
+
"step": 7600
|
| 2208 |
+
},
|
| 2209 |
+
{
|
| 2210 |
+
"epoch": 8.397577092511014,
|
| 2211 |
+
"grad_norm": 0.50752192735672,
|
| 2212 |
+
"learning_rate": 1.2692307692307693e-05,
|
| 2213 |
+
"loss": 0.0558,
|
| 2214 |
+
"step": 7625
|
| 2215 |
+
},
|
| 2216 |
+
{
|
| 2217 |
+
"epoch": 8.42511013215859,
|
| 2218 |
+
"grad_norm": 0.16351301968097687,
|
| 2219 |
+
"learning_rate": 1.2666666666666667e-05,
|
| 2220 |
+
"loss": 0.0479,
|
| 2221 |
+
"step": 7650
|
| 2222 |
+
},
|
| 2223 |
+
{
|
| 2224 |
+
"epoch": 8.452643171806168,
|
| 2225 |
+
"grad_norm": 0.392193078994751,
|
| 2226 |
+
"learning_rate": 1.2641025641025642e-05,
|
| 2227 |
+
"loss": 0.0542,
|
| 2228 |
+
"step": 7675
|
| 2229 |
+
},
|
| 2230 |
+
{
|
| 2231 |
+
"epoch": 8.480176211453745,
|
| 2232 |
+
"grad_norm": 0.23874300718307495,
|
| 2233 |
+
"learning_rate": 1.2615384615384616e-05,
|
| 2234 |
+
"loss": 0.0516,
|
| 2235 |
+
"step": 7700
|
| 2236 |
+
},
|
| 2237 |
+
{
|
| 2238 |
+
"epoch": 8.507709251101321,
|
| 2239 |
+
"grad_norm": 0.31626826524734497,
|
| 2240 |
+
"learning_rate": 1.2589743589743591e-05,
|
| 2241 |
+
"loss": 0.0486,
|
| 2242 |
+
"step": 7725
|
| 2243 |
+
},
|
| 2244 |
+
{
|
| 2245 |
+
"epoch": 8.535242290748899,
|
| 2246 |
+
"grad_norm": 0.5571256875991821,
|
| 2247 |
+
"learning_rate": 1.2564102564102565e-05,
|
| 2248 |
+
"loss": 0.0665,
|
| 2249 |
+
"step": 7750
|
| 2250 |
+
},
|
| 2251 |
+
{
|
| 2252 |
+
"epoch": 8.562775330396477,
|
| 2253 |
+
"grad_norm": 0.3961975872516632,
|
| 2254 |
+
"learning_rate": 1.253846153846154e-05,
|
| 2255 |
+
"loss": 0.0587,
|
| 2256 |
+
"step": 7775
|
| 2257 |
+
},
|
| 2258 |
+
{
|
| 2259 |
+
"epoch": 8.590308370044053,
|
| 2260 |
+
"grad_norm": 0.33032137155532837,
|
| 2261 |
+
"learning_rate": 1.2512820512820514e-05,
|
| 2262 |
+
"loss": 0.0631,
|
| 2263 |
+
"step": 7800
|
| 2264 |
+
},
|
| 2265 |
+
{
|
| 2266 |
+
"epoch": 8.61784140969163,
|
| 2267 |
+
"grad_norm": 0.1791045367717743,
|
| 2268 |
+
"learning_rate": 1.2487179487179487e-05,
|
| 2269 |
+
"loss": 0.0669,
|
| 2270 |
+
"step": 7825
|
| 2271 |
+
},
|
| 2272 |
+
{
|
| 2273 |
+
"epoch": 8.645374449339208,
|
| 2274 |
+
"grad_norm": 0.3944324851036072,
|
| 2275 |
+
"learning_rate": 1.2461538461538463e-05,
|
| 2276 |
+
"loss": 0.0439,
|
| 2277 |
+
"step": 7850
|
| 2278 |
+
},
|
| 2279 |
+
{
|
| 2280 |
+
"epoch": 8.672907488986784,
|
| 2281 |
+
"grad_norm": 0.13560180366039276,
|
| 2282 |
+
"learning_rate": 1.2435897435897436e-05,
|
| 2283 |
+
"loss": 0.0403,
|
| 2284 |
+
"step": 7875
|
| 2285 |
+
},
|
| 2286 |
+
{
|
| 2287 |
+
"epoch": 8.700440528634362,
|
| 2288 |
+
"grad_norm": 0.4744511842727661,
|
| 2289 |
+
"learning_rate": 1.2410256410256412e-05,
|
| 2290 |
+
"loss": 0.0595,
|
| 2291 |
+
"step": 7900
|
| 2292 |
+
},
|
| 2293 |
+
{
|
| 2294 |
+
"epoch": 8.72797356828194,
|
| 2295 |
+
"grad_norm": 0.5268675684928894,
|
| 2296 |
+
"learning_rate": 1.2384615384615385e-05,
