Training in progress, step 12000, checkpoint
Browse files- last-checkpoint/global_step12000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt +3 -0
- last-checkpoint/global_step12000/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_step12000/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 sha256:9b30b063d6826f7465a6da7cfbedf7fcee136d85769f17c3253ff368ea288e73
|
| 3 |
+
size 761059696
|
last-checkpoint/global_step12000/mp_rank_00_model_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b33dbd695a8cf57adef6d3fb2dd06d6e910781e3bb0c506de326284d8a8445fe
|
| 3 |
+
size 129965712
|
last-checkpoint/latest
CHANGED
|
@@ -1 +1 @@
|
|
| 1 |
-
|
|
|
|
| 1 |
+
global_step12000
|
last-checkpoint/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 181508256
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4cfdaaad7311cc963a64154915312068b27d408ccb4dbe0d7a849a3298fcab86
|
| 3 |
size 181508256
|
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 sha256:92fc487271575fb16a5dab13203a006ee4b441af267eba2302c24332dddb9db6
|
| 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 sha256:5884115119708c269ef22db040f254e6fc9b5f2bf91c9b4125193ea4ee22e90a
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 96.
|
| 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,
|
|
@@ -1457,6 +1457,2036 @@
|
|
| 1457 |
"eval_steps_per_second": 3.082,
|
| 1458 |
"eval_wer": 96.94483734087694,
|
| 1459 |
"step": 5000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1460 |
}
|
| 1461 |
],
|
| 1462 |
"logging_steps": 25,
|
|
@@ -1476,7 +3506,7 @@
|
|
| 1476 |
"attributes": {}
|
| 1477 |
}
|
| 1478 |
},
|
| 1479 |
-
"total_flos":
|
| 1480 |
"train_batch_size": 4,
|
| 1481 |
"trial_name": null,
|
| 1482 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 96.82225365393681,
|
| 3 |
+
"best_model_checkpoint": "./iteboshi_temp/checkpoint-12000",
|
| 4 |
+
"epoch": 13.215859030837004,
|
| 5 |
"eval_steps": 1000,
|
| 6 |
+
"global_step": 12000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 1457 |
"eval_steps_per_second": 3.082,
|
| 1458 |
"eval_wer": 96.94483734087694,
|
| 1459 |
"step": 5000
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 5.534140969162996,
|
| 1463 |
+
"grad_norm": 0.8046155571937561,
|
| 1464 |
+
"learning_rate": 1.535897435897436e-05,
|
| 1465 |
+
"loss": 0.2629,
|
| 1466 |
+
"step": 5025
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 5.561674008810573,
|
| 1470 |
+
"grad_norm": 0.7639729976654053,
|
| 1471 |
+
"learning_rate": 1.5333333333333334e-05,
|
| 1472 |
+
"loss": 0.2815,
|
| 1473 |
+
"step": 5050
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 5.5892070484581495,
|
| 1477 |
+
"grad_norm": 1.1292579174041748,
|
| 1478 |
+
"learning_rate": 1.5307692307692308e-05,
|
| 1479 |
+
"loss": 0.2611,
|
| 1480 |
+
"step": 5075
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 5.616740088105727,
|
| 1484 |
+
"grad_norm": 0.8473858833312988,
|
| 1485 |
+
"learning_rate": 1.5282051282051282e-05,
|
| 1486 |
+
"loss": 0.2603,
|
| 1487 |
+
"step": 5100
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 5.644273127753304,
|
| 1491 |
+
"grad_norm": 0.891690194606781,
|
| 1492 |
+
"learning_rate": 1.5256410256410257e-05,
|
| 1493 |
+
"loss": 0.2592,
|
| 1494 |
+
"step": 5125
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 5.671806167400881,
|
| 1498 |
+
"grad_norm": 0.6996191740036011,
|
| 1499 |
+
"learning_rate": 1.523076923076923e-05,
|
| 1500 |
+
"loss": 0.2717,
|
| 1501 |
+
"step": 5150
|
| 1502 |
+
},
|
| 1503 |
+
{
|
| 1504 |
+
"epoch": 5.6993392070484585,
|
| 1505 |
+
"grad_norm": 0.7721630930900574,
|
| 1506 |
+
"learning_rate": 1.5205128205128206e-05,
|
| 1507 |
+
"loss": 0.2383,
|
| 1508 |
+
"step": 5175
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 5.726872246696035,
|
| 1512 |
+
"grad_norm": 0.7740418910980225,
|
| 1513 |
+
"learning_rate": 1.517948717948718e-05,
|
| 1514 |
+
"loss": 0.2795,
|
| 1515 |
+
"step": 5200
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 5.754405286343612,
|
| 1519 |
+
"grad_norm": 0.7397506833076477,
|
| 1520 |
+
"learning_rate": 1.5153846153846155e-05,
|
| 1521 |
+
"loss": 0.2575,
|
| 1522 |
+
"step": 5225
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 5.78193832599119,
|
| 1526 |
+
"grad_norm": 0.8838066458702087,
|
| 1527 |
+
"learning_rate": 1.5128205128205129e-05,
|
| 1528 |
+
"loss": 0.2873,
|
| 1529 |
+
"step": 5250
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 5.809471365638767,
|
| 1533 |
+
"grad_norm": 1.331363320350647,
|
| 1534 |
+
"learning_rate": 1.5102564102564104e-05,
|
| 1535 |
+
"loss": 0.2664,
|
| 1536 |
+
"step": 5275
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 5.8370044052863435,
|
| 1540 |
+
"grad_norm": 0.8736733794212341,
|
| 1541 |
+
"learning_rate": 1.5076923076923078e-05,
|
| 1542 |
+
"loss": 0.2574,
|
| 1543 |
+
"step": 5300
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 5.864537444933921,
|
| 1547 |
+
"grad_norm": 0.8324550986289978,
|
| 1548 |
+
"learning_rate": 1.5051282051282053e-05,
|
| 1549 |
+
"loss": 0.2631,
|
| 1550 |
+
"step": 5325
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 5.892070484581498,
|
| 1554 |
+
"grad_norm": 0.6863672137260437,
|
| 1555 |
+
"learning_rate": 1.5025641025641027e-05,
|
| 1556 |
+
"loss": 0.2693,
|
| 1557 |
+
"step": 5350
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 5.919603524229075,
|
| 1561 |
+
"grad_norm": 0.6372384428977966,
|
| 1562 |
+
"learning_rate": 1.5000000000000002e-05,
|
| 1563 |
+
"loss": 0.2449,
|
| 1564 |
+
"step": 5375
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 5.9471365638766525,
|
| 1568 |
+
"grad_norm": 0.8350533246994019,
|
| 1569 |
+
"learning_rate": 1.4974358974358976e-05,
|
| 1570 |
+
"loss": 0.256,
|
| 1571 |
+
"step": 5400
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 5.974669603524229,
|
| 1575 |
+
"grad_norm": 0.7431644797325134,
|
| 1576 |
+
"learning_rate": 1.494871794871795e-05,
|
| 1577 |
+
"loss": 0.2448,
|
| 1578 |
+
"step": 5425
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 6.002202643171806,
|
| 1582 |
+
"grad_norm": 0.49462586641311646,
|
| 1583 |
+
"learning_rate": 1.4923076923076925e-05,
|
| 1584 |
+
"loss": 0.2238,
|
| 1585 |
+
"step": 5450
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 6.029735682819383,
|
| 1589 |
+
"grad_norm": 0.7905895709991455,
|
| 1590 |
+
"learning_rate": 1.4897435897435898e-05,
|
| 1591 |
+
"loss": 0.1734,
|
| 1592 |
+
"step": 5475
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 6.057268722466961,
|
| 1596 |
+
"grad_norm": 0.6685343384742737,
|
| 1597 |
+
"learning_rate": 1.4871794871794874e-05,
|
| 1598 |
+
"loss": 0.1577,
|
| 1599 |
+
"step": 5500
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 6.084801762114537,
|
| 1603 |
+
"grad_norm": 0.6416425108909607,
|
| 1604 |
+
"learning_rate": 1.4846153846153847e-05,
|
| 1605 |
+
"loss": 0.1845,
|
| 1606 |
+
"step": 5525
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 6.112334801762114,
|
| 1610 |
+
"grad_norm": 0.584987998008728,
|
| 1611 |
+
"learning_rate": 1.4820512820512823e-05,
|
| 1612 |
+
"loss": 0.1899,
|
| 1613 |
+
"step": 5550
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 6.139867841409692,
|
| 1617 |
+
"grad_norm": 0.49699297547340393,
|
| 1618 |
+
"learning_rate": 1.4794871794871796e-05,
|
| 1619 |
+
"loss": 0.1774,
|
| 1620 |
+
"step": 5575
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 6.167400881057269,
|
| 1624 |
+
"grad_norm": 0.7456440925598145,
|
| 1625 |
+
"learning_rate": 1.4769230769230772e-05,
|
| 1626 |
+
"loss": 0.1977,
|
| 1627 |
+
"step": 5600
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 6.1949339207048455,
|
| 1631 |
+
"grad_norm": 0.5305894613265991,
|
| 1632 |
+
"learning_rate": 1.4743589743589745e-05,
|
| 1633 |
+
"loss": 0.1544,
|
| 1634 |
+
"step": 5625
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 6.222466960352423,
|
| 1638 |
+
"grad_norm": 0.6994619369506836,
|
| 1639 |
+
"learning_rate": 1.471794871794872e-05,
|
| 1640 |
+
"loss": 0.1964,
|
| 1641 |
+
"step": 5650
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 6.25,
|
| 1645 |
+
"grad_norm": 0.6842361092567444,
|
| 1646 |
+
"learning_rate": 1.4692307692307694e-05,
|
| 1647 |
+
"loss": 0.1873,
|
| 1648 |
+
"step": 5675
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 6.277533039647577,
|
| 1652 |
+
"grad_norm": 0.5287600159645081,
|
| 1653 |
+
"learning_rate": 1.4666666666666666e-05,
|
| 1654 |
+
"loss": 0.16,
|
| 1655 |
+
"step": 5700
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 6.3050660792951545,
|
| 1659 |
+
