Add model files
Browse files- all_results.json +6 -6
- config.json +1 -1
- pytorch_model.bin +1 -1
- train_results.json +6 -6
- trainer_state.json +1046 -12
- training_args.bin +1 -1
all_results.json
CHANGED
|
@@ -1,15 +1,15 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"init_mem_cpu_alloc_delta":
|
| 4 |
"init_mem_cpu_peaked_delta": 18306,
|
| 5 |
"init_mem_gpu_alloc_delta": 1261972480,
|
| 6 |
"init_mem_gpu_peaked_delta": 0,
|
| 7 |
-
"total_flos":
|
| 8 |
"train_mem_cpu_alloc_delta": 0,
|
| 9 |
"train_mem_cpu_peaked_delta": 0,
|
| 10 |
-
"train_mem_gpu_alloc_delta":
|
| 11 |
"train_mem_gpu_peaked_delta": 0,
|
| 12 |
-
"train_runtime":
|
| 13 |
"train_samples": 7434,
|
| 14 |
-
"train_samples_per_second":
|
| 15 |
}
|
| 1 |
{
|
| 2 |
+
"epoch": 40.0,
|
| 3 |
+
"init_mem_cpu_alloc_delta": 64929,
|
| 4 |
"init_mem_cpu_peaked_delta": 18306,
|
| 5 |
"init_mem_gpu_alloc_delta": 1261972480,
|
| 6 |
"init_mem_gpu_peaked_delta": 0,
|
| 7 |
+
"total_flos": 5.602791262432019e+19,
|
| 8 |
"train_mem_cpu_alloc_delta": 0,
|
| 9 |
"train_mem_cpu_peaked_delta": 0,
|
| 10 |
+
"train_mem_gpu_alloc_delta": 6367137280,
|
| 11 |
"train_mem_gpu_peaked_delta": 0,
|
| 12 |
+
"train_runtime": 9410.6193,
|
| 13 |
"train_samples": 7434,
|
| 14 |
+
"train_samples_per_second": 0.986
|
| 15 |
}
|
config.json
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
{
|
| 2 |
-
"_name_or_path": "/workspace/data2/el/wav2vec2-large-xlsr-greek/checkpoint-
|
| 3 |
"activation_dropout": 0.0,
|
| 4 |
"apply_spec_augment": true,
|
| 5 |
"architectures": [
|
| 1 |
{
|
| 2 |
+
"_name_or_path": "/workspace/data2/el/wav2vec2-large-xlsr-greek/checkpoint-9280",
|
| 3 |
"activation_dropout": 0.0,
|
| 4 |
"apply_spec_augment": true,
|
| 5 |
"architectures": [
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1262151127
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:17f7a6c0b25909b375e4ee72d8cc71e87ddde508311325b2e8562d4bb69c7c91
|
| 3 |
size 1262151127
|
train_results.json
CHANGED
|
@@ -1,15 +1,15 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"init_mem_cpu_alloc_delta":
|
| 4 |
"init_mem_cpu_peaked_delta": 18306,
|
| 5 |
"init_mem_gpu_alloc_delta": 1261972480,
|
| 6 |
"init_mem_gpu_peaked_delta": 0,
|
| 7 |
-
"total_flos":
|
| 8 |
"train_mem_cpu_alloc_delta": 0,
|
| 9 |
"train_mem_cpu_peaked_delta": 0,
|
| 10 |
-
"train_mem_gpu_alloc_delta":
|
| 11 |
"train_mem_gpu_peaked_delta": 0,
|
| 12 |
-
"train_runtime":
|
| 13 |
"train_samples": 7434,
|
| 14 |
-
"train_samples_per_second":
|
| 15 |
}
|
| 1 |
{
|
| 2 |
+
"epoch": 40.0,
|
| 3 |
+
"init_mem_cpu_alloc_delta": 64929,
|
| 4 |
"init_mem_cpu_peaked_delta": 18306,
|
| 5 |
"init_mem_gpu_alloc_delta": 1261972480,
|
| 6 |
"init_mem_gpu_peaked_delta": 0,
|
| 7 |
+
"total_flos": 5.602791262432019e+19,
|
| 8 |
"train_mem_cpu_alloc_delta": 0,
|
| 9 |
"train_mem_cpu_peaked_delta": 0,
|
| 10 |
+
"train_mem_gpu_alloc_delta": 6367137280,
|
| 11 |
"train_mem_gpu_peaked_delta": 0,
|
| 12 |
+
"train_runtime": 9410.6193,
|
| 13 |
"train_samples": 7434,
|
| 14 |
+
"train_samples_per_second": 0.986
|
| 15 |
}
|
trainer_state.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 0.
|
| 3 |
-
"best_model_checkpoint": "/workspace/data2/el/wav2vec2-large-xlsr-greek/checkpoint-
|
| 4 |
-
"epoch":
|
| 5 |
-
"global_step":
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
@@ -4926,16 +4926,1050 @@
|
|
| 4926 |
"step": 7656
|
| 4927 |
},
|
| 4928 |
{
|
| 4929 |
-
"epoch": 33.