|
| 2297 |
+
"loss": 0.0568,
|
| 2298 |
+
"step": 7925
|
| 2299 |
+
},
|
| 2300 |
+
{
|
| 2301 |
+
"epoch": 8.755506607929515,
|
| 2302 |
+
"grad_norm": 0.33561986684799194,
|
| 2303 |
+
"learning_rate": 1.235897435897436e-05,
|
| 2304 |
+
"loss": 0.057,
|
| 2305 |
+
"step": 7950
|
| 2306 |
+
},
|
| 2307 |
+
{
|
| 2308 |
+
"epoch": 8.783039647577093,
|
| 2309 |
+
"grad_norm": 0.23346789181232452,
|
| 2310 |
+
"learning_rate": 1.2333333333333334e-05,
|
| 2311 |
+
"loss": 0.05,
|
| 2312 |
+
"step": 7975
|
| 2313 |
+
},
|
| 2314 |
+
{
|
| 2315 |
+
"epoch": 8.810572687224669,
|
| 2316 |
+
"grad_norm": 0.5446615815162659,
|
| 2317 |
+
"learning_rate": 1.230769230769231e-05,
|
| 2318 |
+
"loss": 0.0647,
|
| 2319 |
+
"step": 8000
|
| 2320 |
+
},
|
| 2321 |
+
{
|
| 2322 |
+
"epoch": 8.810572687224669,
|
| 2323 |
+
"eval_cer": 25.62619650820217,
|
| 2324 |
+
"eval_loss": 0.8131716847419739,
|
| 2325 |
+
"eval_runtime": 1709.9866,
|
| 2326 |
+
"eval_samples_per_second": 6.188,
|
| 2327 |
+
"eval_steps_per_second": 1.547,
|
| 2328 |
+
"eval_wer": 84.64875058934464,
|
| 2329 |
+
"step": 8000
|
| 2330 |
+
},
|
| 2331 |
+
{
|
| 2332 |
+
"epoch": 8.838105726872246,
|
| 2333 |
+
"grad_norm": 0.37270498275756836,
|
| 2334 |
+
"learning_rate": 1.2282051282051283e-05,
|
| 2335 |
+
"loss": 0.0486,
|
| 2336 |
+
"step": 8025
|
| 2337 |
+
},
|
| 2338 |
+
{
|
| 2339 |
+
"epoch": 8.865638766519824,
|
| 2340 |
+
"grad_norm": 0.4193556308746338,
|
| 2341 |
+
"learning_rate": 1.2256410256410259e-05,
|
| 2342 |
+
"loss": 0.0544,
|
| 2343 |
+
"step": 8050
|
| 2344 |
+
},
|
| 2345 |
+
{
|
| 2346 |
+
"epoch": 8.8931718061674,
|
| 2347 |
+
"grad_norm": 0.4220805764198303,
|
| 2348 |
+
"learning_rate": 1.2230769230769232e-05,
|
| 2349 |
+
"loss": 0.0555,
|
| 2350 |
+
"step": 8075
|
| 2351 |
+
},
|
| 2352 |
+
{
|
| 2353 |
+
"epoch": 8.920704845814978,
|
| 2354 |
+
"grad_norm": 0.12924405932426453,
|
| 2355 |
+
"learning_rate": 1.2205128205128208e-05,
|
| 2356 |
+
"loss": 0.0654,
|
| 2357 |
+
"step": 8100
|
| 2358 |
+
},
|
| 2359 |
+
{
|
| 2360 |
+
"epoch": 8.948237885462555,
|
| 2361 |
+
"grad_norm": 0.2983054518699646,
|
| 2362 |
+
"learning_rate": 1.217948717948718e-05,
|
| 2363 |
+
"loss": 0.0481,
|
| 2364 |
+
"step": 8125
|
| 2365 |
+
},
|
| 2366 |
+
{
|
| 2367 |
+
"epoch": 8.975770925110131,
|
| 2368 |
+
"grad_norm": 0.5196457505226135,
|
| 2369 |
+
"learning_rate": 1.2153846153846153e-05,
|
| 2370 |
+
"loss": 0.0455,
|
| 2371 |
+
"step": 8150
|
| 2372 |
+
},
|
| 2373 |
+
{
|
| 2374 |
+
"epoch": 9.003303964757709,
|
| 2375 |
+
"grad_norm": 0.1929197460412979,
|
| 2376 |
+
"learning_rate": 1.2128205128205129e-05,
|
| 2377 |
+
"loss": 0.0587,
|
| 2378 |
+
"step": 8175
|
| 2379 |
+
},
|
| 2380 |
+
{
|
| 2381 |
+
"epoch": 9.030837004405287,
|
| 2382 |
+
"grad_norm": 0.6114550828933716,
|
| 2383 |
+
"learning_rate": 1.2102564102564102e-05,
|
| 2384 |
+
"loss": 0.0375,