"grad_norm": 0.5247631669044495,
|
| 1660 |
+
"learning_rate": 1.4641025641025642e-05,
|
| 1661 |
+
"loss": 0.1849,
|
| 1662 |
+
"step": 5725
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 6.332599118942731,
|
| 1666 |
+
"grad_norm": 0.7973225712776184,
|
| 1667 |
+
"learning_rate": 1.4615384615384615e-05,
|
| 1668 |
+
"loss": 0.17,
|
| 1669 |
+
"step": 5750
|
| 1670 |
+
},
|
| 1671 |
+
{
|
| 1672 |
+
"epoch": 6.360132158590308,
|
| 1673 |
+
"grad_norm": 0.5792837142944336,
|
| 1674 |
+
"learning_rate": 1.458974358974359e-05,
|
| 1675 |
+
"loss": 0.1766,
|
| 1676 |
+
"step": 5775
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 6.387665198237886,
|
| 1680 |
+
"grad_norm": 0.6392947435379028,
|
| 1681 |
+
"learning_rate": 1.4564102564102564e-05,
|
| 1682 |
+
"loss": 0.1888,
|
| 1683 |
+
"step": 5800
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 6.415198237885463,
|
| 1687 |
+
"grad_norm": 0.5919508337974548,
|
| 1688 |
+
"learning_rate": 1.453846153846154e-05,
|
| 1689 |
+
"loss": 0.1729,
|
| 1690 |
+
"step": 5825
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"epoch": 6.442731277533039,
|
| 1694 |
+
"grad_norm": 0.5749716758728027,
|
| 1695 |
+
"learning_rate": 1.4512820512820513e-05,
|
| 1696 |
+
"loss": 0.1929,
|
| 1697 |
+
"step": 5850
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 6.470264317180617,
|
| 1701 |
+
"grad_norm": 0.7120850682258606,
|
| 1702 |
+
"learning_rate": 1.4487179487179489e-05,
|
| 1703 |
+
"loss": 0.2032,
|
| 1704 |
+
"step": 5875
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 6.497797356828194,
|
| 1708 |
+
"grad_norm": 0.7158252596855164,
|
| 1709 |
+
"learning_rate": 1.4461538461538462e-05,
|
| 1710 |
+
"loss": 0.1861,
|
| 1711 |
+
"step": 5900
|
| 1712 |
+
},
|
| 1713 |
+
{
|
| 1714 |
+
"epoch": 6.525330396475771,
|
| 1715 |
+
"grad_norm": 0.6715514659881592,
|
| 1716 |
+
"learning_rate": 1.4435897435897438e-05,
|
| 1717 |
+
"loss": 0.2071,
|
| 1718 |
+
"step": 5925
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 6.5528634361233475,
|
| 1722 |
+
"grad_norm": 0.5938199758529663,
|
| 1723 |
+
"learning_rate": 1.4410256410256411e-05,
|
| 1724 |
+
"loss": 0.1839,
|
| 1725 |
+
"step": 5950
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 6.580396475770925,
|
| 1729 |
+
"grad_norm": 0.7464186549186707,
|
| 1730 |
+
"learning_rate": 1.4384615384615387e-05,
|
| 1731 |
+
"loss": 0.1855,
|
| 1732 |
+
"step": 5975
|
| 1733 |
+
},
|
| 1734 |
+
{
|
| 1735 |
+
"epoch": 6.607929515418502,
|
| 1736 |
+
"grad_norm": 0.6544903516769409,
|
| 1737 |
+
"learning_rate": 1.435897435897436e-05,
|
| 1738 |
+
"loss": 0.1854,
|
| 1739 |
+
"step": 6000
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 6.607929515418502,
|
| 1743 |
+
"eval_cer": 73.37303543383811,
|
| 1744 |
+
"eval_loss": 0.8310097455978394,
|
| 1745 |
+
"eval_runtime": 933.6287,
|
| 1746 |
+
"eval_samples_per_second": 11.333,
|
| 1747 |
+
"eval_steps_per_second": 2.834,
|
| 1748 |
+
"eval_wer": 97.99151343705799,
|
| 1749 |
+
"step": 6000
|
| 1750 |
+
},
|
| 1751 |
+
{
|
| 1752 |
+
"epoch": 6.635462555066079,
|
| 1753 |
+
"grad_norm": 0.625103771686554,
|
| 1754 |
+
"learning_rate": 1.4333333333333334e-05,
|
| 1755 |
+
"loss": 0.1653,
|
| 1756 |
+
"step": 6025
|
| 1757 |
+
},
|
| 1758 |
+
{
|
| 1759 |
+
"epoch": 6.6629955947136565,
|
| 1760 |
+
"grad_norm": 0.540448009967804,
|
| 1761 |
+
"learning_rate": 1.430769230769231e-05,
|
| 1762 |
+
"loss": 0.2075,
|
| 1763 |
+
"step": 6050
|
| 1764 |
+
},
|
| 1765 |
+
{
|
| 1766 |
+
"epoch": 6.690528634361233,
|
| 1767 |
+
"grad_norm": 0.8601000308990479,
|
| 1768 |
+
"learning_rate": 1.4282051282051283e-05,
|
| 1769 |
+
"loss": 0.2174,
|
| 1770 |
+
"step": 6075
|
| 1771 |
+
},
|
| 1772 |
+
{
|
| 1773 |
+
"epoch": 6.71806167400881,
|
| 1774 |
+
"grad_norm": 0.8207329511642456,
|
| 1775 |
+
"learning_rate": 1.4256410256410258e-05,
|
| 1776 |
+
"loss": 0.2157,
|
| 1777 |
+
"step": 6100
|
| 1778 |
+
},
|
| 1779 |
+
{
|
| 1780 |
+
"epoch": 6.745594713656388,
|
| 1781 |
+
"grad_norm": 0.6145961880683899,
|
| 1782 |
+
"learning_rate": 1.4230769230769232e-05,
|
| 1783 |
+
"loss": 0.2074,
|
| 1784 |
+
"step": 6125
|
| 1785 |
+
},
|
| 1786 |
+
{
|
| 1787 |
+
"epoch": 6.773127753303965,
|
| 1788 |
+
"grad_norm": 0.7318241596221924,
|
| 1789 |
+
"learning_rate": 1.4205128205128207e-05,
|
| 1790 |
+
"loss": 0.1896,
|
| 1791 |
+
"step": 6150
|
| 1792 |
+
},
|
| 1793 |
+
{
|
| 1794 |
+
"epoch": 6.8006607929515415,
|
| 1795 |
+
"grad_norm": 0.7682480216026306,
|
| 1796 |
+
"learning_rate": 1.4179487179487181e-05,
|
| 1797 |
+
"loss": 0.199,
|
| 1798 |
+
"step": 6175
|
| 1799 |
+
},
|
| 1800 |
+
{
|
| 1801 |
+
"epoch": 6.828193832599119,
|
| 1802 |
+
"grad_norm": 0.6676972508430481,
|
| 1803 |
+
"learning_rate": 1.4153846153846156e-05,
|
| 1804 |
+
"loss": 0.1742,
|
| 1805 |
+
"step": 6200
|
| 1806 |
+
},
|
| 1807 |
+
{
|
| 1808 |
+
"epoch": 6.855726872246696,
|
| 1809 |
+
"grad_norm": 0.6792747974395752,
|
| 1810 |
+
"learning_rate": 1.412820512820513e-05,
|
| 1811 |
+
"loss": 0.2096,
|
| 1812 |
+
"step": 6225
|
| 1813 |
+
},
|
| 1814 |
+
{
|
| 1815 |
+
"epoch": 6.883259911894273,
|
| 1816 |
+
"grad_norm": 0.5820185542106628,
|
| 1817 |
+
"learning_rate": 1.4102564102564105e-05,
|
| 1818 |
+
"loss": 0.1843,
|
| 1819 |
+
"step": 6250
|
| 1820 |
+
},
|
| 1821 |
+
{
|
| 1822 |
+
"epoch": 6.9107929515418505,
|
| 1823 |
+
"grad_norm": 0.6929712295532227,
|
| 1824 |
+
"learning_rate": 1.4076923076923079e-05,
|
| 1825 |
+
"loss": 0.1797,
|
| 1826 |
+
"step": 6275
|
| 1827 |
+
},
|
| 1828 |
+
{
|
| 1829 |
+
"epoch": 6.938325991189427,
|
| 1830 |
+
"grad_norm": 0.7363738417625427,
|
| 1831 |
+
"learning_rate": 1.405128205128205e-05,
|
| 1832 |
+
"loss": 0.1823,
|
| 1833 |
+
"step": 6300
|
| 1834 |
+
},
|
| 1835 |
+
{
|
| 1836 |
+
"epoch": 6.965859030837004,
|
| 1837 |
+
"grad_norm": 0.708763599395752,
|
| 1838 |
+
"learning_rate": 1.4025641025641026e-05,
|
| 1839 |
+
"loss": 0.1856,
|
| 1840 |
+
"step": 6325
|
| 1841 |
+
},
|
| 1842 |
+
{
|
| 1843 |
+
"epoch": 6.993392070484582,
|
| 1844 |
+
"grad_norm": 0.6946198344230652,
|
| 1845 |
+
"learning_rate": 1.4e-05,
|
| 1846 |
+
"loss": 0.1942,
|
| 1847 |
+
"step": 6350
|
| 1848 |
+
},
|
| 1849 |
+
{
|
| 1850 |
+
"epoch": 7.020925110132159,
|
| 1851 |
+
"grad_norm": 0.5003654956817627,
|
| 1852 |
+
"learning_rate": 1.3974358974358975e-05,
|
| 1853 |
+
"loss": 0.131,
|
| 1854 |
+
"step": 6375
|
| 1855 |
+
},
|
| 1856 |
+
{
|
| 1857 |
+
"epoch": 7.048458149779735,
|
| 1858 |
+
"grad_norm": 0.6626909375190735,
|
| 1859 |
+
"learning_rate": 1.3948717948717949e-05,
|
| 1860 |
+
"loss": 0.1278,
|
| 1861 |
+
"step": 6400
|
| 1862 |
+
},
|
| 1863 |
+
{
|
| 1864 |
+
"epoch": 7.075991189427313,
|
| 1865 |
+
"grad_norm": 0.36942312121391296,
|
| 1866 |
+
"learning_rate": 1.3923076923076924e-05,
|
| 1867 |
+
"loss": 0.1111,
|
| 1868 |
+
"step": 6425
|
| 1869 |
+
},
|
| 1870 |
+
{
|
| 1871 |
+
"epoch": 7.10352422907489,
|
| 1872 |
+
"grad_norm": 0.5009227395057678,
|
| 1873 |
+
"learning_rate": 1.3897435897435898e-05,
|
| 1874 |
+
"loss": 0.1099,
|
| 1875 |
+
"step": 6450
|
| 1876 |
+
},
|
| 1877 |
+
{
|
| 1878 |
+
"epoch": 7.131057268722467,
|
| 1879 |
+
"grad_norm": 0.3141847848892212,
|
| 1880 |
+
"learning_rate": 1.3871794871794873e-05,
|
| 1881 |
+
"loss": 0.1193,
|
| 1882 |
+
"step": 6475
|
| 1883 |
+
},
|
| 1884 |
+
{
|
| 1885 |
+
"epoch": 7.158590308370044,
|
| 1886 |
+
"grad_norm": 0.6159384250640869,
|
| 1887 |
+
"learning_rate": 1.3846153846153847e-05,
|
| 1888 |
+
"loss": 0.1305,
|
| 1889 |
+
"step": 6500
|
| 1890 |
+
},
|
| 1891 |
+
{
|
| 1892 |
+
"epoch": 7.186123348017621,
|
| 1893 |
+
"grad_norm": 0.7549660801887512,
|
| 1894 |
+
"learning_rate": 1.3820512820512822e-05,
|
| 1895 |
+
"loss": 0.1271,
|
| 1896 |
+
"step": 6525
|
| 1897 |
+
},
|
| 1898 |
+
{
|
| 1899 |
+
"epoch": 7.213656387665198,
|
| 1900 |
+
"grad_norm": 0.9365887641906738,
|
| 1901 |
+
"learning_rate": 1.3794871794871796e-05,
|
| 1902 |
+
"loss": 0.12,
|
| 1903 |
+
"step": 6550
|
| 1904 |
+
},
|
| 1905 |
+
{
|
| 1906 |
+
"epoch": 7.241189427312776,
|
| 1907 |
+
"grad_norm": 0.42107388377189636,
|
| 1908 |
+
"learning_rate": 1.3769230769230771e-05,
|
| 1909 |
+
"loss": 0.1303,
|
| 1910 |
+
"step": 6575
|
| 1911 |
+
},
|
| 1912 |
+
{
|
| 1913 |
+
"epoch": 7.2687224669603525,
|
| 1914 |
+
"grad_norm": 0.6491579413414001,
|
| 1915 |
+
"learning_rate": 1.3743589743589745e-05,
|
| 1916 |
+
"loss": 0.1284,
|
| 1917 |
+
"step": 6600
|
| 1918 |
+
},
|
| 1919 |
+
{
|
| 1920 |
+
"epoch": 7.296255506607929,
|
| 1921 |
+
"grad_norm": 0.5241743922233582,
|
| 1922 |