|
| 4930 |
-
"
|
| 4931 |
-
"
|
| 4932 |
-
"
|
| 4933 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 4934 |
}
|
| 4935 |
],
|
| 4936 |
-
"max_steps":
|
| 4937 |
-
"num_train_epochs":
|
| 4938 |
-
"total_flos":
|
| 4939 |
"trial_name": null,
|
| 4940 |
"trial_params": null
|
| 4941 |
}
|
| 1 |
{
|
| 2 |
+
"best_metric": 0.3336529726456041,
|
| 3 |
+
"best_model_checkpoint": "/workspace/data2/el/wav2vec2-large-xlsr-greek/checkpoint-9280",
|
| 4 |
+
"epoch": 39.99784946236559,
|
| 5 |
+
"global_step": 9280,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
| 4926 |
"step": 7656
|
| 4927 |
},
|
| 4928 |
{
|
| 4929 |
+
"epoch": 33.02,
|
| 4930 |
+
"learning_rate": 5.248380129589633e-05,
|
| 4931 |
+
"loss": 0.0542,
|
| 4932 |
+
"step": 7660
|
| 4933 |
+
},
|
| 4934 |
+
{
|
| 4935 |
+
"epoch": 33.06,
|
| 4936 |
+
"learning_rate": 5.2159827213822885e-05,
|
| 4937 |
+
"loss": 0.0352,
|
| 4938 |
+
"step": 7670
|
| 4939 |
+
},
|
| 4940 |
+
{
|
| 4941 |
+
"epoch": 33.1,
|
| 4942 |
+
"learning_rate": 5.183585313174945e-05,
|
| 4943 |
+
"loss": 0.0387,
|
| 4944 |
+
"step": 7680
|
| 4945 |
+
},
|
| 4946 |
+
{
|
| 4947 |
+
"epoch": 33.15,
|
| 4948 |
+
"learning_rate": 5.151187904967602e-05,
|
| 4949 |
+
"loss": 0.065,
|
| 4950 |
+
"step": 7690
|
| 4951 |
+
},
|
| 4952 |
+
{
|
| 4953 |
+
"epoch": 33.19,
|
| 4954 |
+
"learning_rate": 5.118790496760259e-05,
|
| 4955 |
+
"loss": 0.0542,
|
| 4956 |
+
"step": 7700
|
| 4957 |
+
},
|
| 4958 |
+
{
|
| 4959 |
+
"epoch": 33.23,
|
| 4960 |
+
"learning_rate": 5.086393088552915e-05,
|
| 4961 |
+
"loss": 0.0498,
|
| 4962 |
+
"step": 7710
|
| 4963 |
+
},
|
| 4964 |
+
{
|
| 4965 |
+
"epoch": 33.28,
|
| 4966 |
+
"learning_rate": 5.053995680345572e-05,
|
| 4967 |
+
"loss": 0.0771,
|
| 4968 |
+
"step": 7720
|
| 4969 |
+
},
|
| 4970 |
+
{
|
| 4971 |
+
"epoch": 33.32,
|
| 4972 |
+
"learning_rate": 5.021598272138229e-05,
|
| 4973 |
+
"loss": 0.0614,
|
| 4974 |
+
"step": 7730
|
| 4975 |
+
},
|
| 4976 |
+
{
|
| 4977 |
+
"epoch": 33.36,
|
| 4978 |
+
"learning_rate": 4.9892008639308855e-05,
|
| 4979 |
+
"loss": 0.0633,
|
| 4980 |
+
"step": 7740
|
| 4981 |
+
},
|
| 4982 |
+
{
|
| 4983 |
+
"epoch": 33.4,
|
| 4984 |
+
"learning_rate": 4.956803455723542e-05,
|
| 4985 |
+
"loss": 0.0574,
|
| 4986 |
+
"step": 7750
|
| 4987 |
+
},
|
| 4988 |
+
{
|
| 4989 |
+
"epoch": 33.45,
|
| 4990 |
+
"learning_rate": 4.924406047516198e-05,
|
| 4991 |
+
"loss": 0.038,
|
| 4992 |
+
"step": 7760
|
| 4993 |
+
},
|
| 4994 |
+
{
|
| 4995 |
+
"epoch": 33.49,
|
| 4996 |
+
"learning_rate": 4.8920086393088544e-05,
|
| 4997 |
+
"loss": 0.0588,
|
| 4998 |
+
"step": 7770
|
| 4999 |
+
},
|
| 5000 |
+
{
|
| 5001 |
+
"epoch": 33.53,
|
| 5002 |
+
"learning_rate": 4.8596112311015116e-05,
|
| 5003 |
+
"loss": 0.0484,
|
| 5004 |
+
"step": 7780
|
| 5005 |
+
},
|
| 5006 |
+
{
|
| 5007 |
+
"epoch": 33.58,
|
| 5008 |
+
"learning_rate": 4.827213822894168e-05,
|
| 5009 |
+
"loss": 0.0645,
|
| 5010 |
+
"step": 7790
|
| 5011 |
+
},
|
| 5012 |
+
{
|
| 5013 |
+
"epoch": 33.62,
|
| 5014 |
+
"learning_rate": 4.7948164146868247e-05,
|
| 5015 |
+
"loss": 0.0545,
|
| 5016 |
+
"step": 7800
|
| 5017 |
+
},
|
| 5018 |
+
{
|
| 5019 |
+