|
| 2385 |
+
"step": 8200
|
| 2386 |
+
},
|
| 2387 |
+
{
|
| 2388 |
+
"epoch": 9.058370044052863,
|
| 2389 |
+
"grad_norm": 0.3071931004524231,
|
| 2390 |
+
"learning_rate": 1.2076923076923078e-05,
|
| 2391 |
+
"loss": 0.032,
|
| 2392 |
+
"step": 8225
|
| 2393 |
+
},
|
| 2394 |
+
{
|
| 2395 |
+
"epoch": 9.08590308370044,
|
| 2396 |
+
"grad_norm": 0.3815561532974243,
|
| 2397 |
+
"learning_rate": 1.2051282051282051e-05,
|
| 2398 |
+
"loss": 0.0297,
|
| 2399 |
+
"step": 8250
|
| 2400 |
+
},
|
| 2401 |
+
{
|
| 2402 |
+
"epoch": 9.113436123348018,
|
| 2403 |
+
"grad_norm": 0.3786160349845886,
|
| 2404 |
+
"learning_rate": 1.2025641025641027e-05,
|
| 2405 |
+
"loss": 0.0357,
|
| 2406 |
+
"step": 8275
|
| 2407 |
+
},
|
| 2408 |
+
{
|
| 2409 |
+
"epoch": 9.140969162995594,
|
| 2410 |
+
"grad_norm": 0.32819467782974243,
|
| 2411 |
+
"learning_rate": 1.2e-05,
|
| 2412 |
+
"loss": 0.0362,
|
| 2413 |
+
"step": 8300
|
| 2414 |
+
},
|
| 2415 |
+
{
|
| 2416 |
+
"epoch": 9.168502202643172,
|
| 2417 |
+
"grad_norm": 0.2228395938873291,
|
| 2418 |
+
"learning_rate": 1.1974358974358976e-05,
|
| 2419 |
+
"loss": 0.0259,
|
| 2420 |
+
"step": 8325
|
| 2421 |
+
},
|
| 2422 |
+
{
|
| 2423 |
+
"epoch": 9.19603524229075,
|
| 2424 |
+
"grad_norm": 0.3630577325820923,
|
| 2425 |
+
"learning_rate": 1.194871794871795e-05,
|
| 2426 |
+
"loss": 0.0388,
|
| 2427 |
+
"step": 8350
|
| 2428 |
+
},
|
| 2429 |
+
{
|
| 2430 |
+
"epoch": 9.223568281938325,
|
| 2431 |
+
"grad_norm": 0.24628213047981262,
|
| 2432 |
+
"learning_rate": 1.1923076923076925e-05,
|
| 2433 |
+
"loss": 0.0344,
|
| 2434 |
+
"step": 8375
|
| 2435 |
+
},
|
| 2436 |
+
{
|
| 2437 |
+
"epoch": 9.251101321585903,
|
| 2438 |
+
"grad_norm": 0.4021029472351074,
|
| 2439 |
+
"learning_rate": 1.1897435897435898e-05,
|
| 2440 |
+
"loss": 0.0322,
|
| 2441 |
+
"step": 8400
|
| 2442 |
+
},
|
| 2443 |
+
{
|
| 2444 |
+
"epoch": 9.27863436123348,
|
| 2445 |
+
"grad_norm": 0.1852307766675949,
|
| 2446 |
+
"learning_rate": 1.1871794871794872e-05,
|
| 2447 |
+
"loss": 0.0327,
|
| 2448 |
+
"step": 8425
|
| 2449 |
+
},
|
| 2450 |
+
{
|
| 2451 |
+
"epoch": 9.306167400881057,
|
| 2452 |
+
"grad_norm": 0.8122808337211609,
|
| 2453 |
+
"learning_rate": 1.1846153846153847e-05,
|
| 2454 |
+
"loss": 0.0373,
|
| 2455 |
+
"step": 8450
|
| 2456 |
+
},
|
| 2457 |
+
{
|
| 2458 |
+
"epoch": 9.333700440528634,
|
| 2459 |
+
"grad_norm": 0.4952065348625183,
|
| 2460 |
+
"learning_rate": 1.1820512820512821e-05,
|
| 2461 |
+
"loss": 0.0344,
|
| 2462 |
+
"step": 8475
|
| 2463 |
+
},
|
| 2464 |
+
{
|
| 2465 |
+
"epoch": 9.361233480176212,
|
| 2466 |
+
"grad_norm": 0.2687460482120514,
|
| 2467 |
+
"learning_rate": 1.1794871794871796e-05,
|
| 2468 |
+
"loss": 0.0449,
|
| 2469 |
+
"step": 8500
|
| 2470 |
+
},
|
| 2471 |
+
{
|
| 2472 |
+
"epoch": 9.388766519823788,
|
| 2473 |
+
"grad_norm": 0.46548476815223694,
|
| 2474 |
+
"learning_rate": 1.176923076923077e-05,