+
"learning_rate": 1.3717948717948718e-05,
|
| 1923 |
+
"loss": 0.1352,
|
| 1924 |
+
"step": 6625
|
| 1925 |
+
},
|
| 1926 |
+
{
|
| 1927 |
+
"epoch": 7.323788546255507,
|
| 1928 |
+
"grad_norm": 0.5401379466056824,
|
| 1929 |
+
"learning_rate": 1.3692307692307694e-05,
|
| 1930 |
+
"loss": 0.1305,
|
| 1931 |
+
"step": 6650
|
| 1932 |
+
},
|
| 1933 |
+
{
|
| 1934 |
+
"epoch": 7.351321585903084,
|
| 1935 |
+
"grad_norm": 0.8926748037338257,
|
| 1936 |
+
"learning_rate": 1.3666666666666667e-05,
|
| 1937 |
+
"loss": 0.1414,
|
| 1938 |
+
"step": 6675
|
| 1939 |
+
},
|
| 1940 |
+
{
|
| 1941 |
+
"epoch": 7.378854625550661,
|
| 1942 |
+
"grad_norm": 0.4713541567325592,
|
| 1943 |
+
"learning_rate": 1.3641025641025643e-05,
|
| 1944 |
+
"loss": 0.1152,
|
| 1945 |
+
"step": 6700
|
| 1946 |
+
},
|
| 1947 |
+
{
|
| 1948 |
+
"epoch": 7.406387665198238,
|
| 1949 |
+
"grad_norm": 0.4151066839694977,
|
| 1950 |
+
"learning_rate": 1.3615384615384616e-05,
|
| 1951 |
+
"loss": 0.1123,
|
| 1952 |
+
"step": 6725
|
| 1953 |
+
},
|
| 1954 |
+
{
|
| 1955 |
+
"epoch": 7.433920704845815,
|
| 1956 |
+
"grad_norm": 0.5809412598609924,
|
| 1957 |
+
"learning_rate": 1.3589743589743592e-05,
|
| 1958 |
+
"loss": 0.1342,
|
| 1959 |
+
"step": 6750
|
| 1960 |
+
},
|
| 1961 |
+
{
|
| 1962 |
+
"epoch": 7.461453744493392,
|
| 1963 |
+
"grad_norm": 0.5300701856613159,
|
| 1964 |
+
"learning_rate": 1.3564102564102565e-05,
|
| 1965 |
+
"loss": 0.1394,
|
| 1966 |
+
"step": 6775
|
| 1967 |
+
},
|
| 1968 |
+
{
|
| 1969 |
+
"epoch": 7.48898678414097,
|
| 1970 |
+
"grad_norm": 0.5370484590530396,
|
| 1971 |
+
"learning_rate": 1.353846153846154e-05,
|
| 1972 |
+
"loss": 0.1337,
|
| 1973 |
+
"step": 6800
|
| 1974 |
+
},
|
| 1975 |
+
{
|
| 1976 |
+
"epoch": 7.516519823788546,
|
| 1977 |
+
"grad_norm": 0.6923957467079163,
|
| 1978 |
+
"learning_rate": 1.3512820512820514e-05,
|
| 1979 |
+
"loss": 0.1314,
|
| 1980 |
+
"step": 6825
|
| 1981 |
+
},
|
| 1982 |
+
{
|
| 1983 |
+
"epoch": 7.544052863436123,
|
| 1984 |
+
"grad_norm": 0.35029956698417664,
|
| 1985 |
+
"learning_rate": 1.348717948717949e-05,
|
| 1986 |
+
"loss": 0.128,
|
| 1987 |
+
"step": 6850
|
| 1988 |
+
},
|
| 1989 |
+
{
|
| 1990 |
+
"epoch": 7.5715859030837,
|
| 1991 |
+
"grad_norm": 0.5914937853813171,
|
| 1992 |
+
"learning_rate": 1.3461538461538463e-05,
|
| 1993 |
+
"loss": 0.121,
|
| 1994 |
+
"step": 6875
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 7.599118942731278,
|
| 1998 |
+
"grad_norm": 0.5608924627304077,
|
| 1999 |
+
"learning_rate": 1.3435897435897435e-05,
|
| 2000 |
+
"loss": 0.1439,
|
| 2001 |
+
"step": 6900
|
| 2002 |
+
},
|
| 2003 |
+
{
|
| 2004 |
+
"epoch": 7.6266519823788546,
|
| 2005 |
+
"grad_norm": 0.38419216871261597,
|
| 2006 |
+
"learning_rate": 1.341025641025641e-05,
|
| 2007 |
+
"loss": 0.1195,
|
| 2008 |
+
"step": 6925
|
| 2009 |
+
},
|
| 2010 |
+
{
|
| 2011 |
+
"epoch": 7.654185022026431,
|
| 2012 |
+
"grad_norm": 0.6647460460662842,
|
| 2013 |
+
"learning_rate": 1.3384615384615384e-05,
|
| 2014 |
+
"loss": 0.1501,
|
| 2015 |
+
"step": 6950
|
| 2016 |
+
},
|
| 2017 |
+
{
|
| 2018 |
+
"epoch": 7.681718061674009,
|
| 2019 |
+
"grad_norm": 0.5829716324806213,
|
| 2020 |
+
"learning_rate": 1.335897435897436e-05,
|
| 2021 |
+
"loss": 0.1402,
|
| 2022 |
+
"step": 6975
|
| 2023 |
+
},
|
| 2024 |
+
{
|
| 2025 |
+
"epoch": 7.709251101321586,
|
| 2026 |
+
"grad_norm": 0.7427933216094971,
|
| 2027 |
+
"learning_rate": 1.3333333333333333e-05,
|
| 2028 |
+
"loss": 0.1173,
|
| 2029 |
+
"step": 7000
|
| 2030 |
+
},
|
| 2031 |
+
{
|
| 2032 |
+
"epoch": 7.709251101321586,
|
| 2033 |
+
"eval_cer": 64.11449795375603,
|
| 2034 |
+
"eval_loss": 0.8565544486045837,
|
| 2035 |
+
"eval_runtime": 853.5046,
|
| 2036 |
+
"eval_samples_per_second": 12.397,
|
| 2037 |
+
"eval_steps_per_second": 3.1,
|
| 2038 |
+
"eval_wer": 97.64262140499764,
|
| 2039 |
+
"step": 7000
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 7.736784140969163,
|
| 2043 |
+
"grad_norm": 0.7469947338104248,
|
| 2044 |
+
"learning_rate": 1.3307692307692309e-05,
|
| 2045 |
+
"loss": 0.146,
|
| 2046 |
+
"step": 7025
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"epoch": 7.76431718061674,
|
| 2050 |
+
"grad_norm": 0.962263822555542,
|
| 2051 |
+
"learning_rate": 1.3282051282051282e-05,
|
| 2052 |
+
"loss": 0.1349,
|
| 2053 |
+
"step": 7050
|
| 2054 |
+
},
|
| 2055 |
+
{
|
| 2056 |
+
"epoch": 7.791850220264317,
|
| 2057 |
+
"grad_norm": 0.7025478482246399,
|
| 2058 |
+
"learning_rate": 1.3256410256410258e-05,
|
| 2059 |
+
"loss": 0.129,
|
| 2060 |
+
"step": 7075
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 7.819383259911894,
|
| 2064 |
+
"grad_norm": 0.749464750289917,
|
| 2065 |
+
"learning_rate": 1.3230769230769231e-05,
|
| 2066 |
+
"loss": 0.1276,
|
| 2067 |
+
"step": 7100
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 7.846916299559472,
|
| 2071 |
+
"grad_norm": 0.6161931157112122,
|
| 2072 |
+
"learning_rate": 1.3205128205128207e-05,
|
| 2073 |
+
"loss": 0.1196,
|
| 2074 |
+
"step": 7125
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 7.8744493392070485,
|
| 2078 |
+
"grad_norm": 0.43383580446243286,
|
| 2079 |
+
"learning_rate": 1.317948717948718e-05,
|
| 2080 |
+
"loss": 0.1511,
|
| 2081 |
+
"step": 7150
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 7.901982378854625,
|
| 2085 |
+
"grad_norm": 0.5156465768814087,
|
| 2086 |
+
"learning_rate": 1.3153846153846156e-05,
|
| 2087 |
+
"loss": 0.141,
|
| 2088 |
+
"step": 7175
|
| 2089 |
+
},
|
| 2090 |
+
{
|
| 2091 |
+
"epoch": 7.929515418502203,
|
| 2092 |
+
"grad_norm": 0.6636006236076355,
|
| 2093 |
+
"learning_rate": 1.312820512820513e-05,
|
| 2094 |
+
"loss": 0.152,
|
| 2095 |
+
"step": 7200
|
| 2096 |
+
},
|
| 2097 |
+
{
|
| 2098 |
+
"epoch": 7.95704845814978,
|
| 2099 |
+
"grad_norm": 0.6761754751205444,
|
| 2100 |
+
"learning_rate": 1.3102564102564103e-05,
|
| 2101 |
+
"loss": 0.1379,
|
| 2102 |
+
"step": 7225
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 7.984581497797357,
|
| 2106 |
+
"grad_norm": 0.5854198336601257,
|
| 2107 |
+
"learning_rate": 1.3076923076923078e-05,
|
| 2108 |
+
"loss": 0.1406,
|
| 2109 |
+
"step": 7250
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 8.012114537444933,
|
| 2113 |
+
"grad_norm": 0.3060972988605499,
|
| 2114 |
+
"learning_rate": 1.3051282051282052e-05,
|
| 2115 |
+
"loss": 0.1423,
|
| 2116 |
+
"step": 7275
|
| 2117 |
+
},
|
| 2118 |
+
{
|
| 2119 |
+
"epoch": 8.039647577092511,
|
| 2120 |
+
"grad_norm": 0.37324124574661255,
|
| 2121 |
+
"learning_rate": 1.3025641025641027e-05,
|
| 2122 |
+
"loss": 0.0909,
|
| 2123 |
+
"step": 7300
|
| 2124 |
+
},
|
| 2125 |
+
{
|
| 2126 |
+
"epoch": 8.067180616740089,
|
| 2127 |
+
"grad_norm": 0.4420897960662842,
|
| 2128 |
+
"learning_rate": 1.3000000000000001e-05,
|
| 2129 |
+
"loss": 0.091,
|
| 2130 |
+
"step": 7325
|
| 2131 |
+
},
|
| 2132 |
+
{
|
| 2133 |
+
"epoch": 8.094713656387665,
|
| 2134 |
+
"grad_norm": 0.5292598605155945,
|
| 2135 |
+
"learning_rate": 1.2974358974358976e-05,
|
| 2136 |
+
"loss": 0.088,
|
| 2137 |
+
"step": 7350
|
| 2138 |
+
},
|
| 2139 |
+
{
|
| 2140 |
+
"epoch": 8.122246696035242,
|
| 2141 |
+
"grad_norm": 0.6191997528076172,
|
| 2142 |
+
"learning_rate": 1.294871794871795e-05,
|
| 2143 |
+
"loss": 0.0857,
|
| 2144 |
+
"step": 7375
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 8.14977973568282,
|
| 2148 |
+
"grad_norm": 0.4383416473865509,
|
| 2149 |
+
"learning_rate": 1.2923076923076925e-05,
|
| 2150 |
+
"loss": 0.0832,
|
| 2151 |
+
"step": 7400
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 8.177312775330396,
|
| 2155 |
+
"grad_norm": 0.2875024378299713,
|
| 2156 |
+
"learning_rate": 1.2897435897435899e-05,
|
| 2157 |
+
"loss": 0.0872,
|
| 2158 |
+
"step": 7425
|
| 2159 |
+
},
|
| 2160 |
+
{
|
| 2161 |
+
"epoch": 8.204845814977974,
|
| 2162 |
+
"grad_norm": 0.2749011218547821,
|
| 2163 |
+
"learning_rate": 1.2871794871794874e-05,
|
| 2164 |
+
"loss": 0.0784,
|
| 2165 |
+
"step": 7450
|
| 2166 |
+
},
|
| 2167 |
+
{
|
| 2168 |
+
"epoch": 8.232378854625551,
|
| 2169 |
+
"grad_norm": 0.6115961074829102,
|
| 2170 |
+
"learning_rate": 1.2846153846153848e-05,
|
| 2171 |
+
"loss": 0.1014,
|
| 2172 |
+
"step": 7475
|
| 2173 |
+
},
|
| 2174 |
+
{
|
| 2175 |
+
"epoch": 8.259911894273127,
|
| 2176 |
+
"grad_norm": 0.30915337800979614,
|
| 2177 |
+
"learning_rate": 1.2820512820512823e-05,
|
| 2178 |
+
"loss": 0.0788,
|
| 2179 |
+
"step": 7500
|
| 2180 |
+
},
|
| 2181 |
+
{
|
| 2182 |
+
"epoch": 8.287444933920705,
|
| 2183 |
+
"grad_norm": 0.8303879499435425,
|
| 2184 |
+
"learning_rate": 1.2794871794871795e-05,