"epoch": 33.66,
|
| 5020 |
+
"learning_rate": 4.762419006479481e-05,
|
| 5021 |
+
"loss": 0.0617,
|
| 5022 |
+
"step": 7810
|
| 5023 |
+
},
|
| 5024 |
+
{
|
| 5025 |
+
"epoch": 33.71,
|
| 5026 |
+
"learning_rate": 4.7300215982721384e-05,
|
| 5027 |
+
"loss": 0.0899,
|
| 5028 |
+
"step": 7820
|
| 5029 |
+
},
|
| 5030 |
+
{
|
| 5031 |
+
"epoch": 33.75,
|
| 5032 |
+
"learning_rate": 4.697624190064795e-05,
|
| 5033 |
+
"loss": 0.0613,
|
| 5034 |
+
"step": 7830
|
| 5035 |
+
},
|
| 5036 |
+
{
|
| 5037 |
+
"epoch": 33.79,
|
| 5038 |
+
"learning_rate": 4.665226781857451e-05,
|
| 5039 |
+
"loss": 0.065,
|
| 5040 |
+
"step": 7840
|
| 5041 |
+
},
|
| 5042 |
+
{
|
| 5043 |
+
"epoch": 33.83,
|
| 5044 |
+
"learning_rate": 4.632829373650107e-05,
|
| 5045 |
+
"loss": 0.0541,
|
| 5046 |
+
"step": 7850
|
| 5047 |
+
},
|
| 5048 |
+
{
|
| 5049 |
+
"epoch": 33.88,
|
| 5050 |
+
"learning_rate": 4.600431965442764e-05,
|
| 5051 |
+
"loss": 0.0532,
|
| 5052 |
+
"step": 7860
|
| 5053 |
+
},
|
| 5054 |
+
{
|
| 5055 |
+
"epoch": 33.92,
|
| 5056 |
+
"learning_rate": 4.568034557235421e-05,
|
| 5057 |
+
"loss": 0.0704,
|
| 5058 |
+
"step": 7870
|
| 5059 |
+
},
|
| 5060 |
+
{
|
| 5061 |
+
"epoch": 33.96,
|
| 5062 |
+
"learning_rate": 4.5356371490280775e-05,
|
| 5063 |
+
"loss": 0.0565,
|
| 5064 |
+
"step": 7880
|
| 5065 |
+
},
|
| 5066 |
+
{
|
| 5067 |
+
"epoch": 34.0,
|
| 5068 |
+
"eval_loss": 0.3123264014720917,
|
| 5069 |
+
"eval_runtime": 273.8278,
|
| 5070 |
+
"eval_samples_per_second": 11.116,
|
| 5071 |
+
"eval_wer": 0.3369839507418997,
|
| 5072 |
+
"step": 7888
|
| 5073 |
+
},
|
| 5074 |
+
{
|
| 5075 |
+
"epoch": 34.01,
|
| 5076 |
+
"learning_rate": 4.503239740820734e-05,
|
| 5077 |
+
"loss": 0.0576,
|
| 5078 |
+
"step": 7890
|
| 5079 |
+
},
|
| 5080 |
+
{
|
| 5081 |
+
"epoch": 34.05,
|
| 5082 |
+
"learning_rate": 4.47084233261339e-05,
|
| 5083 |
+
"loss": 0.0737,
|
| 5084 |
+
"step": 7900
|
| 5085 |
+
},
|
| 5086 |
+
{
|
| 5087 |
+
"epoch": 34.09,
|
| 5088 |
+
"learning_rate": 4.438444924406048e-05,
|
| 5089 |
+
"loss": 0.0541,
|
| 5090 |
+
"step": 7910
|
| 5091 |
+
},
|
| 5092 |
+
{
|
| 5093 |
+
"epoch": 34.14,
|
| 5094 |
+
"learning_rate": 4.4060475161987036e-05,
|
| 5095 |
+
"loss": 0.0416,
|
| 5096 |
+
"step": 7920
|
| 5097 |
+
},
|
| 5098 |
+
{
|
| 5099 |
+
"epoch": 34.18,
|
| 5100 |
+
"learning_rate": 4.37365010799136e-05,
|
| 5101 |
+
"loss": 0.0541,
|
| 5102 |
+
"step": 7930
|
| 5103 |
+
},
|
| 5104 |
+
{
|
| 5105 |
+
"epoch": 34.22,
|
| 5106 |
+
"learning_rate": 4.3412526997840166e-05,
|
| 5107 |
+
"loss": 0.059,
|
| 5108 |
+
"step": 7940
|
| 5109 |
+
},
|
| 5110 |
+
{
|
| 5111 |
+
"epoch": 34.27,
|
| 5112 |
+
"learning_rate": 4.308855291576674e-05,
|
| 5113 |
+
"loss": 0.0538,
|
| 5114 |
+
"step": 7950
|
| 5115 |
+
},
|
| 5116 |
+
{
|
| 5117 |
+
"epoch": 34.31,
|
| 5118 |
+
"learning_rate": 4.2764578833693303e-05,
|
| 5119 |
+
"loss": 0.0795,
|
| 5120 |
+
"step": 7960
|
| 5121 |
+
},
|
| 5122 |
+
{
|
| 5123 |
+
"epoch": 34.35,
|
| 5124 |
+
"learning_rate": 4.244060475161987e-05,
|
| 5125 |
+
"loss": 0.0618,
|
| 5126 |
+
"step": 7970
|
| 5127 |
+
},
|
| 5128 |
+
{
|
| 5129 |
+
"epoch": 34.4,
|
| 5130 |
+