|
| 2475 |
+
"loss": 0.0347,
|
| 2476 |
+
"step": 8525
|
| 2477 |
+
},
|
| 2478 |
+
{
|
| 2479 |
+
"epoch": 9.416299559471366,
|
| 2480 |
+
"grad_norm": 0.271555632352829,
|
| 2481 |
+
"learning_rate": 1.1743589743589745e-05,
|
| 2482 |
+
"loss": 0.0372,
|
| 2483 |
+
"step": 8550
|
| 2484 |
+
},
|
| 2485 |
+
{
|
| 2486 |
+
"epoch": 9.443832599118943,
|
| 2487 |
+
"grad_norm": 0.2767009735107422,
|
| 2488 |
+
"learning_rate": 1.1717948717948719e-05,
|
| 2489 |
+
"loss": 0.0341,
|
| 2490 |
+
"step": 8575
|
| 2491 |
+
},
|
| 2492 |
+
{
|
| 2493 |
+
"epoch": 9.47136563876652,
|
| 2494 |
+
"grad_norm": 0.5092193484306335,
|
| 2495 |
+
"learning_rate": 1.1692307692307694e-05,
|
| 2496 |
+
"loss": 0.0344,
|
| 2497 |
+
"step": 8600
|
| 2498 |
+
},
|
| 2499 |
+
{
|
| 2500 |
+
"epoch": 9.498898678414097,
|
| 2501 |
+
"grad_norm": 0.4733143448829651,
|
| 2502 |
+
"learning_rate": 1.1666666666666668e-05,
|
| 2503 |
+
"loss": 0.0442,
|
| 2504 |
+
"step": 8625
|
| 2505 |
+
},
|
| 2506 |
+
{
|
| 2507 |
+
"epoch": 9.526431718061675,
|
| 2508 |
+
"grad_norm": 0.3013637661933899,
|
| 2509 |
+
"learning_rate": 1.1641025641025643e-05,
|
| 2510 |
+
"loss": 0.0437,
|
| 2511 |
+
"step": 8650
|
| 2512 |
+
},
|
| 2513 |
+
{
|
| 2514 |
+
"epoch": 9.55396475770925,
|
| 2515 |
+
"grad_norm": 0.23593106865882874,
|
| 2516 |
+
"learning_rate": 1.1615384615384617e-05,
|
| 2517 |
+
"loss": 0.0345,
|
| 2518 |
+
"step": 8675
|
| 2519 |
+
},
|
| 2520 |
+
{
|
| 2521 |
+
"epoch": 9.581497797356828,
|
| 2522 |
+
"grad_norm": 0.42476484179496765,
|
| 2523 |
+
"learning_rate": 1.1589743589743592e-05,
|
| 2524 |
+
"loss": 0.033,
|
| 2525 |
+
"step": 8700
|
| 2526 |
+
},
|
| 2527 |
+
{
|
| 2528 |
+
"epoch": 9.609030837004406,
|
| 2529 |
+
"grad_norm": 0.1042499765753746,
|
| 2530 |
+
"learning_rate": 1.1564102564102566e-05,
|
| 2531 |
+
"loss": 0.0452,
|
| 2532 |
+
"step": 8725
|
| 2533 |
+
},
|
| 2534 |
+
{
|
| 2535 |
+
"epoch": 9.636563876651982,
|
| 2536 |
+
"grad_norm": 0.4548315405845642,
|
| 2537 |
+
"learning_rate": 1.1538461538461538e-05,
|
| 2538 |
+
"loss": 0.049,
|
| 2539 |
+
"step": 8750
|
| 2540 |
+
},
|
| 2541 |
+
{
|
| 2542 |
+
"epoch": 9.66409691629956,
|
| 2543 |
+
"grad_norm": 0.749373733997345,
|
| 2544 |
+
"learning_rate": 1.1512820512820513e-05,
|
| 2545 |
+
"loss": 0.0397,
|
| 2546 |
+
"step": 8775
|
| 2547 |
+
},
|
| 2548 |
+
{
|
| 2549 |
+
"epoch": 9.691629955947137,
|
| 2550 |
+
"grad_norm": 0.24911393225193024,
|
| 2551 |
+
"learning_rate": 1.1487179487179487e-05,
|
| 2552 |
+
"loss": 0.0327,
|
| 2553 |
+
"step": 8800
|
| 2554 |
+
},
|
| 2555 |
+
{
|
| 2556 |
+
"epoch": 9.719162995594713,
|
| 2557 |
+
"grad_norm": 0.3840325176715851,
|
| 2558 |
+
"learning_rate": 1.1461538461538462e-05,
|
| 2559 |
+
"loss": 0.0378,
|
| 2560 |
+
"step": 8825
|
| 2561 |
+
},
|
| 2562 |
+
{
|
| 2563 |
+
"epoch": 9.746696035242291,
|
| 2564 |
+
"grad_norm": 0.3074113726615906,