|
| 2185 |
+
"loss": 0.0815,
|
| 2186 |
+
"step": 7525
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 8.314977973568283,
|
| 2190 |
+
"grad_norm": 0.48807233572006226,
|
| 2191 |
+
"learning_rate": 1.2769230769230769e-05,
|
| 2192 |
+
"loss": 0.083,
|
| 2193 |
+
"step": 7550
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 8.342511013215859,
|
| 2197 |
+
"grad_norm": 0.5944768786430359,
|
| 2198 |
+
"learning_rate": 1.2743589743589744e-05,
|
| 2199 |
+
"loss": 0.0873,
|
| 2200 |
+
"step": 7575
|
| 2201 |
+
},
|
| 2202 |
+
{
|
| 2203 |
+
"epoch": 8.370044052863436,
|
| 2204 |
+
"grad_norm": 0.4144470691680908,
|
| 2205 |
+
"learning_rate": 1.2717948717948718e-05,
|
| 2206 |
+
"loss": 0.0749,
|
| 2207 |
+
"step": 7600
|
| 2208 |
+
},
|
| 2209 |
+
{
|
| 2210 |
+
"epoch": 8.397577092511014,
|
| 2211 |
+
"grad_norm": 0.6508732438087463,
|
| 2212 |
+
"learning_rate": 1.2692307692307693e-05,
|
| 2213 |
+
"loss": 0.0946,
|
| 2214 |
+
"step": 7625
|
| 2215 |
+
},
|
| 2216 |
+
{
|
| 2217 |
+
"epoch": 8.42511013215859,
|
| 2218 |
+
"grad_norm": 0.3488626778125763,
|
| 2219 |
+
"learning_rate": 1.2666666666666667e-05,
|
| 2220 |
+
"loss": 0.081,
|
| 2221 |
+
"step": 7650
|
| 2222 |
+
},
|
| 2223 |
+
{
|
| 2224 |
+
"epoch": 8.452643171806168,
|
| 2225 |
+
"grad_norm": 0.6012532114982605,
|
| 2226 |
+
"learning_rate": 1.2641025641025642e-05,
|
| 2227 |
+
"loss": 0.0875,
|
| 2228 |
+
"step": 7675
|
| 2229 |
+
},
|
| 2230 |
+
{
|
| 2231 |
+
"epoch": 8.480176211453745,
|
| 2232 |
+
"grad_norm": 0.43506327271461487,
|
| 2233 |
+
"learning_rate": 1.2615384615384616e-05,
|
| 2234 |
+
"loss": 0.0868,
|
| 2235 |
+
"step": 7700
|
| 2236 |
+
},
|
| 2237 |
+
{
|
| 2238 |
+
"epoch": 8.507709251101321,
|
| 2239 |
+
"grad_norm": 0.35185477137565613,
|
| 2240 |
+
"learning_rate": 1.2589743589743591e-05,
|
| 2241 |
+
"loss": 0.0842,
|
| 2242 |
+
"step": 7725
|
| 2243 |
+
},
|
| 2244 |
+
{
|
| 2245 |
+
"epoch": 8.535242290748899,
|
| 2246 |
+
"grad_norm": 0.6967858076095581,
|
| 2247 |
+
"learning_rate": 1.2564102564102565e-05,
|
| 2248 |
+
"loss": 0.0957,
|
| 2249 |
+
"step": 7750
|
| 2250 |
+
},
|
| 2251 |
+
{
|
| 2252 |
+
"epoch": 8.562775330396477,
|
| 2253 |
+
"grad_norm": 0.5462283492088318,
|
| 2254 |
+
"learning_rate": 1.253846153846154e-05,
|
| 2255 |
+
"loss": 0.0942,
|
| 2256 |
+
"step": 7775
|
| 2257 |
+
},
|
| 2258 |
+
{
|
| 2259 |
+
"epoch": 8.590308370044053,
|
| 2260 |
+
"grad_norm": 0.524748682975769,
|
| 2261 |
+
"learning_rate": 1.2512820512820514e-05,
|
| 2262 |
+
"loss": 0.1017,
|
| 2263 |
+
"step": 7800
|
| 2264 |
+
},
|
| 2265 |
+
{
|
| 2266 |
+
"epoch": 8.61784140969163,
|
| 2267 |
+
"grad_norm": 0.3966872990131378,
|
| 2268 |
+
"learning_rate": 1.2487179487179487e-05,
|
| 2269 |
+
"loss": 0.1023,
|
| 2270 |
+
"step": 7825
|
| 2271 |
+
},
|
| 2272 |
+
{
|
| 2273 |
+
"epoch": 8.645374449339208,
|
| 2274 |
+
"grad_norm": 0.9616275429725647,
|
| 2275 |
+
"learning_rate": 1.2461538461538463e-05,
|
| 2276 |
+
"loss": 0.0797,
|
| 2277 |
+
"step": 7850
|
| 2278 |
+
},
|
| 2279 |
+
{
|
| 2280 |
+
"epoch": 8.672907488986784,
|
| 2281 |
+
"grad_norm": 0.3389146029949188,
|
| 2282 |
+
"learning_rate": 1.2435897435897436e-05,
|
| 2283 |
+
"loss": 0.0753,
|
| 2284 |
+
"step": 7875
|
| 2285 |
+
},
|
| 2286 |
+
{
|
| 2287 |
+
"epoch": 8.700440528634362,
|
| 2288 |
+
"grad_norm": 0.7579898834228516,
|
| 2289 |
+
"learning_rate": 1.2410256410256412e-05,
|
| 2290 |
+
"loss": 0.0971,
|
| 2291 |
+
"step": 7900
|
| 2292 |
+
},
|
| 2293 |
+
{
|
| 2294 |
+
"epoch": 8.72797356828194,
|
| 2295 |
+
"grad_norm": 0.43244099617004395,
|
| 2296 |
+
"learning_rate": 1.2384615384615385e-05,
|
| 2297 |
+
"loss": 0.0939,
|
| 2298 |
+
"step": 7925
|
| 2299 |
+
},
|
| 2300 |
+
{
|
| 2301 |
+
"epoch": 8.755506607929515,
|
| 2302 |
+
"grad_norm": 0.4598844051361084,
|
| 2303 |
+
"learning_rate": 1.235897435897436e-05,
|
| 2304 |
+
"loss": 0.0944,
|
| 2305 |
+
"step": 7950
|
| 2306 |
+
},
|
| 2307 |
+
{
|
| 2308 |
+
"epoch": 8.783039647577093,
|
| 2309 |
+
"grad_norm": 0.5808974504470825,
|
| 2310 |
+
"learning_rate": 1.2333333333333334e-05,
|
| 2311 |
+
"loss": 0.0958,
|
| 2312 |
+
"step": 7975
|
| 2313 |
+
},
|
| 2314 |
+
{
|
| 2315 |
+
"epoch": 8.810572687224669,
|
| 2316 |
+
"grad_norm": 0.8299708366394043,
|
| 2317 |
+
"learning_rate": 1.230769230769231e-05,
|
| 2318 |
+
"loss": 0.1049,
|
| 2319 |
+
"step": 8000
|
| 2320 |
+
},
|
| 2321 |
+
{
|
| 2322 |
+
"epoch": 8.810572687224669,
|
| 2323 |
+
"eval_cer": 70.65044191991564,
|
| 2324 |
+
"eval_loss": 0.8805521130561829,
|
| 2325 |
+
"eval_runtime": 902.4134,
|
| 2326 |
+
"eval_samples_per_second": 11.725,
|
| 2327 |
+
"eval_steps_per_second": 2.932,
|
| 2328 |
+
"eval_wer": 97.72748703441772,
|
| 2329 |
+
"step": 8000
|
| 2330 |
+
},
|
| 2331 |
+
{
|
| 2332 |
+
"epoch": 8.838105726872246,
|
| 2333 |
+
"grad_norm": 0.5200393795967102,
|
| 2334 |
+
"learning_rate": 1.2282051282051283e-05,
|
| 2335 |
+
"loss": 0.0886,
|
| 2336 |
+
"step": 8025
|
| 2337 |
+
},
|
| 2338 |
+
{
|
| 2339 |
+
"epoch": 8.865638766519824,
|
| 2340 |
+
"grad_norm": 0.610178530216217,
|
| 2341 |
+
"learning_rate": 1.2256410256410259e-05,
|
| 2342 |
+
"loss": 0.0957,
|
| 2343 |
+
"step": 8050
|
| 2344 |
+
},
|
| 2345 |
+
{
|
| 2346 |
+
"epoch": 8.8931718061674,
|
| 2347 |
+
"grad_norm": 0.4926294684410095,
|
| 2348 |
+
"learning_rate": 1.2230769230769232e-05,
|
| 2349 |
+
"loss": 0.0893,
|
| 2350 |
+
"step": 8075
|
| 2351 |
+
},
|
| 2352 |
+
{
|
| 2353 |
+
"epoch": 8.920704845814978,
|
| 2354 |
+
"grad_norm": 0.3815154731273651,
|
| 2355 |
+
"learning_rate": 1.2205128205128208e-05,
|
| 2356 |
+
"loss": 0.1048,
|
| 2357 |
+
"step": 8100
|
| 2358 |
+
},
|
| 2359 |
+
{
|
| 2360 |
+
"epoch": 8.948237885462555,
|
| 2361 |
+
"grad_norm": 0.44610753655433655,
|
| 2362 |
+
"learning_rate": 1.217948717948718e-05,
|
| 2363 |
+
"loss": 0.0903,
|
| 2364 |
+
"step": 8125
|
| 2365 |
+
},
|
| 2366 |
+
{
|
| 2367 |
+
"epoch": 8.975770925110131,
|
| 2368 |
+
"grad_norm": 0.5451329946517944,
|
| 2369 |
+
"learning_rate": 1.2153846153846153e-05,
|
| 2370 |
+
"loss": 0.0779,
|
| 2371 |
+
"step": 8150
|
| 2372 |
+
},
|
| 2373 |
+
{
|
| 2374 |
+
"epoch": 9.003303964757709,
|
| 2375 |
+
"grad_norm": 0.27041003108024597,
|
| 2376 |
+
"learning_rate": 1.2128205128205129e-05,
|
| 2377 |
+
"loss": 0.0943,
|
| 2378 |
+
"step": 8175
|
| 2379 |
+
},
|
| 2380 |
+
{
|
| 2381 |
+
"epoch": 9.030837004405287,
|
| 2382 |
+
"grad_norm": 0.5706424117088318,
|
| 2383 |
+
"learning_rate": 1.2102564102564102e-05,
|
| 2384 |
+
"loss": 0.0544,
|
| 2385 |
+
"step": 8200
|
| 2386 |
+
},
|
| 2387 |
+
{
|
| 2388 |
+
"epoch": 9.058370044052863,
|
| 2389 |
+
"grad_norm": 0.2460174858570099,
|
| 2390 |
+
"learning_rate": 1.2076923076923078e-05,
|
| 2391 |
+
"loss": 0.0542,
|
| 2392 |
+
"step": 8225
|
| 2393 |
+
},
|
| 2394 |
+
{
|
| 2395 |
+
"epoch": 9.08590308370044,
|
| 2396 |
+
"grad_norm": 0.44156837463378906,
|
| 2397 |
+
"learning_rate": 1.2051282051282051e-05,
|
| 2398 |
+
"loss": 0.048,
|
| 2399 |
+
"step": 8250
|
| 2400 |
+
},
|
| 2401 |
+
{
|
| 2402 |
+
"epoch": 9.113436123348018,
|
| 2403 |
+
"grad_norm": 0.5182222127914429,
|
| 2404 |
+
"learning_rate": 1.2025641025641027e-05,
|
| 2405 |
+
"loss": 0.0551,
|
| 2406 |
+
"step": 8275
|
| 2407 |
+
},
|
| 2408 |
+
{
|
| 2409 |
+
"epoch": 9.140969162995594,
|
| 2410 |
+
"grad_norm": 0.25801339745521545,
|
| 2411 |
+
"learning_rate": 1.2e-05,
|
| 2412 |
+
"loss": 0.0574,
|
| 2413 |
+
"step": 8300
|
| 2414 |
+
},
|
| 2415 |
+
{
|
| 2416 |
+
"epoch": 9.168502202643172,
|
| 2417 |
+
"grad_norm": 0.32541367411613464,
|
| 2418 |
+
"learning_rate": 1.1974358974358976e-05,
|
| 2419 |
+
"loss": 0.0465,
|
| 2420 |
+
"step": 8325
|
| 2421 |
+
},
|
| 2422 |
+
{
|
| 2423 |
+
"epoch": 9.19603524229075,
|
| 2424 |
+
"grad_norm": 0.3748158812522888,
|
| 2425 |
+
"learning_rate": 1.194871794871795e-05,
|
| 2426 |
+
"loss": 0.0628,
|
| 2427 |
+
"step": 8350
|
| 2428 |
+
},
|
| 2429 |
+
{
|
| 2430 |
+
"epoch": 9.223568281938325,
|
| 2431 |
+
"grad_norm": 0.37736037373542786,
|
| 2432 |
+
"learning_rate": 1.1923076923076925e-05,
|
| 2433 |
+
"loss": 0.0563,
|
| 2434 |
+
"step": 8375
|
| 2435 |
+
},
|
| 2436 |
+
{
|
| 2437 |
+
"epoch": 9.251101321585903,
|
| 2438 |
+
"grad_norm": 0.37839052081108093,
|
| 2439 |
+
"learning_rate": 1.1897435897435898e-05,
|
| 2440 |
+
"loss": 0.0535,
|
| 2441 |
+
"step": 8400
|
| 2442 |
+
},
|
| 2443 |
+
{
|
| 2444 |
+
"epoch": 9.27863436123348,
|
| 2445 |
+
"grad_norm": 0.3533216714859009,
|
| 2446 |
+
"learning_rate": 1.1871794871794872e-05,
|
| 2447 |
+
"loss": 0.053,
|
| 2448 |
+