"learning_rate": 4.2116630669546434e-05,
|
| 5131 |
+
"loss": 0.0667,
|
| 5132 |
+
"step": 7980
|
| 5133 |
+
},
|
| 5134 |
+
{
|
| 5135 |
+
"epoch": 34.44,
|
| 5136 |
+
"learning_rate": 4.179265658747299e-05,
|
| 5137 |
+
"loss": 0.0501,
|
| 5138 |
+
"step": 7990
|
| 5139 |
+
},
|
| 5140 |
+
{
|
| 5141 |
+
"epoch": 34.48,
|
| 5142 |
+
"learning_rate": 4.1468682505399564e-05,
|
| 5143 |
+
"loss": 0.0564,
|
| 5144 |
+
"step": 8000
|
| 5145 |
+
},
|
| 5146 |
+
{
|
| 5147 |
+
"epoch": 34.52,
|
| 5148 |
+
"learning_rate": 4.114470842332613e-05,
|
| 5149 |
+
"loss": 0.0507,
|
| 5150 |
+
"step": 8010
|
| 5151 |
+
},
|
| 5152 |
+
{
|
| 5153 |
+
"epoch": 34.57,
|
| 5154 |
+
"learning_rate": 4.0820734341252695e-05,
|
| 5155 |
+
"loss": 0.0785,
|
| 5156 |
+
"step": 8020
|
| 5157 |
+
},
|
| 5158 |
+
{
|
| 5159 |
+
"epoch": 34.61,
|
| 5160 |
+
"learning_rate": 4.049676025917926e-05,
|
| 5161 |
+
"loss": 0.0697,
|
| 5162 |
+
"step": 8030
|
| 5163 |
+
},
|
| 5164 |
+
{
|
| 5165 |
+
"epoch": 34.65,
|
| 5166 |
+
"learning_rate": 4.017278617710583e-05,
|
| 5167 |
+
"loss": 0.0675,
|
| 5168 |
+
"step": 8040
|
| 5169 |
+
},
|
| 5170 |
+
{
|
| 5171 |
+
"epoch": 34.7,
|
| 5172 |
+
"learning_rate": 3.98488120950324e-05,
|
| 5173 |
+
"loss": 0.0516,
|
| 5174 |
+
"step": 8050
|
| 5175 |
+
},
|
| 5176 |
+
{
|
| 5177 |
+
"epoch": 34.74,
|
| 5178 |
+
"learning_rate": 3.952483801295896e-05,
|
| 5179 |
+
"loss": 0.0556,
|
| 5180 |
+
"step": 8060
|
| 5181 |
+
},
|
| 5182 |
+
{
|
| 5183 |
+
"epoch": 34.78,
|
| 5184 |
+
"learning_rate": 3.920086393088552e-05,
|
| 5185 |
+
"loss": 0.0739,
|
| 5186 |
+
"step": 8070
|
| 5187 |
+
},
|
| 5188 |
+
{
|
| 5189 |
+
"epoch": 34.83,
|
| 5190 |
+
"learning_rate": 3.887688984881209e-05,
|
| 5191 |
+
"loss": 0.0584,
|
| 5192 |
+
"step": 8080
|
| 5193 |
+
},
|
| 5194 |
+
{
|
| 5195 |
+
"epoch": 34.87,
|
| 5196 |
+
"learning_rate": 3.855291576673866e-05,
|
| 5197 |
+
"loss": 0.0827,
|
| 5198 |
+
"step": 8090
|
| 5199 |
+
},
|
| 5200 |
+
{
|
| 5201 |
+
"epoch": 34.91,
|
| 5202 |
+
"learning_rate": 3.822894168466522e-05,
|
| 5203 |
+
"loss": 0.0504,
|
| 5204 |
+
"step": 8100
|
| 5205 |
+
},
|
| 5206 |
+
{
|
| 5207 |
+
"epoch": 34.95,
|
| 5208 |
+
"learning_rate": 3.790496760259179e-05,
|
| 5209 |
+
"loss": 0.0521,
|
| 5210 |
+
"step": 8110
|
| 5211 |
+
},
|
| 5212 |
+
{
|
| 5213 |
+
"epoch": 35.0,
|
| 5214 |
+
"learning_rate": 3.7580993520518354e-05,
|
| 5215 |
+
"loss": 0.0568,
|
| 5216 |
+
"step": 8120
|
| 5217 |
+
},
|
| 5218 |
+
{
|
| 5219 |
+
"epoch": 35.0,
|
| 5220 |
+
"eval_loss": 0.30912891030311584,
|
| 5221 |
+
"eval_runtime": 273.0561,
|
| 5222 |
+
"eval_samples_per_second": 11.148,
|
| 5223 |
+
"eval_wer": 0.3364287877258504,
|
| 5224 |
+
"step": 8120
|
| 5225 |
+
},
|
| 5226 |
+
{
|
| 5227 |
+
"epoch": 35.04,
|
| 5228 |
+
"learning_rate": 3.725701943844492e-05,
|
| 5229 |
+
"loss": 0.0615,
|
| 5230 |
+
"step": 8130
|
| 5231 |
+
},
|
| 5232 |
+
{
|
| 5233 |
+
"epoch": 35.09,
|
| 5234 |
+
"learning_rate": 3.693304535637149e-05,
|
| 5235 |
+
"loss": 0.0569,
|
| 5236 |
+
"step": 8140
|
| 5237 |
+
},
|
| 5238 |
+
{
|
| 5239 |
+
"epoch": 35.13,
|
| 5240 |
+
"learning_rate": 3.660907127429805e-05,
|
| 5241 |
+