|
| 2565 |
+
"learning_rate": 1.1435897435897436e-05,
|
| 2566 |
+
"loss": 0.0424,
|
| 2567 |
+
"step": 8850
|
| 2568 |
+
},
|
| 2569 |
+
{
|
| 2570 |
+
"epoch": 9.774229074889869,
|
| 2571 |
+
"grad_norm": 0.3504096269607544,
|
| 2572 |
+
"learning_rate": 1.1410256410256411e-05,
|
| 2573 |
+
"loss": 0.0408,
|
| 2574 |
+
"step": 8875
|
| 2575 |
+
},
|
| 2576 |
+
{
|
| 2577 |
+
"epoch": 9.801762114537445,
|
| 2578 |
+
"grad_norm": 0.7079716324806213,
|
| 2579 |
+
"learning_rate": 1.1384615384615385e-05,
|
| 2580 |
+
"loss": 0.0362,
|
| 2581 |
+
"step": 8900
|
| 2582 |
+
},
|
| 2583 |
+
{
|
| 2584 |
+
"epoch": 9.829295154185022,
|
| 2585 |
+
"grad_norm": 0.46274736523628235,
|
| 2586 |
+
"learning_rate": 1.135897435897436e-05,
|
| 2587 |
+
"loss": 0.0444,
|
| 2588 |
+
"step": 8925
|
| 2589 |
+
},
|
| 2590 |
+
{
|
| 2591 |
+
"epoch": 9.8568281938326,
|
| 2592 |
+
"grad_norm": 0.30366581678390503,
|
| 2593 |
+
"learning_rate": 1.1333333333333334e-05,
|
| 2594 |
+
"loss": 0.0411,
|
| 2595 |
+
"step": 8950
|
| 2596 |
+
},
|
| 2597 |
+
{
|
| 2598 |
+
"epoch": 9.884361233480176,
|
| 2599 |
+
"grad_norm": 0.394626647233963,
|
| 2600 |
+
"learning_rate": 1.1307692307692309e-05,
|
| 2601 |
+
"loss": 0.0347,
|
| 2602 |
+
"step": 8975
|
| 2603 |
+
},
|
| 2604 |
+
{
|
| 2605 |
+
"epoch": 9.911894273127754,
|
| 2606 |
+
"grad_norm": 0.2617024779319763,
|
| 2607 |
+
"learning_rate": 1.1282051282051283e-05,
|
| 2608 |
+
"loss": 0.0343,
|
| 2609 |
+
"step": 9000
|
| 2610 |
+
},
|
| 2611 |
+
{
|
| 2612 |
+
"epoch": 9.911894273127754,
|
| 2613 |
+
"eval_cer": 24.69478293764974,
|
| 2614 |
+
"eval_loss": 0.8281993865966797,
|
| 2615 |
+
"eval_runtime": 1707.5191,
|
| 2616 |
+
"eval_samples_per_second": 6.197,
|
| 2617 |
+
"eval_steps_per_second": 1.55,
|
| 2618 |
+
"eval_wer": 84.82791136256482,
|
| 2619 |
+
"step": 9000
|
| 2620 |
+
},
|
| 2621 |
+
{
|
| 2622 |
+
"epoch": 9.939427312775331,
|
| 2623 |
+
"grad_norm": 0.5920683145523071,
|
| 2624 |
+
"learning_rate": 1.1256410256410258e-05,
|
| 2625 |
+
"loss": 0.0415,
|
| 2626 |
+
"step": 9025
|
| 2627 |
+
},
|
| 2628 |
+
{
|
| 2629 |
+
"epoch": 9.966960352422907,
|
| 2630 |
+
"grad_norm": 0.5178209543228149,
|
| 2631 |
+
"learning_rate": 1.1230769230769232e-05,
|
| 2632 |
+
"loss": 0.0385,
|
| 2633 |
+
"step": 9050
|
| 2634 |
+
},
|
| 2635 |
+
{
|
| 2636 |
+
"epoch": 9.994493392070485,
|
| 2637 |
+
"grad_norm": 0.33022773265838623,
|
| 2638 |
+
"learning_rate": 1.1205128205128205e-05,
|
| 2639 |
+
"loss": 0.038,
|
| 2640 |
+
"step": 9075
|
| 2641 |
+
},
|
| 2642 |
+
{
|
| 2643 |
+
"epoch": 10.022026431718063,
|
| 2644 |
+
"grad_norm": 0.5157202482223511,
|
| 2645 |
+
"learning_rate": 1.117948717948718e-05,
|
| 2646 |
+
"loss": 0.0304,
|
| 2647 |
+
"step": 9100
|
| 2648 |
+
},
|
| 2649 |
+
{
|
| 2650 |
+
"epoch": 10.049559471365638,
|
| 2651 |
+
"grad_norm": 0.28111565113067627,
|
| 2652 |
+
"learning_rate": 1.1153846153846154e-05,