"step": 8425
|
| 2449 |
+
},
|
| 2450 |
+
{
|
| 2451 |
+
"epoch": 9.306167400881057,
|
| 2452 |
+
"grad_norm": 0.3839458227157593,
|
| 2453 |
+
"learning_rate": 1.1846153846153847e-05,
|
| 2454 |
+
"loss": 0.0621,
|
| 2455 |
+
"step": 8450
|
| 2456 |
+
},
|
| 2457 |
+
{
|
| 2458 |
+
"epoch": 9.333700440528634,
|
| 2459 |
+
"grad_norm": 0.4322117567062378,
|
| 2460 |
+
"learning_rate": 1.1820512820512821e-05,
|
| 2461 |
+
"loss": 0.0523,
|
| 2462 |
+
"step": 8475
|
| 2463 |
+
},
|
| 2464 |
+
{
|
| 2465 |
+
"epoch": 9.361233480176212,
|
| 2466 |
+
"grad_norm": 0.29340413212776184,
|
| 2467 |
+
"learning_rate": 1.1794871794871796e-05,
|
| 2468 |
+
"loss": 0.0618,
|
| 2469 |
+
"step": 8500
|
| 2470 |
+
},
|
| 2471 |
+
{
|
| 2472 |
+
"epoch": 9.388766519823788,
|
| 2473 |
+
"grad_norm": 0.1959504932165146,
|
| 2474 |
+
"learning_rate": 1.176923076923077e-05,
|
| 2475 |
+
"loss": 0.0593,
|
| 2476 |
+
"step": 8525
|
| 2477 |
+
},
|
| 2478 |
+
{
|
| 2479 |
+
"epoch": 9.416299559471366,
|
| 2480 |
+
"grad_norm": 0.3742198050022125,
|
| 2481 |
+
"learning_rate": 1.1743589743589745e-05,
|
| 2482 |
+
"loss": 0.0615,
|
| 2483 |
+
"step": 8550
|
| 2484 |
+
},
|
| 2485 |
+
{
|
| 2486 |
+
"epoch": 9.443832599118943,
|
| 2487 |
+
"grad_norm": 0.2696450650691986,
|
| 2488 |
+
"learning_rate": 1.1717948717948719e-05,
|
| 2489 |
+
"loss": 0.0552,
|
| 2490 |
+
"step": 8575
|
| 2491 |
+
},
|
| 2492 |
+
{
|
| 2493 |
+
"epoch": 9.47136563876652,
|
| 2494 |
+
"grad_norm": 0.5510464310646057,
|
| 2495 |
+
"learning_rate": 1.1692307692307694e-05,
|
| 2496 |
+
"loss": 0.0572,
|
| 2497 |
+
"step": 8600
|
| 2498 |
+
},
|
| 2499 |
+
{
|
| 2500 |
+
"epoch": 9.498898678414097,
|
| 2501 |
+
"grad_norm": 0.6056195497512817,
|
| 2502 |
+
"learning_rate": 1.1666666666666668e-05,
|
| 2503 |
+
"loss": 0.0694,
|
| 2504 |
+
"step": 8625
|
| 2505 |
+
},
|
| 2506 |
+
{
|
| 2507 |
+
"epoch": 9.526431718061675,
|
| 2508 |
+
"grad_norm": 0.2911369204521179,
|
| 2509 |
+
"learning_rate": 1.1641025641025643e-05,
|
| 2510 |
+
"loss": 0.0683,
|
| 2511 |
+
"step": 8650
|
| 2512 |
+
},
|
| 2513 |
+
{
|
| 2514 |
+
"epoch": 9.55396475770925,
|
| 2515 |
+
"grad_norm": 0.3952886760234833,
|
| 2516 |
+
"learning_rate": 1.1615384615384617e-05,
|
| 2517 |
+
"loss": 0.0641,
|
| 2518 |
+
"step": 8675
|
| 2519 |
+
},
|
| 2520 |
+
{
|
| 2521 |
+
"epoch": 9.581497797356828,
|
| 2522 |
+
"grad_norm": 0.4399433732032776,
|
| 2523 |
+
"learning_rate": 1.1589743589743592e-05,
|
| 2524 |
+
"loss": 0.0507,
|
| 2525 |
+
"step": 8700
|
| 2526 |
+
},
|
| 2527 |
+
{
|
| 2528 |
+
"epoch": 9.609030837004406,
|
| 2529 |
+
"grad_norm": 0.19196133315563202,
|
| 2530 |
+
"learning_rate": 1.1564102564102566e-05,
|
| 2531 |
+
"loss": 0.0637,
|
| 2532 |
+
"step": 8725
|
| 2533 |
+
},
|
| 2534 |
+
{
|
| 2535 |
+
"epoch": 9.636563876651982,
|
| 2536 |
+
"grad_norm": 0.6704760193824768,
|
| 2537 |
+
"learning_rate": 1.1538461538461538e-05,
|
| 2538 |
+
"loss": 0.0628,
|
| 2539 |
+
"step": 8750
|
| 2540 |
+
},
|
| 2541 |
+
{
|
| 2542 |
+
"epoch": 9.66409691629956,
|
| 2543 |
+
"grad_norm": 0.4588681757450104,
|
| 2544 |
+
"learning_rate": 1.1512820512820513e-05,
|
| 2545 |
+
"loss": 0.0587,
|
| 2546 |
+
"step": 8775
|
| 2547 |
+
},
|
| 2548 |
+
{
|
| 2549 |
+
"epoch": 9.691629955947137,
|
| 2550 |
+
"grad_norm": 0.49800875782966614,
|
| 2551 |
+
"learning_rate": 1.1487179487179487e-05,
|
| 2552 |
+
"loss": 0.0534,
|
| 2553 |
+
"step": 8800
|
| 2554 |
+
},
|
| 2555 |
+
{
|
| 2556 |
+
"epoch": 9.719162995594713,
|
| 2557 |
+
"grad_norm": 0.5300059914588928,
|
| 2558 |
+
"learning_rate": 1.1461538461538462e-05,
|
| 2559 |
+
"loss": 0.0659,
|
| 2560 |
+
"step": 8825
|
| 2561 |
+
},
|
| 2562 |
+
{
|
| 2563 |
+
"epoch": 9.746696035242291,
|
| 2564 |
+
"grad_norm": 0.4262904226779938,
|
| 2565 |
+
"learning_rate": 1.1435897435897436e-05,
|
| 2566 |
+
"loss": 0.0688,
|
| 2567 |
+
"step": 8850
|
| 2568 |
+
},
|
| 2569 |
+
{
|
| 2570 |
+
"epoch": 9.774229074889869,
|
| 2571 |
+
"grad_norm": 0.5263908505439758,
|
| 2572 |
+
"learning_rate": 1.1410256410256411e-05,
|
| 2573 |
+
"loss": 0.068,
|
| 2574 |
+
"step": 8875
|
| 2575 |
+
},
|
| 2576 |
+
{
|
| 2577 |
+
"epoch": 9.801762114537445,
|
| 2578 |
+
"grad_norm": 0.6390700340270996,
|
| 2579 |
+
"learning_rate": 1.1384615384615385e-05,
|
| 2580 |
+
"loss": 0.0595,
|
| 2581 |
+
"step": 8900
|
| 2582 |
+
},
|
| 2583 |
+
{
|
| 2584 |
+
"epoch": 9.829295154185022,
|
| 2585 |
+
"grad_norm": 0.3971773684024811,
|
| 2586 |
+
"learning_rate": 1.135897435897436e-05,
|
| 2587 |
+
"loss": 0.0715,
|
| 2588 |
+
"step": 8925
|
| 2589 |
+
},
|
| 2590 |
+
{
|
| 2591 |
+
"epoch": 9.8568281938326,
|
| 2592 |
+
"grad_norm": 0.2393706738948822,
|
| 2593 |
+
"learning_rate": 1.1333333333333334e-05,
|
| 2594 |
+
"loss": 0.0619,
|
| 2595 |
+
"step": 8950
|
| 2596 |
+
},
|
| 2597 |
+
{
|
| 2598 |
+
"epoch": 9.884361233480176,
|
| 2599 |
+
"grad_norm": 0.7413886785507202,
|
| 2600 |
+
"learning_rate": 1.1307692307692309e-05,
|
| 2601 |
+
"loss": 0.0543,
|
| 2602 |
+
"step": 8975
|
| 2603 |
+
},
|
| 2604 |
+
{
|
| 2605 |
+
"epoch": 9.911894273127754,
|
| 2606 |
+
"grad_norm": 0.4195553660392761,
|
| 2607 |
+
"learning_rate": 1.1282051282051283e-05,
|
| 2608 |
+
"loss": 0.0566,
|
| 2609 |
+
"step": 9000
|
| 2610 |
+
},
|
| 2611 |
+
{
|
| 2612 |
+
"epoch": 9.911894273127754,
|
| 2613 |
+
"eval_cer": 66.49834351678837,
|
| 2614 |
+
"eval_loss": 0.9025006294250488,
|
| 2615 |
+
"eval_runtime": 869.4795,
|
| 2616 |
+
"eval_samples_per_second": 12.169,
|
| 2617 |
+
"eval_steps_per_second": 3.043,
|
| 2618 |
+
"eval_wer": 97.7934936350778,
|
| 2619 |
+
"step": 9000
|
| 2620 |
+
},
|
| 2621 |
+
{
|
| 2622 |
+
"epoch": 9.939427312775331,
|
| 2623 |
+
"grad_norm": 0.26785850524902344,
|
| 2624 |
+
"learning_rate": 1.1256410256410258e-05,
|
| 2625 |
+
"loss": 0.0617,
|
| 2626 |
+
"step": 9025
|
| 2627 |
+
},
|
| 2628 |
+
{
|
| 2629 |
+
"epoch": 9.966960352422907,
|
| 2630 |
+
"grad_norm": 0.7492098808288574,
|
| 2631 |
+
"learning_rate": 1.1230769230769232e-05,
|
| 2632 |
+
"loss": 0.0673,
|
| 2633 |
+
"step": 9050
|
| 2634 |
+
},
|
| 2635 |
+
{
|
| 2636 |
+
"epoch": 9.994493392070485,
|
| 2637 |
+
"grad_norm": 0.4261627495288849,
|
| 2638 |
+
"learning_rate": 1.1205128205128205e-05,
|
| 2639 |
+
"loss": 0.0596,
|
| 2640 |
+
"step": 9075
|
| 2641 |
+
},
|
| 2642 |
+
{
|
| 2643 |
+
"epoch": 10.022026431718063,
|
| 2644 |
+
"grad_norm": 0.271913081407547,
|
| 2645 |
+
"learning_rate": 1.117948717948718e-05,
|
| 2646 |
+
"loss": 0.0387,
|
| 2647 |
+
"step": 9100
|
| 2648 |
+
},
|
| 2649 |
+
{
|
| 2650 |
+
"epoch": 10.049559471365638,
|
| 2651 |
+
"grad_norm": 0.3687051832675934,
|
| 2652 |
+
"learning_rate": 1.1153846153846154e-05,
|
| 2653 |
+
"loss": 0.0355,
|
| 2654 |
+
"step": 9125
|
| 2655 |
+
},
|
| 2656 |
+
{
|
| 2657 |
+
"epoch": 10.077092511013216,
|
| 2658 |
+
"grad_norm": 0.5124948620796204,
|
| 2659 |
+
"learning_rate": 1.112820512820513e-05,
|
| 2660 |
+
"loss": 0.047,
|
| 2661 |
+
"step": 9150
|
| 2662 |
+
},
|
| 2663 |
+
{
|
| 2664 |
+
"epoch": 10.104625550660794,
|
| 2665 |
+
"grad_norm": 0.5072323679924011,
|
| 2666 |
+
"learning_rate": 1.1102564102564103e-05,
|
| 2667 |
+
"loss": 0.0311,
|
| 2668 |
+
"step": 9175
|
| 2669 |
+
},
|
| 2670 |
+
{
|
| 2671 |
+
"epoch": 10.13215859030837,
|
| 2672 |
+
"grad_norm": 0.7538577914237976,
|
| 2673 |
+
"learning_rate": 1.1076923076923079e-05,
|
| 2674 |
+
"loss": 0.0392,
|
| 2675 |
+
"step": 9200
|
| 2676 |
+
},
|
| 2677 |
+
{
|
| 2678 |
+
"epoch": 10.159691629955947,
|
| 2679 |
+
"grad_norm": 0.14610818028450012,
|
| 2680 |
+
"learning_rate": 1.1051282051282052e-05,
|
| 2681 |
+
"loss": 0.0337,
|
| 2682 |
+
"step": 9225
|
| 2683 |
+
},
|
| 2684 |
+
{
|
| 2685 |
+
"epoch": 10.187224669603523,
|
| 2686 |
+
"grad_norm": 0.2696049213409424,
|
| 2687 |
+
"learning_rate": 1.1025641025641028e-05,
|
| 2688 |
+
"loss": 0.0424,
|
| 2689 |
+
"step": 9250
|
| 2690 |
+
},
|
| 2691 |
+
{
|
| 2692 |
+
"epoch": 10.214757709251101,
|
| 2693 |
+
"grad_norm": 0.1662781834602356,
|
| 2694 |
+
"learning_rate": 1.1000000000000001e-05,
|
| 2695 |
+
"loss": 0.0329,
|
| 2696 |
+
"step": 9275
|
| 2697 |
+
},
|
| 2698 |
+
{
|
| 2699 |
+
"epoch": 10.242290748898679,
|
| 2700 |
+
"grad_norm": 0.23444922268390656,
|
| 2701 |
+
"learning_rate": 1.0974358974358977e-05,
|
| 2702 |
+
"loss": 0.0329,
|
| 2703 |
+
"step": 9300
|
| 2704 |
+
},
|
| 2705 |
+
{
|
| 2706 |
+
"epoch": 10.269823788546255,
|
| 2707 |
+
"grad_norm": 0.5237591862678528,
|
| 2708 |
+
"learning_rate": 1.094871794871795e-05,
|
| 2709 |
+
"loss": 0.0403,
|
| 2710 |
+
"step": 9325
|
| 2711 |
+
},
|
| 2712 |