"loss": 0.0747,
|
| 5242 |
+
"step": 8150
|
| 5243 |
+
},
|
| 5244 |
+
{
|
| 5245 |
+
"epoch": 35.17,
|
| 5246 |
+
"learning_rate": 3.628509719222462e-05,
|
| 5247 |
+
"loss": 0.0527,
|
| 5248 |
+
"step": 8160
|
| 5249 |
+
},
|
| 5250 |
+
{
|
| 5251 |
+
"epoch": 35.22,
|
| 5252 |
+
"learning_rate": 3.5961123110151187e-05,
|
| 5253 |
+
"loss": 0.0622,
|
| 5254 |
+
"step": 8170
|
| 5255 |
+
},
|
| 5256 |
+
{
|
| 5257 |
+
"epoch": 35.26,
|
| 5258 |
+
"learning_rate": 3.563714902807775e-05,
|
| 5259 |
+
"loss": 0.0689,
|
| 5260 |
+
"step": 8180
|
| 5261 |
+
},
|
| 5262 |
+
{
|
| 5263 |
+
"epoch": 35.3,
|
| 5264 |
+
"learning_rate": 3.531317494600432e-05,
|
| 5265 |
+
"loss": 0.0658,
|
| 5266 |
+
"step": 8190
|
| 5267 |
+
},
|
| 5268 |
+
{
|
| 5269 |
+
"epoch": 35.34,
|
| 5270 |
+
"learning_rate": 3.498920086393088e-05,
|
| 5271 |
+
"loss": 0.0411,
|
| 5272 |
+
"step": 8200
|
| 5273 |
+
},
|
| 5274 |
+
{
|
| 5275 |
+
"epoch": 35.39,
|
| 5276 |
+
"learning_rate": 3.466522678185745e-05,
|
| 5277 |
+
"loss": 0.0732,
|
| 5278 |
+
"step": 8210
|
| 5279 |
+
},
|
| 5280 |
+
{
|
| 5281 |
+
"epoch": 35.43,
|
| 5282 |
+
"learning_rate": 3.434125269978401e-05,
|
| 5283 |
+
"loss": 0.0568,
|
| 5284 |
+
"step": 8220
|
| 5285 |
+
},
|
| 5286 |
+
{
|
| 5287 |
+
"epoch": 35.47,
|
| 5288 |
+
"learning_rate": 3.401727861771058e-05,
|
| 5289 |
+
"loss": 0.0512,
|
| 5290 |
+
"step": 8230
|
| 5291 |
+
},
|
| 5292 |
+
{
|
| 5293 |
+
"epoch": 35.52,
|
| 5294 |
+
"learning_rate": 3.369330453563714e-05,
|
| 5295 |
+
"loss": 0.0373,
|
| 5296 |
+
"step": 8240
|
| 5297 |
+
},
|
| 5298 |
+
{
|
| 5299 |
+
"epoch": 35.56,
|
| 5300 |
+
"learning_rate": 3.3369330453563715e-05,
|
| 5301 |
+
"loss": 0.0667,
|
| 5302 |
+
"step": 8250
|
| 5303 |
+
},
|
| 5304 |
+
{
|
| 5305 |
+
"epoch": 35.6,
|
| 5306 |
+
"learning_rate": 3.3045356371490274e-05,
|
| 5307 |
+
"loss": 0.0433,
|
| 5308 |
+
"step": 8260
|
| 5309 |
+
},
|
| 5310 |
+
{
|
| 5311 |
+
"epoch": 35.65,
|
| 5312 |
+
"learning_rate": 3.2721382289416845e-05,
|
| 5313 |
+
"loss": 0.0595,
|
| 5314 |
+
"step": 8270
|
| 5315 |
+
},
|
| 5316 |
+
{
|
| 5317 |
+
"epoch": 35.69,
|
| 5318 |
+
"learning_rate": 3.239740820734341e-05,
|
| 5319 |
+
"loss": 0.0569,
|
| 5320 |
+
"step": 8280
|
| 5321 |
+
},
|
| 5322 |
+
{
|
| 5323 |
+
"epoch": 35.73,
|
| 5324 |
+
"learning_rate": 3.2073434125269976e-05,
|
| 5325 |
+
"loss": 0.0612,
|
| 5326 |
+
"step": 8290
|
| 5327 |
+
},
|
| 5328 |
+
{
|
| 5329 |
+
"epoch": 35.77,
|
| 5330 |
+
"learning_rate": 3.174946004319654e-05,
|
| 5331 |
+
"loss": 0.0681,
|
| 5332 |
+
"step": 8300
|
| 5333 |
+
},
|
| 5334 |
+
{
|
| 5335 |
+
"epoch": 35.82,
|
| 5336 |
+
"learning_rate": 3.1425485961123106e-05,
|
| 5337 |
+
"loss": 0.0579,
|
| 5338 |
+
"step": 8310
|
| 5339 |
+
},
|
| 5340 |
+
{
|
| 5341 |
+
"epoch": 35.86,
|
| 5342 |
+
"learning_rate": 3.110151187904967e-05,
|
| 5343 |
+
"loss": 0.0397,
|
| 5344 |
+
"step": 8320
|
| 5345 |
+
},
|
| 5346 |
+
{
|
| 5347 |
+
"epoch": 35.9,
|
| 5348 |
+
"learning_rate": 3.0777537796976244e-05,
|
| 5349 |
+
"loss": 0.0604,
|
| 5350 |
+
"step": 8330
|
| 5351 |
+
},
|
| 5352 |
+
{
|
| 5353 |
+
"epoch": 35.95,
|
| 5354 |
+