|
| 2653 |
+
"loss": 0.0271,
|
| 2654 |
+
"step": 9125
|
| 2655 |
+
},
|
| 2656 |
+
{
|
| 2657 |
+
"epoch": 10.077092511013216,
|
| 2658 |
+
"grad_norm": 0.3325743079185486,
|
| 2659 |
+
"learning_rate": 1.112820512820513e-05,
|
| 2660 |
+
"loss": 0.0348,
|
| 2661 |
+
"step": 9150
|
| 2662 |
+
},
|
| 2663 |
+
{
|
| 2664 |
+
"epoch": 10.104625550660794,
|
| 2665 |
+
"grad_norm": 0.39298519492149353,
|
| 2666 |
+
"learning_rate": 1.1102564102564103e-05,
|
| 2667 |
+
"loss": 0.0221,
|
| 2668 |
+
"step": 9175
|
| 2669 |
+
},
|
| 2670 |
+
{
|
| 2671 |
+
"epoch": 10.13215859030837,
|
| 2672 |
+
"grad_norm": 0.7207234501838684,
|
| 2673 |
+
"learning_rate": 1.1076923076923079e-05,
|
| 2674 |
+
"loss": 0.0316,
|
| 2675 |
+
"step": 9200
|
| 2676 |
+
},
|
| 2677 |
+
{
|
| 2678 |
+
"epoch": 10.159691629955947,
|
| 2679 |
+
"grad_norm": 0.12185712903738022,
|
| 2680 |
+
"learning_rate": 1.1051282051282052e-05,
|
| 2681 |
+
"loss": 0.0333,
|
| 2682 |
+
"step": 9225
|
| 2683 |
+
},
|
| 2684 |
+
{
|
| 2685 |
+
"epoch": 10.187224669603523,
|
| 2686 |
+
"grad_norm": 0.18927183747291565,
|
| 2687 |
+
"learning_rate": 1.1025641025641028e-05,
|
| 2688 |
+
"loss": 0.0317,
|
| 2689 |
+
"step": 9250
|
| 2690 |
+
},
|
| 2691 |
+
{
|
| 2692 |
+
"epoch": 10.214757709251101,
|
| 2693 |
+
"grad_norm": 0.5243281722068787,
|
| 2694 |
+
"learning_rate": 1.1000000000000001e-05,
|
| 2695 |
+
"loss": 0.0238,
|
| 2696 |
+
"step": 9275
|
| 2697 |
+
},
|
| 2698 |
+
{
|
| 2699 |
+
"epoch": 10.242290748898679,
|
| 2700 |
+
"grad_norm": 0.25754281878471375,
|
| 2701 |
+
"learning_rate": 1.0974358974358977e-05,
|
| 2702 |
+
"loss": 0.0265,
|
| 2703 |
+
"step": 9300
|
| 2704 |
+
},
|
| 2705 |
+
{
|
| 2706 |
+
"epoch": 10.269823788546255,
|
| 2707 |
+
"grad_norm": 0.3999498188495636,
|
| 2708 |
+
"learning_rate": 1.094871794871795e-05,
|
| 2709 |
+
"loss": 0.0271,
|
| 2710 |
+
"step": 9325
|
| 2711 |
+
},
|
| 2712 |
+
{
|
| 2713 |
+
"epoch": 10.297356828193832,
|
| 2714 |
+
"grad_norm": 0.5361197590827942,
|
| 2715 |
+
"learning_rate": 1.0923076923076922e-05,
|
| 2716 |
+
"loss": 0.0299,
|
| 2717 |
+
"step": 9350
|
| 2718 |
+
},
|
| 2719 |
+
{
|
| 2720 |
+
"epoch": 10.32488986784141,
|
| 2721 |
+
"grad_norm": 0.4799012839794159,
|
| 2722 |
+
"learning_rate": 1.0897435897435898e-05,
|
| 2723 |
+
"loss": 0.0337,
|
| 2724 |
+
"step": 9375
|
| 2725 |
+
},
|
| 2726 |
+
{
|
| 2727 |
+
"epoch": 10.352422907488986,
|
| 2728 |
+
"grad_norm": 0.29171785712242126,
|
| 2729 |
+
"learning_rate": 1.0871794871794871e-05,
|
| 2730 |
+
"loss": 0.0194,
|
| 2731 |
+
"step": 9400
|
| 2732 |
+
},
|
| 2733 |
+
{
|
| 2734 |
+
"epoch": 10.379955947136564,
|
| 2735 |
+
"grad_norm": 0.158452108502388,
|
| 2736 |
+
"learning_rate": 1.0846153846153847e-05,
|
| 2737 |
+
"loss": 0.0293,
|
| 2738 |
+
"step": 9425
|
| 2739 |
+
},
|
| 2740 |
+
{
|
| 2741 |
+
"epoch": 10.407488986784141,
|
| 2742 |
+