+
{
|
| 2713 |
+
"epoch": 10.297356828193832,
|
| 2714 |
+
"grad_norm": 0.46771717071533203,
|
| 2715 |
+
"learning_rate": 1.0923076923076922e-05,
|
| 2716 |
+
"loss": 0.0396,
|
| 2717 |
+
"step": 9350
|
| 2718 |
+
},
|
| 2719 |
+
{
|
| 2720 |
+
"epoch": 10.32488986784141,
|
| 2721 |
+
"grad_norm": 0.24174697697162628,
|
| 2722 |
+
"learning_rate": 1.0897435897435898e-05,
|
| 2723 |
+
"loss": 0.0483,
|
| 2724 |
+
"step": 9375
|
| 2725 |
+
},
|
| 2726 |
+
{
|
| 2727 |
+
"epoch": 10.352422907488986,
|
| 2728 |
+
"grad_norm": 0.38117629289627075,
|
| 2729 |
+
"learning_rate": 1.0871794871794871e-05,
|
| 2730 |
+
"loss": 0.0337,
|
| 2731 |
+
"step": 9400
|
| 2732 |
+
},
|
| 2733 |
+
{
|
| 2734 |
+
"epoch": 10.379955947136564,
|
| 2735 |
+
"grad_norm": 0.2335430532693863,
|
| 2736 |
+
"learning_rate": 1.0846153846153847e-05,
|
| 2737 |
+
"loss": 0.046,
|
| 2738 |
+
"step": 9425
|
| 2739 |
+
},
|
| 2740 |
+
{
|
| 2741 |
+
"epoch": 10.407488986784141,
|
| 2742 |
+
"grad_norm": 0.1994549036026001,
|
| 2743 |
+
"learning_rate": 1.082051282051282e-05,
|
| 2744 |
+
"loss": 0.0389,
|
| 2745 |
+
"step": 9450
|
| 2746 |
+
},
|
| 2747 |
+
{
|
| 2748 |
+
"epoch": 10.435022026431717,
|
| 2749 |
+
"grad_norm": 0.42172032594680786,
|
| 2750 |
+
"learning_rate": 1.0794871794871796e-05,
|
| 2751 |
+
"loss": 0.0332,
|
| 2752 |
+
"step": 9475
|
| 2753 |
+
},
|
| 2754 |
+
{
|
| 2755 |
+
"epoch": 10.462555066079295,
|
| 2756 |
+
"grad_norm": 0.32804402709007263,
|
| 2757 |
+
"learning_rate": 1.076923076923077e-05,
|
| 2758 |
+
"loss": 0.0356,
|
| 2759 |
+
"step": 9500
|
| 2760 |
+
},
|
| 2761 |
+
{
|
| 2762 |
+
"epoch": 10.490088105726873,
|
| 2763 |
+
"grad_norm": 0.26200738549232483,
|
| 2764 |
+
"learning_rate": 1.0743589743589745e-05,
|
| 2765 |
+
"loss": 0.0495,
|
| 2766 |
+
"step": 9525
|
| 2767 |
+
},
|
| 2768 |
+
{
|
| 2769 |
+
"epoch": 10.517621145374449,
|
| 2770 |
+
"grad_norm": 0.24157360196113586,
|
| 2771 |
+
"learning_rate": 1.0717948717948718e-05,
|
| 2772 |
+
"loss": 0.0364,
|
| 2773 |
+
"step": 9550
|
| 2774 |
+
},
|
| 2775 |
+
{
|
| 2776 |
+
"epoch": 10.545154185022026,
|
| 2777 |
+
"grad_norm": 0.20026101171970367,
|
| 2778 |
+
"learning_rate": 1.0692307692307694e-05,
|
| 2779 |
+
"loss": 0.0374,
|
| 2780 |
+
"step": 9575
|
| 2781 |
+
},
|
| 2782 |
+
{
|
| 2783 |
+
"epoch": 10.572687224669604,
|
| 2784 |
+
"grad_norm": 0.24734953045845032,
|
| 2785 |
+
"learning_rate": 1.0666666666666667e-05,
|
| 2786 |
+
"loss": 0.0373,
|
| 2787 |
+
"step": 9600
|
| 2788 |
+
},
|
| 2789 |
+
{
|
| 2790 |
+
"epoch": 10.60022026431718,
|
| 2791 |
+
"grad_norm": 0.4558570384979248,
|
| 2792 |
+
"learning_rate": 1.0641025641025643e-05,
|
| 2793 |
+
"loss": 0.0399,
|
| 2794 |
+
"step": 9625
|
| 2795 |
+
},
|
| 2796 |
+
{
|
| 2797 |
+
"epoch": 10.627753303964758,
|
| 2798 |
+
"grad_norm": 0.18524861335754395,
|
| 2799 |
+
"learning_rate": 1.0615384615384616e-05,
|
| 2800 |
+
"loss": 0.0386,
|
| 2801 |
+
"step": 9650
|
| 2802 |
+
},
|
| 2803 |
+
{
|
| 2804 |
+
"epoch": 10.655286343612335,
|
| 2805 |
+
"grad_norm": 0.25492531061172485,
|
| 2806 |
+
"learning_rate": 1.058974358974359e-05,
|
| 2807 |
+
"loss": 0.043,
|
| 2808 |
+
"step": 9675
|
| 2809 |
+
},
|
| 2810 |
+
{
|
| 2811 |
+
"epoch": 10.682819383259911,
|
| 2812 |
+
"grad_norm": 0.4922308921813965,
|
| 2813 |
+
"learning_rate": 1.0564102564102565e-05,
|
| 2814 |
+
"loss": 0.0457,
|
| 2815 |
+
"step": 9700
|
| 2816 |
+
},
|
| 2817 |
+
{
|
| 2818 |
+
"epoch": 10.710352422907489,
|
| 2819 |
+
"grad_norm": 0.2943499982357025,
|
| 2820 |
+
"learning_rate": 1.0538461538461539e-05,
|
| 2821 |
+
"loss": 0.0431,
|
| 2822 |
+
"step": 9725
|
| 2823 |
+
},
|
| 2824 |
+
{
|
| 2825 |
+
"epoch": 10.737885462555067,
|
| 2826 |
+
"grad_norm": 0.1940225064754486,
|
| 2827 |
+
"learning_rate": 1.0512820512820514e-05,
|
| 2828 |
+
"loss": 0.0403,
|
| 2829 |
+
"step": 9750
|
| 2830 |
+
},
|
| 2831 |
+
{
|
| 2832 |
+
"epoch": 10.765418502202643,
|
| 2833 |
+
"grad_norm": 0.1552239954471588,
|
| 2834 |
+
"learning_rate": 1.0487179487179488e-05,
|
| 2835 |
+
"loss": 0.0367,
|
| 2836 |
+
"step": 9775
|
| 2837 |
+
},
|
| 2838 |
+
{
|
| 2839 |
+
"epoch": 10.79295154185022,
|
| 2840 |
+
"grad_norm": 0.49517762660980225,
|
| 2841 |
+
"learning_rate": 1.0461538461538463e-05,
|
| 2842 |
+
"loss": 0.0425,
|
| 2843 |
+
"step": 9800
|
| 2844 |
+
},
|
| 2845 |
+
{
|
| 2846 |
+
"epoch": 10.820484581497798,
|
| 2847 |
+
"grad_norm": 0.611227810382843,
|
| 2848 |
+
"learning_rate": 1.0435897435897437e-05,
|
| 2849 |
+
"loss": 0.0385,
|
| 2850 |
+
"step": 9825
|
| 2851 |
+
},
|
| 2852 |
+
{
|
| 2853 |
+
"epoch": 10.848017621145374,
|
| 2854 |
+
"grad_norm": 0.19623863697052002,
|
| 2855 |
+
"learning_rate": 1.0410256410256412e-05,
|
| 2856 |
+
"loss": 0.0364,
|
| 2857 |
+
"step": 9850
|
| 2858 |
+
},
|
| 2859 |
+
{
|
| 2860 |
+
"epoch": 10.875550660792952,
|
| 2861 |
+
"grad_norm": 0.4200350344181061,
|
| 2862 |
+
"learning_rate": 1.0384615384615386e-05,
|
| 2863 |
+
"loss": 0.0408,
|
| 2864 |
+
"step": 9875
|
| 2865 |
+
},
|
| 2866 |
+
{
|
| 2867 |
+
"epoch": 10.90308370044053,
|
| 2868 |
+
"grad_norm": 0.5670203566551208,
|
| 2869 |
+
"learning_rate": 1.0358974358974361e-05,
|
| 2870 |
+
"loss": 0.0333,
|
| 2871 |
+
"step": 9900
|
| 2872 |
+
},
|
| 2873 |
+
{
|
| 2874 |
+
"epoch": 10.930616740088105,
|
| 2875 |
+
"grad_norm": 0.4078093469142914,
|
| 2876 |
+
"learning_rate": 1.0333333333333335e-05,
|
| 2877 |
+
"loss": 0.0319,
|
| 2878 |
+
"step": 9925
|
| 2879 |
+
},
|
| 2880 |
+
{
|
| 2881 |
+
"epoch": 10.958149779735683,
|
| 2882 |
+
"grad_norm": 0.24493761360645294,
|
| 2883 |
+
"learning_rate": 1.0307692307692307e-05,
|
| 2884 |
+
"loss": 0.0414,
|
| 2885 |
+
"step": 9950
|
| 2886 |
+
},
|
| 2887 |
+
{
|
| 2888 |
+
"epoch": 10.98568281938326,
|
| 2889 |
+
"grad_norm": 0.39487889409065247,
|
| 2890 |
+
"learning_rate": 1.0282051282051282e-05,
|
| 2891 |
+
"loss": 0.046,
|
| 2892 |
+
"step": 9975
|
| 2893 |
+
},
|
| 2894 |
+
{
|
| 2895 |
+
"epoch": 11.013215859030836,
|
| 2896 |
+
"grad_norm": 0.09149141609668732,
|
| 2897 |
+
"learning_rate": 1.0256410256410256e-05,
|
| 2898 |
+
"loss": 0.037,
|
| 2899 |
+
"step": 10000
|
| 2900 |
+
},
|
| 2901 |
+
{
|
| 2902 |
+
"epoch": 11.013215859030836,
|
| 2903 |
+
"eval_cer": 63.11544942854195,
|
| 2904 |
+
"eval_loss": 0.9283918738365173,
|
| 2905 |
+
"eval_runtime": 860.3587,
|
| 2906 |
+
"eval_samples_per_second": 12.298,
|
| 2907 |
+
"eval_steps_per_second": 3.075,
|
| 2908 |
+
"eval_wer": 97.53889674681754,
|
| 2909 |
+
"step": 10000
|
| 2910 |
+
},
|
| 2911 |
+
{
|
| 2912 |
+
"epoch": 11.040748898678414,
|
| 2913 |
+
"grad_norm": 0.4592762887477875,
|
| 2914 |
+
"learning_rate": 1.0230769230769231e-05,
|
| 2915 |
+
"loss": 0.0276,
|
| 2916 |
+
"step": 10025
|
| 2917 |
+
},
|
| 2918 |
+
{
|
| 2919 |
+
"epoch": 11.068281938325992,
|
| 2920 |
+
"grad_norm": 0.15113013982772827,
|
| 2921 |
+
"learning_rate": 1.0205128205128205e-05,
|
| 2922 |
+
"loss": 0.0259,
|
| 2923 |
+
"step": 10050
|
| 2924 |
+
},
|
| 2925 |
+
{
|
| 2926 |
+
"epoch": 11.095814977973568,
|
| 2927 |
+
"grad_norm": 0.1503317505121231,
|
| 2928 |
+
"learning_rate": 1.017948717948718e-05,
|
| 2929 |
+
"loss": 0.0276,
|
| 2930 |
+
"step": 10075
|
| 2931 |
+
},
|
| 2932 |
+
{
|
| 2933 |
+
"epoch": 11.123348017621145,
|
| 2934 |
+
"grad_norm": 0.14569109678268433,
|
| 2935 |
+
"learning_rate": 1.0153846153846154e-05,
|
| 2936 |
+
"loss": 0.0238,
|
| 2937 |
+
"step": 10100
|
| 2938 |
+
},
|
| 2939 |
+
{
|
| 2940 |
+
"epoch": 11.150881057268723,
|
| 2941 |
+
"grad_norm": 0.37791919708251953,
|
| 2942 |
+
"learning_rate": 1.012820512820513e-05,
|
| 2943 |
+
"loss": 0.022,
|
| 2944 |
+
"step": 10125
|
| 2945 |
+
},
|
| 2946 |
+
{
|
| 2947 |
+
"epoch": 11.178414096916299,
|
| 2948 |
+
"grad_norm": 0.2850872874259949,
|
| 2949 |
+
"learning_rate": 1.0102564102564103e-05,
|
| 2950 |
+
"loss": 0.0245,
|
| 2951 |
+
"step": 10150
|
| 2952 |
+
},
|
| 2953 |
+
{
|
| 2954 |
+
"epoch": 11.205947136563877,
|
| 2955 |
+
"grad_norm": 0.3191249370574951,
|
| 2956 |
+
"learning_rate": 1.0076923076923078e-05,
|
| 2957 |
+
"loss": 0.0247,
|
| 2958 |
+
"step": 10175
|
| 2959 |
+
},
|
| 2960 |
+
{
|
| 2961 |
+
"epoch": 11.233480176211454,
|
| 2962 |
+
"grad_norm": 0.23166881501674652,
|
| 2963 |
+
"learning_rate": 1.0051282051282052e-05,
|
| 2964 |
+
"loss": 0.0247,
|
| 2965 |
+
"step": 10200
|
| 2966 |
+
},
|
| 2967 |
+
{
|
| 2968 |
+
"epoch": 11.26101321585903,
|
| 2969 |
+
"grad_norm": 0.24626369774341583,
|
| 2970 |
+
"learning_rate": 1.0025641025641027e-05,
|
| 2971 |
+
"loss": 0.0246,
|
| 2972 |
+
"step": 10225
|
| 2973 |