"learning_rate": 3.0453563714902805e-05,
|
| 5355 |
+
"loss": 0.0669,
|
| 5356 |
+
"step": 8340
|
| 5357 |
+
},
|
| 5358 |
+
{
|
| 5359 |
+
"epoch": 35.99,
|
| 5360 |
+
"learning_rate": 3.012958963282937e-05,
|
| 5361 |
+
"loss": 0.0616,
|
| 5362 |
+
"step": 8350
|
| 5363 |
+
},
|
| 5364 |
+
{
|
| 5365 |
+
"epoch": 36.0,
|
| 5366 |
+
"eval_loss": 0.31008386611938477,
|
| 5367 |
+
"eval_runtime": 276.5254,
|
| 5368 |
+
"eval_samples_per_second": 11.008,
|
| 5369 |
+
"eval_wer": 0.3399616432825275,
|
| 5370 |
+
"step": 8352
|
| 5371 |
+
},
|
| 5372 |
+
{
|
| 5373 |
+
"epoch": 36.03,
|
| 5374 |
+
"learning_rate": 2.9805615550755936e-05,
|
| 5375 |
+
"loss": 0.0434,
|
| 5376 |
+
"step": 8360
|
| 5377 |
+
},
|
| 5378 |
+
{
|
| 5379 |
+
"epoch": 36.08,
|
| 5380 |
+
"learning_rate": 2.94816414686825e-05,
|
| 5381 |
+
"loss": 0.0598,
|
| 5382 |
+
"step": 8370
|
| 5383 |
+
},
|
| 5384 |
+
{
|
| 5385 |
+
"epoch": 36.12,
|
| 5386 |
+
"learning_rate": 2.915766738660907e-05,
|
| 5387 |
+
"loss": 0.0565,
|
| 5388 |
+
"step": 8380
|
| 5389 |
+
},
|
| 5390 |
+
{
|
| 5391 |
+
"epoch": 36.16,
|
| 5392 |
+
"learning_rate": 2.8833693304535635e-05,
|
| 5393 |
+
"loss": 0.056,
|
| 5394 |
+
"step": 8390
|
| 5395 |
+
},
|
| 5396 |
+
{
|
| 5397 |
+
"epoch": 36.21,
|
| 5398 |
+
"learning_rate": 2.85097192224622e-05,
|
| 5399 |
+
"loss": 0.0563,
|
| 5400 |
+
"step": 8400
|
| 5401 |
+
},
|
| 5402 |
+
{
|
| 5403 |
+
"epoch": 36.25,
|
| 5404 |
+
"learning_rate": 2.8185745140388765e-05,
|
| 5405 |
+
"loss": 0.064,
|
| 5406 |
+
"step": 8410
|
| 5407 |
+
},
|
| 5408 |
+
{
|
| 5409 |
+
"epoch": 36.29,
|
| 5410 |
+
"learning_rate": 2.7861771058315334e-05,
|
| 5411 |
+
"loss": 0.0466,
|
| 5412 |
+
"step": 8420
|
| 5413 |
+
},
|
| 5414 |
+
{
|
| 5415 |
+
"epoch": 36.34,
|
| 5416 |
+
"learning_rate": 2.75377969762419e-05,
|
| 5417 |
+
"loss": 0.0445,
|
| 5418 |
+
"step": 8430
|
| 5419 |
+
},
|
| 5420 |
+
{
|
| 5421 |
+
"epoch": 36.38,
|
| 5422 |
+
"learning_rate": 2.7213822894168468e-05,
|
| 5423 |
+
"loss": 0.0504,
|
| 5424 |
+
"step": 8440
|
| 5425 |
+
},
|
| 5426 |
+
{
|
| 5427 |
+
"epoch": 36.42,
|
| 5428 |
+
"learning_rate": 2.688984881209503e-05,
|
| 5429 |
+
"loss": 0.0715,
|
| 5430 |
+
"step": 8450
|
| 5431 |
+
},
|
| 5432 |
+
{
|
| 5433 |
+
"epoch": 36.46,
|
| 5434 |
+
"learning_rate": 2.6565874730021595e-05,
|
| 5435 |
+
"loss": 0.0488,
|
| 5436 |
+
"step": 8460
|
| 5437 |
+
},
|
| 5438 |
+
{
|
| 5439 |
+
"epoch": 36.51,
|
| 5440 |
+
"learning_rate": 2.6241900647948163e-05,
|
| 5441 |
+
"loss": 0.0493,
|
| 5442 |
+
"step": 8470
|
| 5443 |
+
},
|
| 5444 |
+
{
|
| 5445 |
+
"epoch": 36.55,
|
| 5446 |
+
"learning_rate": 2.5917926565874725e-05,
|
| 5447 |
+
"loss": 0.0395,
|
| 5448 |
+
"step": 8480
|
| 5449 |
+
},
|
| 5450 |
+
{
|
| 5451 |
+
"epoch": 36.59,
|
| 5452 |
+
"learning_rate": 2.5593952483801294e-05,
|
| 5453 |
+
"loss": 0.0683,
|
| 5454 |
+
"step": 8490
|
| 5455 |
+
},
|
| 5456 |
+
{
|
| 5457 |
+
"epoch": 36.64,
|
| 5458 |
+
"learning_rate": 2.526997840172786e-05,
|
| 5459 |
+
"loss": 0.0584,
|
| 5460 |
+
"step": 8500
|
| 5461 |
+
},
|
| 5462 |
+
{
|
| 5463 |
+
"epoch": 36.68,
|
| 5464 |
+
"learning_rate": 2.4946004319654428e-05,
|
| 5465 |