"grad_norm": 0.1673414409160614,
|
| 2743 |
+
"learning_rate": 1.082051282051282e-05,
|
| 2744 |
+
"loss": 0.0245,
|
| 2745 |
+
"step": 9450
|
| 2746 |
+
},
|
| 2747 |
+
{
|
| 2748 |
+
"epoch": 10.435022026431717,
|
| 2749 |
+
"grad_norm": 0.4289039075374603,
|
| 2750 |
+
"learning_rate": 1.0794871794871796e-05,
|
| 2751 |
+
"loss": 0.0251,
|
| 2752 |
+
"step": 9475
|
| 2753 |
+
},
|
| 2754 |
+
{
|
| 2755 |
+
"epoch": 10.462555066079295,
|
| 2756 |
+
"grad_norm": 0.43541353940963745,
|
| 2757 |
+
"learning_rate": 1.076923076923077e-05,
|
| 2758 |
+
"loss": 0.0283,
|
| 2759 |
+
"step": 9500
|
| 2760 |
+
},
|
| 2761 |
+
{
|
| 2762 |
+
"epoch": 10.490088105726873,
|
| 2763 |
+
"grad_norm": 0.15916317701339722,
|
| 2764 |
+
"learning_rate": 1.0743589743589745e-05,
|
| 2765 |
+
"loss": 0.0336,
|
| 2766 |
+
"step": 9525
|
| 2767 |
+
},
|
| 2768 |
+
{
|
| 2769 |
+
"epoch": 10.517621145374449,
|
| 2770 |
+
"grad_norm": 0.3281770944595337,
|
| 2771 |
+
"learning_rate": 1.0717948717948718e-05,
|
| 2772 |
+
"loss": 0.0217,
|
| 2773 |
+
"step": 9550
|
| 2774 |
+
},
|
| 2775 |
+
{
|
| 2776 |
+
"epoch": 10.545154185022026,
|
| 2777 |
+
"grad_norm": 0.12852789461612701,
|
| 2778 |
+
"learning_rate": 1.0692307692307694e-05,
|
| 2779 |
+
"loss": 0.0261,
|
| 2780 |
+
"step": 9575
|
| 2781 |
+
},
|
| 2782 |
+
{
|
| 2783 |
+
"epoch": 10.572687224669604,
|
| 2784 |
+
"grad_norm": 0.3133266270160675,
|
| 2785 |
+
"learning_rate": 1.0666666666666667e-05,
|
| 2786 |
+
"loss": 0.0258,
|
| 2787 |
+
"step": 9600
|
| 2788 |
+
},
|
| 2789 |
+
{
|
| 2790 |
+
"epoch": 10.60022026431718,
|
| 2791 |
+
"grad_norm": 0.49409055709838867,
|
| 2792 |
+
"learning_rate": 1.0641025641025643e-05,
|
| 2793 |
+
"loss": 0.0241,
|
| 2794 |
+
"step": 9625
|
| 2795 |
+
},
|
| 2796 |
+
{
|
| 2797 |
+
"epoch": 10.627753303964758,
|
| 2798 |
+
"grad_norm": 0.1980574131011963,
|
| 2799 |
+
"learning_rate": 1.0615384615384616e-05,
|
| 2800 |
+
"loss": 0.0264,
|
| 2801 |
+
"step": 9650
|
| 2802 |
+
},
|
| 2803 |
+
{
|
| 2804 |
+
"epoch": 10.655286343612335,
|
| 2805 |
+
"grad_norm": 0.3089769184589386,
|
| 2806 |
+
"learning_rate": 1.058974358974359e-05,
|
| 2807 |
+
"loss": 0.0301,
|
| 2808 |
+
"step": 9675
|
| 2809 |
+
},
|
| 2810 |
+
{
|
| 2811 |
+
"epoch": 10.682819383259911,
|
| 2812 |
+
"grad_norm": 0.4599588215351105,
|
| 2813 |
+
"learning_rate": 1.0564102564102565e-05,
|
| 2814 |
+
"loss": 0.0297,
|
| 2815 |
+
"step": 9700
|
| 2816 |
+
},
|
| 2817 |
+
{
|
| 2818 |
+
"epoch": 10.710352422907489,
|
| 2819 |
+
"grad_norm": 0.3897559344768524,
|
| 2820 |
+
"learning_rate": 1.0538461538461539e-05,
|
| 2821 |
+
"loss": 0.026,
|
| 2822 |
+
"step": 9725
|
| 2823 |
+
},
|
| 2824 |
+
{
|
| 2825 |
+
"epoch": 10.737885462555067,
|
| 2826 |
+
"grad_norm": 0.26833340525627136,
|
| 2827 |
+
"learning_rate": 1.0512820512820514e-05,
|
| 2828 |
+
"loss": 0.0297,
|
| 2829 |
+
"step": 9750
|
| 2830 |
+
},
|