+
},
|
| 2974 |
+
{
|
| 2975 |
+
"epoch": 11.288546255506608,
|
| 2976 |
+
"grad_norm": 0.2534284293651581,
|
| 2977 |
+
"learning_rate": 1e-05,
|
| 2978 |
+
"loss": 0.0219,
|
| 2979 |
+
"step": 10250
|
| 2980 |
+
},
|
| 2981 |
+
{
|
| 2982 |
+
"epoch": 11.316079295154186,
|
| 2983 |
+
"grad_norm": 0.170914426445961,
|
| 2984 |
+
"learning_rate": 9.974358974358974e-06,
|
| 2985 |
+
"loss": 0.0263,
|
| 2986 |
+
"step": 10275
|
| 2987 |
+
},
|
| 2988 |
+
{
|
| 2989 |
+
"epoch": 11.343612334801762,
|
| 2990 |
+
"grad_norm": 0.20018179714679718,
|
| 2991 |
+
"learning_rate": 9.94871794871795e-06,
|
| 2992 |
+
"loss": 0.0258,
|
| 2993 |
+
"step": 10300
|
| 2994 |
+
},
|
| 2995 |
+
{
|
| 2996 |
+
"epoch": 11.37114537444934,
|
| 2997 |
+
"grad_norm": 0.25566136837005615,
|
| 2998 |
+
"learning_rate": 9.923076923076923e-06,
|
| 2999 |
+
"loss": 0.0229,
|
| 3000 |
+
"step": 10325
|
| 3001 |
+
},
|
| 3002 |
+
{
|
| 3003 |
+
"epoch": 11.398678414096917,
|
| 3004 |
+
"grad_norm": 0.16708512604236603,
|
| 3005 |
+
"learning_rate": 9.897435897435899e-06,
|
| 3006 |
+
"loss": 0.0288,
|
| 3007 |
+
"step": 10350
|
| 3008 |
+
},
|
| 3009 |
+
{
|
| 3010 |
+
"epoch": 11.426211453744493,
|
| 3011 |
+
"grad_norm": 0.1267959177494049,
|
| 3012 |
+
"learning_rate": 9.871794871794872e-06,
|
| 3013 |
+
"loss": 0.0227,
|
| 3014 |
+
"step": 10375
|
| 3015 |
+
},
|
| 3016 |
+
{
|
| 3017 |
+
"epoch": 11.45374449339207,
|
| 3018 |
+
"grad_norm": 0.14873263239860535,
|
| 3019 |
+
"learning_rate": 9.846153846153848e-06,
|
| 3020 |
+
"loss": 0.0293,
|
| 3021 |
+
"step": 10400
|
| 3022 |
+
},
|
| 3023 |
+
{
|
| 3024 |
+
"epoch": 11.481277533039648,
|
| 3025 |
+
"grad_norm": 0.4109511077404022,
|
| 3026 |
+
"learning_rate": 9.820512820512821e-06,
|
| 3027 |
+
"loss": 0.0254,
|
| 3028 |
+
"step": 10425
|
| 3029 |
+
},
|
| 3030 |
+
{
|
| 3031 |
+
"epoch": 11.508810572687224,
|
| 3032 |
+
"grad_norm": 0.16823522746562958,
|
| 3033 |
+
"learning_rate": 9.794871794871795e-06,
|
| 3034 |
+
"loss": 0.0307,
|
| 3035 |
+
"step": 10450
|
| 3036 |
+
},
|
| 3037 |
+
{
|
| 3038 |
+
"epoch": 11.536343612334802,
|
| 3039 |
+
"grad_norm": 0.24721799790859222,
|
| 3040 |
+
"learning_rate": 9.76923076923077e-06,
|
| 3041 |
+
"loss": 0.0226,
|
| 3042 |
+
"step": 10475
|
| 3043 |
+
},
|
| 3044 |
+
{
|
| 3045 |
+
"epoch": 11.56387665198238,
|
| 3046 |
+
"grad_norm": 0.1115945503115654,
|
| 3047 |
+
"learning_rate": 9.743589743589744e-06,
|
| 3048 |
+
"loss": 0.0243,
|
| 3049 |
+
"step": 10500
|
| 3050 |
+
},
|
| 3051 |
+
{
|
| 3052 |
+
"epoch": 11.591409691629956,
|
| 3053 |
+
"grad_norm": 0.1863705813884735,
|
| 3054 |
+
"learning_rate": 9.71794871794872e-06,
|
| 3055 |
+
"loss": 0.0283,
|
| 3056 |
+
"step": 10525
|
| 3057 |
+
},
|
| 3058 |
+
{
|
| 3059 |
+
"epoch": 11.618942731277533,
|
| 3060 |
+
"grad_norm": 0.1940588504076004,
|
| 3061 |
+
"learning_rate": 9.692307692307693e-06,
|
| 3062 |
+
"loss": 0.0243,
|
| 3063 |
+
"step": 10550
|
| 3064 |
+
},
|
| 3065 |
+
{
|
| 3066 |
+
"epoch": 11.646475770925111,
|
| 3067 |
+
"grad_norm": 0.4765683114528656,
|
| 3068 |
+
"learning_rate": 9.666666666666667e-06,
|
| 3069 |
+
"loss": 0.0286,
|
| 3070 |
+
"step": 10575
|
| 3071 |
+
},
|
| 3072 |
+
{
|
| 3073 |
+
"epoch": 11.674008810572687,
|
| 3074 |
+
"grad_norm": 0.22577764093875885,
|
| 3075 |
+
"learning_rate": 9.641025641025642e-06,
|
| 3076 |
+
"loss": 0.0251,
|
| 3077 |
+
"step": 10600
|
| 3078 |
+
},
|
| 3079 |
+
{
|
| 3080 |
+
"epoch": 11.701541850220265,
|
| 3081 |
+
"grad_norm": 0.351157009601593,
|
| 3082 |
+
"learning_rate": 9.615384615384616e-06,
|
| 3083 |
+
"loss": 0.0268,
|
| 3084 |
+
"step": 10625
|
| 3085 |
+
},
|
| 3086 |
+
{
|
| 3087 |
+
"epoch": 11.729074889867842,
|
| 3088 |
+
"grad_norm": 0.10466675460338593,
|
| 3089 |
+
"learning_rate": 9.589743589743591e-06,
|
| 3090 |
+
"loss": 0.0248,
|
| 3091 |
+
"step": 10650
|
| 3092 |
+
},
|
| 3093 |
+
{
|
| 3094 |
+
"epoch": 11.756607929515418,
|
| 3095 |
+
"grad_norm": 0.14528048038482666,
|
| 3096 |
+
"learning_rate": 9.564102564102565e-06,
|
| 3097 |
+
"loss": 0.0243,
|
| 3098 |
+
"step": 10675
|
| 3099 |
+
},
|
| 3100 |
+
{
|
| 3101 |
+
"epoch": 11.784140969162996,
|
| 3102 |
+
"grad_norm": 0.48743611574172974,
|
| 3103 |
+
"learning_rate": 9.53846153846154e-06,
|
| 3104 |
+
"loss": 0.0273,
|
| 3105 |
+
"step": 10700
|
| 3106 |
+
},
|
| 3107 |
+
{
|
| 3108 |
+
"epoch": 11.811674008810574,
|
| 3109 |
+
"grad_norm": 0.1229957789182663,
|
| 3110 |
+
"learning_rate": 9.512820512820514e-06,
|
| 3111 |
+
"loss": 0.0227,
|
| 3112 |
+
"step": 10725
|
| 3113 |
+
},
|
| 3114 |
+
{
|
| 3115 |
+
"epoch": 11.83920704845815,
|
| 3116 |
+
"grad_norm": 0.13707463443279266,
|
| 3117 |
+
"learning_rate": 9.487179487179487e-06,
|
| 3118 |
+
"loss": 0.028,
|
| 3119 |
+
"step": 10750
|
| 3120 |
+
},
|
| 3121 |
+
{
|
| 3122 |
+
"epoch": 11.866740088105727,
|
| 3123 |
+
"grad_norm": 0.1293453723192215,
|
| 3124 |
+
"learning_rate": 9.461538461538463e-06,
|
| 3125 |
+
"loss": 0.0203,
|
| 3126 |
+
"step": 10775
|
| 3127 |
+
},
|
| 3128 |
+
{
|
| 3129 |
+
"epoch": 11.894273127753303,
|
| 3130 |
+
"grad_norm": 0.30864667892456055,
|
| 3131 |
+
"learning_rate": 9.435897435897436e-06,
|
| 3132 |
+
"loss": 0.0271,
|
| 3133 |
+
"step": 10800
|
| 3134 |
+
},
|
| 3135 |
+
{
|
| 3136 |
+
"epoch": 11.92180616740088,
|
| 3137 |
+
"grad_norm": 0.19544030725955963,
|
| 3138 |
+
"learning_rate": 9.410256410256412e-06,
|
| 3139 |
+
"loss": 0.0235,
|
| 3140 |
+
"step": 10825
|
| 3141 |
+
},
|
| 3142 |
+
{
|
| 3143 |
+
"epoch": 11.949339207048459,
|
| 3144 |
+
"grad_norm": 0.24226927757263184,
|
| 3145 |
+
"learning_rate": 9.384615384615385e-06,
|
| 3146 |
+
"loss": 0.0275,
|
| 3147 |
+
"step": 10850
|
| 3148 |
+
},
|
| 3149 |
+
{
|
| 3150 |
+
"epoch": 11.976872246696034,
|
| 3151 |
+
"grad_norm": 0.31063777208328247,
|
| 3152 |
+
"learning_rate": 9.358974358974359e-06,
|
| 3153 |
+
"loss": 0.0336,
|
| 3154 |
+
"step": 10875
|
| 3155 |
+
},
|
| 3156 |
+
{
|
| 3157 |
+
"epoch": 12.004405286343612,
|
| 3158 |
+
"grad_norm": 0.08906027674674988,
|
| 3159 |
+
"learning_rate": 9.333333333333334e-06,
|
| 3160 |
+
"loss": 0.03,
|
| 3161 |
+
"step": 10900
|
| 3162 |
+
},
|
| 3163 |
+
{
|
| 3164 |
+
"epoch": 12.03193832599119,
|
| 3165 |
+
"grad_norm": 0.17722788453102112,
|
| 3166 |
+
"learning_rate": 9.307692307692308e-06,
|
| 3167 |
+
"loss": 0.016,
|
| 3168 |
+
"step": 10925
|
| 3169 |
+
},
|
| 3170 |
+
{
|
| 3171 |
+
"epoch": 12.059471365638766,
|
| 3172 |
+
"grad_norm": 0.1557055562734604,
|
| 3173 |
+
"learning_rate": 9.282051282051283e-06,
|
| 3174 |
+
"loss": 0.0164,
|
| 3175 |
+
"step": 10950
|
| 3176 |
+
},
|
| 3177 |
+
{
|
| 3178 |
+
"epoch": 12.087004405286343,
|
| 3179 |
+
"grad_norm": 0.1017475500702858,
|
| 3180 |
+
"learning_rate": 9.256410256410257e-06,
|
| 3181 |
+
"loss": 0.015,
|
| 3182 |
+
"step": 10975
|
| 3183 |
+
},
|
| 3184 |
+
{
|
| 3185 |
+
"epoch": 12.114537444933921,
|
| 3186 |
+
"grad_norm": 0.1605156809091568,
|
| 3187 |
+
"learning_rate": 9.230769230769232e-06,
|
| 3188 |
+
"loss": 0.0139,
|
| 3189 |
+
"step": 11000
|
| 3190 |
+
},
|
| 3191 |
+
{
|
| 3192 |
+
"epoch": 12.114537444933921,
|
| 3193 |
+
"eval_cer": 60.90584985040065,
|
| 3194 |
+
"eval_loss": 0.9458209872245789,
|
| 3195 |
+
"eval_runtime": 832.2415,
|
| 3196 |
+
"eval_samples_per_second": 12.714,
|
| 3197 |
+
"eval_steps_per_second": 3.179,
|
| 3198 |
+
"eval_wer": 97.02970297029702,
|
| 3199 |
+
"step": 11000
|
| 3200 |
+
},
|
| 3201 |
+
{
|
| 3202 |
+
"epoch": 12.142070484581497,
|
| 3203 |
+
"grad_norm": 0.24084699153900146,
|
| 3204 |
+
"learning_rate": 9.205128205128206e-06,
|
| 3205 |
+
"loss": 0.0161,
|
| 3206 |
+
"step": 11025
|
| 3207 |
+
},
|
| 3208 |
+
{
|
| 3209 |
+
"epoch": 12.169603524229075,
|
| 3210 |
+
"grad_norm": 0.10124333202838898,
|
| 3211 |
+
"learning_rate": 9.17948717948718e-06,
|
| 3212 |
+
"loss": 0.0145,
|
| 3213 |
+
"step": 11050
|
| 3214 |
+
},
|
| 3215 |
+
{
|
| 3216 |
+
"epoch": 12.197136563876652,
|
| 3217 |
+
"grad_norm": 0.6229532957077026,
|
| 3218 |
+
"learning_rate": 9.153846153846155e-06,
|
| 3219 |
+
"loss": 0.0203,
|
| 3220 |
+
"step": 11075
|
| 3221 |
+
},
|
| 3222 |
+
{
|
| 3223 |
+
"epoch": 12.224669603524228,
|
| 3224 |
+
"grad_norm": 0.13394711911678314,
|
| 3225 |
+
"learning_rate": 9.128205128205129e-06,
|
| 3226 |
+
"loss": 0.0184,
|
| 3227 |
+
"step": 11100
|
| 3228 |
+
},
|
| 3229 |
+
{
|
| 3230 |
+
"epoch": 12.252202643171806,
|
| 3231 |
+
"grad_norm": 0.200868159532547,
|
| 3232 |
+
"learning_rate": 9.102564102564104e-06,
|
| 3233 |
+
"loss": 0.0176,
|
| 3234 |
+
"step": 11125
|
| 3235 |
+
},
|