+
"loss": 0.07,
|
| 5466 |
+
"step": 8510
|
| 5467 |
+
},
|
| 5468 |
+
{
|
| 5469 |
+
"epoch": 36.72,
|
| 5470 |
+
"learning_rate": 2.462203023758099e-05,
|
| 5471 |
+
"loss": 0.0448,
|
| 5472 |
+
"step": 8520
|
| 5473 |
+
},
|
| 5474 |
+
{
|
| 5475 |
+
"epoch": 36.77,
|
| 5476 |
+
"learning_rate": 2.4298056155507558e-05,
|
| 5477 |
+
"loss": 0.0443,
|
| 5478 |
+
"step": 8530
|
| 5479 |
+
},
|
| 5480 |
+
{
|
| 5481 |
+
"epoch": 36.81,
|
| 5482 |
+
"learning_rate": 2.3974082073434123e-05,
|
| 5483 |
+
"loss": 0.0586,
|
| 5484 |
+
"step": 8540
|
| 5485 |
+
},
|
| 5486 |
+
{
|
| 5487 |
+
"epoch": 36.85,
|
| 5488 |
+
"learning_rate": 2.3650107991360692e-05,
|
| 5489 |
+
"loss": 0.069,
|
| 5490 |
+
"step": 8550
|
| 5491 |
+
},
|
| 5492 |
+
{
|
| 5493 |
+
"epoch": 36.89,
|
| 5494 |
+
"learning_rate": 2.3326133909287254e-05,
|
| 5495 |
+
"loss": 0.0621,
|
| 5496 |
+
"step": 8560
|
| 5497 |
+
},
|
| 5498 |
+
{
|
| 5499 |
+
"epoch": 36.94,
|
| 5500 |
+
"learning_rate": 2.300215982721382e-05,
|
| 5501 |
+
"loss": 0.0617,
|
| 5502 |
+
"step": 8570
|
| 5503 |
+
},
|
| 5504 |
+
{
|
| 5505 |
+
"epoch": 36.98,
|
| 5506 |
+
"learning_rate": 2.2678185745140387e-05,
|
| 5507 |
+
"loss": 0.0589,
|
| 5508 |
+
"step": 8580
|
| 5509 |
+
},
|
| 5510 |
+
{
|
| 5511 |
+
"epoch": 37.0,
|
| 5512 |
+
"eval_loss": 0.3041793406009674,
|
| 5513 |
+
"eval_runtime": 277.9516,
|
| 5514 |
+
"eval_samples_per_second": 10.952,
|
| 5515 |
+
"eval_wer": 0.3337034420106995,
|
| 5516 |
+
"step": 8584
|
| 5517 |
+
},
|
| 5518 |
+
{
|
| 5519 |
+
"epoch": 37.03,
|
| 5520 |
+
"learning_rate": 2.235421166306695e-05,
|
| 5521 |
+
"loss": 0.0452,
|
| 5522 |
+
"step": 8590
|
| 5523 |
+
},
|
| 5524 |
+
{
|
| 5525 |
+
"epoch": 37.07,
|
| 5526 |
+
"learning_rate": 2.2030237580993518e-05,
|
| 5527 |
+
"loss": 0.0503,
|
| 5528 |
+
"step": 8600
|
| 5529 |
+
},
|
| 5530 |
+
{
|
| 5531 |
+
"epoch": 37.11,
|
| 5532 |
+
"learning_rate": 2.1706263498920083e-05,
|
| 5533 |
+
"loss": 0.0464,
|
| 5534 |
+
"step": 8610
|
| 5535 |
+
},
|
| 5536 |
+
{
|
| 5537 |
+
"epoch": 37.15,
|
| 5538 |
+
"learning_rate": 2.1382289416846652e-05,
|
| 5539 |
+
"loss": 0.0423,
|
| 5540 |
+
"step": 8620
|
| 5541 |
+
},
|
| 5542 |
+
{
|
| 5543 |
+
"epoch": 37.2,
|
| 5544 |
+
"learning_rate": 2.1058315334773217e-05,
|
| 5545 |
+
"loss": 0.0625,
|
| 5546 |
+
"step": 8630
|
| 5547 |
+
},
|
| 5548 |
+
{
|
| 5549 |
+
"epoch": 37.24,
|
| 5550 |
+
"learning_rate": 2.0734341252699782e-05,
|
| 5551 |
+
"loss": 0.0497,
|
| 5552 |
+
"step": 8640
|
| 5553 |
+
},
|
| 5554 |
+
{
|
| 5555 |
+
"epoch": 37.28,
|
| 5556 |
+
"learning_rate": 2.0410367170626347e-05,
|
| 5557 |
+
"loss": 0.0427,
|
| 5558 |
+
"step": 8650
|
| 5559 |
+
},
|
| 5560 |
+
{
|
| 5561 |
+
"epoch": 37.33,
|
| 5562 |
+
"learning_rate": 2.0086393088552916e-05,
|
| 5563 |
+
"loss": 0.0562,
|
| 5564 |
+
"step": 8660
|
| 5565 |
+
},
|
| 5566 |
+
{
|
| 5567 |
+
"epoch": 37.37,
|
| 5568 |
+
"learning_rate": 1.976241900647948e-05,
|
| 5569 |
+
"loss": 0.0371,
|
| 5570 |
+
"step": 8670
|
| 5571 |
+
},
|
| 5572 |
+
{
|
| 5573 |
+
"epoch": 37.41,
|
| 5574 |
+
"learning_rate": 1.9438444924406046e-05,
|
| 5575 |
+
"loss": 0.0661,
|
| 5576 |
+