| 2831 |
+
{
|
| 2832 |
+
"epoch": 10.765418502202643,
|
| 2833 |
+
"grad_norm": 0.1307441145181656,
|
| 2834 |
+
"learning_rate": 1.0487179487179488e-05,
|
| 2835 |
+
"loss": 0.0269,
|
| 2836 |
+
"step": 9775
|
| 2837 |
+
},
|
| 2838 |
+
{
|
| 2839 |
+
"epoch": 10.79295154185022,
|
| 2840 |
+
"grad_norm": 0.5322534441947937,
|
| 2841 |
+
"learning_rate": 1.0461538461538463e-05,
|
| 2842 |
+
"loss": 0.0308,
|
| 2843 |
+
"step": 9800
|
| 2844 |
+
},
|
| 2845 |
+
{
|
| 2846 |
+
"epoch": 10.820484581497798,
|
| 2847 |
+
"grad_norm": 0.42848142981529236,
|
| 2848 |
+
"learning_rate": 1.0435897435897437e-05,
|
| 2849 |
+
"loss": 0.0264,
|
| 2850 |
+
"step": 9825
|
| 2851 |
+
},
|
| 2852 |
+
{
|
| 2853 |
+
"epoch": 10.848017621145374,
|
| 2854 |
+
"grad_norm": 0.24415309727191925,
|
| 2855 |
+
"learning_rate": 1.0410256410256412e-05,
|
| 2856 |
+
"loss": 0.0258,
|
| 2857 |
+
"step": 9850
|
| 2858 |
+
},
|
| 2859 |
+
{
|
| 2860 |
+
"epoch": 10.875550660792952,
|
| 2861 |
+
"grad_norm": 0.2621011734008789,
|
| 2862 |
+
"learning_rate": 1.0384615384615386e-05,
|
| 2863 |
+
"loss": 0.025,
|
| 2864 |
+
"step": 9875
|
| 2865 |
+
},
|
| 2866 |
+
{
|
| 2867 |
+
"epoch": 10.90308370044053,
|
| 2868 |
+
"grad_norm": 0.3474062383174896,
|
| 2869 |
+
"learning_rate": 1.0358974358974361e-05,
|
| 2870 |
+
"loss": 0.0239,
|
| 2871 |
+
"step": 9900
|
| 2872 |
+
},
|
| 2873 |
+
{
|
| 2874 |
+
"epoch": 10.930616740088105,
|
| 2875 |
+
"grad_norm": 0.16342134773731232,
|
| 2876 |
+
"learning_rate": 1.0333333333333335e-05,
|
| 2877 |
+
"loss": 0.0213,
|
| 2878 |
+
"step": 9925
|
| 2879 |
+
},
|
| 2880 |
+
{
|
| 2881 |
+
"epoch": 10.958149779735683,
|
| 2882 |
+
"grad_norm": 0.12468931823968887,
|
| 2883 |
+
"learning_rate": 1.0307692307692307e-05,
|
| 2884 |
+
"loss": 0.0233,
|
| 2885 |
+
"step": 9950
|
| 2886 |
+
},
|
| 2887 |
+
{
|
| 2888 |
+
"epoch": 10.98568281938326,
|
| 2889 |
+
"grad_norm": 0.42761552333831787,
|
| 2890 |
+
"learning_rate": 1.0282051282051282e-05,
|
| 2891 |
+
"loss": 0.0253,
|
| 2892 |
+
"step": 9975
|
| 2893 |
+
},
|
| 2894 |
+
{
|
| 2895 |
+
"epoch": 11.013215859030836,
|
| 2896 |
+
"grad_norm": 0.16973541676998138,
|
| 2897 |
+
"learning_rate": 1.0256410256410256e-05,
|
| 2898 |
+
"loss": 0.0181,
|
| 2899 |
+
"step": 10000
|
| 2900 |
+
},
|
| 2901 |
+
{
|
| 2902 |
+
"epoch": 11.013215859030836,
|
| 2903 |
+
"eval_cer": 24.361766762578384,
|
| 2904 |
+
"eval_loss": 0.8395898342132568,
|
| 2905 |
+
"eval_runtime": 1712.1817,
|
| 2906 |
+
"eval_samples_per_second": 6.18,
|
| 2907 |
+
"eval_steps_per_second": 1.545,
|
| 2908 |
+
"eval_wer": 83.80009429514381,
|
| 2909 |
+
"step": 10000
|
| 2910 |
}
|
| 2911 |
],
|
| 2912 |
"logging_steps": 25,
|
|
|
|
| 2926 |
"attributes": {}
|
| 2927 |
}
|
| 2928 |
},
|
| 2929 |
+
"total_flos": 1.7181431864900977e+20,
|
| 2930 |
"train_batch_size": 4,
|
| 2931 |
"trial_name": null,
|
| 2932 |
"trial_params": null
|