| 3236 |
+
{
|
| 3237 |
+
"epoch": 12.279735682819384,
|
| 3238 |
+
"grad_norm": 0.2978365123271942,
|
| 3239 |
+
"learning_rate": 9.076923076923078e-06,
|
| 3240 |
+
"loss": 0.017,
|
| 3241 |
+
"step": 11150
|
| 3242 |
+
},
|
| 3243 |
+
{
|
| 3244 |
+
"epoch": 12.30726872246696,
|
| 3245 |
+
"grad_norm": 0.07799817621707916,
|
| 3246 |
+
"learning_rate": 9.051282051282051e-06,
|
| 3247 |
+
"loss": 0.017,
|
| 3248 |
+
"step": 11175
|
| 3249 |
+
},
|
| 3250 |
+
{
|
| 3251 |
+
"epoch": 12.334801762114537,
|
| 3252 |
+
"grad_norm": 0.20449844002723694,
|
| 3253 |
+
"learning_rate": 9.025641025641027e-06,
|
| 3254 |
+
"loss": 0.0174,
|
| 3255 |
+
"step": 11200
|
| 3256 |
+
},
|
| 3257 |
+
{
|
| 3258 |
+
"epoch": 12.362334801762115,
|
| 3259 |
+
"grad_norm": 0.21774497628211975,
|
| 3260 |
+
"learning_rate": 9e-06,
|
| 3261 |
+
"loss": 0.0147,
|
| 3262 |
+
"step": 11225
|
| 3263 |
+
},
|
| 3264 |
+
{
|
| 3265 |
+
"epoch": 12.389867841409691,
|
| 3266 |
+
"grad_norm": 0.2724749445915222,
|
| 3267 |
+
"learning_rate": 8.974358974358976e-06,
|
| 3268 |
+
"loss": 0.0161,
|
| 3269 |
+
"step": 11250
|
| 3270 |
+
},
|
| 3271 |
+
{
|
| 3272 |
+
"epoch": 12.417400881057269,
|
| 3273 |
+
"grad_norm": 0.09058816730976105,
|
| 3274 |
+
"learning_rate": 8.94871794871795e-06,
|
| 3275 |
+
"loss": 0.0161,
|
| 3276 |
+
"step": 11275
|
| 3277 |
+
},
|
| 3278 |
+
{
|
| 3279 |
+
"epoch": 12.444933920704846,
|
| 3280 |
+
"grad_norm": 0.15494456887245178,
|
| 3281 |
+
"learning_rate": 8.923076923076925e-06,
|
| 3282 |
+
"loss": 0.0183,
|
| 3283 |
+
"step": 11300
|
| 3284 |
+
},
|
| 3285 |
+
{
|
| 3286 |
+
"epoch": 12.472466960352422,
|
| 3287 |
+
"grad_norm": 0.16612644493579865,
|
| 3288 |
+
"learning_rate": 8.897435897435898e-06,
|
| 3289 |
+
"loss": 0.0164,
|
| 3290 |
+
"step": 11325
|
| 3291 |
+
},
|
| 3292 |
+
{
|
| 3293 |
+
"epoch": 12.5,
|
| 3294 |
+
"grad_norm": 0.20393668115139008,
|
| 3295 |
+
"learning_rate": 8.871794871794872e-06,
|
| 3296 |
+
"loss": 0.0193,
|
| 3297 |
+
"step": 11350
|
| 3298 |
+
},
|
| 3299 |
+
{
|
| 3300 |
+
"epoch": 12.527533039647578,
|
| 3301 |
+
"grad_norm": 0.1749681979417801,
|
| 3302 |
+
"learning_rate": 8.846153846153847e-06,
|
| 3303 |
+
"loss": 0.0196,
|
| 3304 |
+
"step": 11375
|
| 3305 |
+
},
|
| 3306 |
+
{
|
| 3307 |
+
"epoch": 12.555066079295154,
|
| 3308 |
+
"grad_norm": 0.1079695075750351,
|
| 3309 |
+
"learning_rate": 8.820512820512821e-06,
|
| 3310 |
+
"loss": 0.0203,
|
| 3311 |
+
"step": 11400
|
| 3312 |
+
},
|
| 3313 |
+
{
|
| 3314 |
+
"epoch": 12.582599118942731,
|
| 3315 |
+
"grad_norm": 0.2510460317134857,
|
| 3316 |
+
"learning_rate": 8.794871794871796e-06,
|
| 3317 |
+
"loss": 0.0174,
|
| 3318 |
+
"step": 11425
|
| 3319 |
+
},
|
| 3320 |
+
{
|
| 3321 |
+
"epoch": 12.610132158590309,
|
| 3322 |
+
"grad_norm": 0.11694565415382385,
|
| 3323 |
+
"learning_rate": 8.76923076923077e-06,
|
| 3324 |
+
"loss": 0.0163,
|
| 3325 |
+
"step": 11450
|
| 3326 |
+
},
|
| 3327 |
+
{
|
| 3328 |
+
"epoch": 12.637665198237885,
|
| 3329 |
+
"grad_norm": 0.37202000617980957,
|
| 3330 |
+
"learning_rate": 8.743589743589743e-06,
|
| 3331 |
+
"loss": 0.0181,
|
| 3332 |
+
"step": 11475
|
| 3333 |
+
},
|
| 3334 |
+
{
|
| 3335 |
+
"epoch": 12.665198237885463,
|
| 3336 |
+
"grad_norm": 0.16289299726486206,
|
| 3337 |
+
"learning_rate": 8.717948717948719e-06,
|
| 3338 |
+
"loss": 0.0184,
|
| 3339 |
+
"step": 11500
|
| 3340 |
+
},
|
| 3341 |
+
{
|
| 3342 |
+
"epoch": 12.69273127753304,
|
| 3343 |
+
"grad_norm": 0.22636614739894867,
|
| 3344 |
+
"learning_rate": 8.692307692307692e-06,
|
| 3345 |
+
"loss": 0.0183,
|
| 3346 |
+
"step": 11525
|
| 3347 |
+
},
|
| 3348 |
+
{
|
| 3349 |
+
"epoch": 12.720264317180616,
|
| 3350 |
+
"grad_norm": 0.1628304421901703,
|
| 3351 |
+
"learning_rate": 8.666666666666668e-06,
|
| 3352 |
+
"loss": 0.0195,
|
| 3353 |
+
"step": 11550
|
| 3354 |
+
},
|
| 3355 |
+
{
|
| 3356 |
+
"epoch": 12.747797356828194,
|
| 3357 |
+
"grad_norm": 0.1048491895198822,
|
| 3358 |
+
"learning_rate": 8.641025641025641e-06,
|
| 3359 |
+
"loss": 0.0154,
|
| 3360 |
+
"step": 11575
|
| 3361 |
+
},
|
| 3362 |
+
{
|
| 3363 |
+
"epoch": 12.775330396475772,
|
| 3364 |
+
"grad_norm": 0.4491131007671356,
|
| 3365 |
+
"learning_rate": 8.615384615384617e-06,
|
| 3366 |
+
"loss": 0.0179,
|
| 3367 |
+
"step": 11600
|
| 3368 |
+
},
|
| 3369 |
+
{
|
| 3370 |
+
"epoch": 12.802863436123348,
|
| 3371 |
+
"grad_norm": 0.28522998094558716,
|
| 3372 |
+
"learning_rate": 8.58974358974359e-06,
|
| 3373 |
+
"loss": 0.0189,
|
| 3374 |
+
"step": 11625
|
| 3375 |
+
},
|
| 3376 |
+
{
|
| 3377 |
+
"epoch": 12.830396475770925,
|
| 3378 |
+
"grad_norm": 0.24292542040348053,
|
| 3379 |
+
"learning_rate": 8.564102564102564e-06,
|
| 3380 |
+
"loss": 0.0165,
|
| 3381 |
+
"step": 11650
|
| 3382 |
+
},
|
| 3383 |
+
{
|
| 3384 |
+
"epoch": 12.857929515418503,
|
| 3385 |
+
"grad_norm": 0.44718441367149353,
|
| 3386 |
+
"learning_rate": 8.53846153846154e-06,
|
| 3387 |
+
"loss": 0.0182,
|
| 3388 |
+
"step": 11675
|
| 3389 |
+
},
|
| 3390 |
+
{
|
| 3391 |
+
"epoch": 12.885462555066079,
|
| 3392 |
+
"grad_norm": 0.1012931615114212,
|
| 3393 |
+
"learning_rate": 8.512820512820513e-06,
|
| 3394 |
+
"loss": 0.0183,
|
| 3395 |
+
"step": 11700
|
| 3396 |
+
},
|
| 3397 |
+
{
|
| 3398 |
+
"epoch": 12.912995594713657,
|
| 3399 |
+
"grad_norm": 0.14866550266742706,
|
| 3400 |
+
"learning_rate": 8.487179487179488e-06,
|
| 3401 |
+
"loss": 0.0191,
|
| 3402 |
+
"step": 11725
|
| 3403 |
+
},
|
| 3404 |
+
{
|
| 3405 |
+
"epoch": 12.940528634361234,
|
| 3406 |
+
"grad_norm": 0.2306414097547531,
|
| 3407 |
+
"learning_rate": 8.461538461538462e-06,
|
| 3408 |
+
"loss": 0.0202,
|
| 3409 |
+
"step": 11750
|
| 3410 |
+
},
|
| 3411 |
+
{
|
| 3412 |
+
"epoch": 12.96806167400881,
|
| 3413 |
+
"grad_norm": 0.2795422077178955,
|
| 3414 |
+
"learning_rate": 8.435897435897436e-06,
|
| 3415 |
+
"loss": 0.0209,
|
| 3416 |
+
"step": 11775
|
| 3417 |
+
},
|
| 3418 |
+
{
|
| 3419 |
+
"epoch": 12.995594713656388,
|
| 3420 |
+
"grad_norm": 0.22507907450199127,
|
| 3421 |
+
"learning_rate": 8.410256410256411e-06,
|
| 3422 |
+
"loss": 0.021,
|
| 3423 |
+
"step": 11800
|
| 3424 |
+
},
|
| 3425 |
+
{
|
| 3426 |
+
"epoch": 13.023127753303966,
|
| 3427 |
+
"grad_norm": 0.10657211393117905,
|
| 3428 |
+
"learning_rate": 8.384615384615385e-06,
|
| 3429 |
+
"loss": 0.0133,
|
| 3430 |
+
"step": 11825
|
| 3431 |
+
},
|
| 3432 |
+
{
|
| 3433 |
+
"epoch": 13.050660792951541,
|
| 3434 |
+
"grad_norm": 0.3987150490283966,
|
| 3435 |
+
"learning_rate": 8.35897435897436e-06,
|
| 3436 |
+
"loss": 0.0123,
|
| 3437 |
+
"step": 11850
|
| 3438 |
+
},
|
| 3439 |
+
{
|
| 3440 |
+
"epoch": 13.07819383259912,
|
| 3441 |
+
"grad_norm": 0.16128714382648468,
|
| 3442 |
+
"learning_rate": 8.333333333333334e-06,
|
| 3443 |
+
"loss": 0.0122,
|
| 3444 |
+
"step": 11875
|
| 3445 |
+
},
|
| 3446 |
+
{
|
| 3447 |
+
"epoch": 13.105726872246697,
|
| 3448 |
+
"grad_norm": 0.13785377144813538,
|
| 3449 |
+
"learning_rate": 8.307692307692309e-06,
|
| 3450 |
+
"loss": 0.0101,
|
| 3451 |
+
"step": 11900
|
| 3452 |
+
},
|
| 3453 |
+
{
|
| 3454 |
+
"epoch": 13.133259911894273,
|
| 3455 |
+
"grad_norm": 0.180278480052948,
|
| 3456 |
+
"learning_rate": 8.282051282051283e-06,
|
| 3457 |
+
"loss": 0.0126,
|
| 3458 |
+
"step": 11925
|
| 3459 |
+
},
|
| 3460 |
+
{
|
| 3461 |
+
"epoch": 13.16079295154185,
|
| 3462 |
+
"grad_norm": 0.06596548855304718,
|
| 3463 |
+
"learning_rate": 8.256410256410256e-06,
|
| 3464 |
+
"loss": 0.0105,
|
| 3465 |
+
"step": 11950
|
| 3466 |
+
},
|
| 3467 |
+
{
|
| 3468 |
+
"epoch": 13.188325991189428,
|
| 3469 |
+
"grad_norm": 0.17440514266490936,
|
| 3470 |
+
"learning_rate": 8.230769230769232e-06,
|
| 3471 |
+
"loss": 0.0109,
|
| 3472 |
+
"step": 11975
|
| 3473 |
+
},
|
| 3474 |
+
{
|
| 3475 |
+
"epoch": 13.215859030837004,
|
| 3476 |
+
"grad_norm": 0.16661639511585236,
|
| 3477 |
+
"learning_rate": 8.205128205128205e-06,
|
| 3478 |
+
"loss": 0.013,
|
| 3479 |
+
"step": 12000
|
| 3480 |
+
},
|
| 3481 |
+
{
|
| 3482 |
+
"epoch": 13.215859030837004,
|
| 3483 |
+
"eval_cer": 57.880618573246366,
|
| 3484 |
+
"eval_loss": 0.9624491930007935,
|
| 3485 |
+
"eval_runtime": 831.8616,
|
| 3486 |
+
"eval_samples_per_second": 12.72,
|
| 3487 |
+
"eval_steps_per_second": 3.181,
|
| 3488 |
+
"eval_wer": 96.82225365393681,
|
| 3489 |
+
"step": 12000
|
| 3490 |
}
|
| 3491 |
],
|
| 3492 |
"logging_steps": 25,
|
|
|
|
| 3506 |
"attributes": {}
|
| 3507 |
}
|
| 3508 |
},
|
| 3509 |
+
"total_flos": 2.025804888042111e+19,
|
| 3510 |
"train_batch_size": 4,
|
| 3511 |
"trial_name": null,
|
| 3512 |
"trial_params": null
|