"step": 8680
|
| 5577 |
+
},
|
| 5578 |
+
{
|
| 5579 |
+
"epoch": 37.46,
|
| 5580 |
+
"learning_rate": 1.911447084233261e-05,
|
| 5581 |
+
"loss": 0.0556,
|
| 5582 |
+
"step": 8690
|
| 5583 |
+
},
|
| 5584 |
+
{
|
| 5585 |
+
"epoch": 37.5,
|
| 5586 |
+
"learning_rate": 1.8790496760259177e-05,
|
| 5587 |
+
"loss": 0.0711,
|
| 5588 |
+
"step": 8700
|
| 5589 |
+
},
|
| 5590 |
+
{
|
| 5591 |
+
"epoch": 37.54,
|
| 5592 |
+
"learning_rate": 1.8466522678185745e-05,
|
| 5593 |
+
"loss": 0.0568,
|
| 5594 |
+
"step": 8710
|
| 5595 |
+
},
|
| 5596 |
+
{
|
| 5597 |
+
"epoch": 37.58,
|
| 5598 |
+
"learning_rate": 1.814254859611231e-05,
|
| 5599 |
+
"loss": 0.0389,
|
| 5600 |
+
"step": 8720
|
| 5601 |
+
},
|
| 5602 |
+
{
|
| 5603 |
+
"epoch": 37.63,
|
| 5604 |
+
"learning_rate": 1.7818574514038876e-05,
|
| 5605 |
+
"loss": 0.0668,
|
| 5606 |
+
"step": 8730
|
| 5607 |
+
},
|
| 5608 |
+
{
|
| 5609 |
+
"epoch": 37.67,
|
| 5610 |
+
"learning_rate": 1.749460043196544e-05,
|
| 5611 |
+
"loss": 0.0694,
|
| 5612 |
+
"step": 8740
|
| 5613 |
+
},
|
| 5614 |
+
{
|
| 5615 |
+
"epoch": 37.71,
|
| 5616 |
+
"learning_rate": 1.7170626349892006e-05,
|
| 5617 |
+
"loss": 0.0704,
|
| 5618 |
+
"step": 8750
|
| 5619 |
+
},
|
| 5620 |
+
{
|
| 5621 |
+
"epoch": 37.76,
|
| 5622 |
+
"learning_rate": 1.684665226781857e-05,
|
| 5623 |
+
"loss": 0.0373,
|
| 5624 |
+
"step": 8760
|
| 5625 |
+
},
|
| 5626 |
+
{
|
| 5627 |
+
"epoch": 37.8,
|
| 5628 |
+
"learning_rate": 1.6522678185745137e-05,
|
| 5629 |
+
"loss": 0.0595,
|
| 5630 |
+
"step": 8770
|
| 5631 |
+
},
|
| 5632 |
+
{
|
| 5633 |
+
"epoch": 37.84,
|
| 5634 |
+
"learning_rate": 1.6198704103671705e-05,
|
| 5635 |
+
"loss": 0.0489,
|
| 5636 |
+
"step": 8780
|
| 5637 |
+
},
|
| 5638 |
+
{
|
| 5639 |
+
"epoch": 37.89,
|
| 5640 |
+
"learning_rate": 1.587473002159827e-05,
|
| 5641 |
+
"loss": 0.0448,
|
| 5642 |
+
"step": 8790
|
| 5643 |
+
},
|
| 5644 |
+
{
|
| 5645 |
+
"epoch": 37.93,
|
| 5646 |
+
"learning_rate": 1.5550755939524836e-05,
|
| 5647 |
+
"loss": 0.0524,
|
| 5648 |
+
"step": 8800
|
| 5649 |
+
},
|
| 5650 |
+
{
|
| 5651 |
+
"epoch": 37.97,
|
| 5652 |
+
"learning_rate": 1.5226781857451403e-05,
|
| 5653 |
+
"loss": 0.0548,
|
| 5654 |
+
"step": 8810
|
| 5655 |
+
},
|
| 5656 |
+
{
|
| 5657 |
+
"epoch": 38.0,
|
| 5658 |
+
"eval_loss": 0.3078618049621582,
|
| 5659 |
+
"eval_runtime": 282.4379,
|
| 5660 |
+
"eval_samples_per_second": 10.778,
|
| 5661 |
+
"eval_wer": 0.3337539113757949,
|
| 5662 |
+
"step": 8816
|
| 5663 |
+
},
|
| 5664 |
+
{
|
| 5665 |
+
"epoch": 38.02,
|
| 5666 |
+
"learning_rate": 1.4902807775377968e-05,
|
| 5667 |
+
"loss": 0.0635,
|
| 5668 |
+
"step": 8820
|
| 5669 |
+
},
|
| 5670 |
+
{
|
| 5671 |
+
"epoch": 38.06,
|
| 5672 |
+
"learning_rate": 1.4578833693304535e-05,
|
| 5673 |
+
"loss": 0.0457,
|
| 5674 |
+
"step": 8830
|
| 5675 |
+
},
|
| 5676 |
+
{
|
| 5677 |
+
"epoch": 38.1,
|
| 5678 |
+
"learning_rate": 1.42548596112311e-05,
|
| 5679 |
+
"loss": 0.0474,
|
| 5680 |
+
"step": 8840
|
| 5681 |
+
},
|
| 5682 |
+
{
|
| 5683 |
+
"epoch": 38.15,
|
| 5684 |
+
"learning_rate": 1.3930885529157667e-05,
|
| 5685 |
+
"loss": 0.0921,
|
| 5686 |
+
"step": 8850
|
| 5687 |
+
},
|
| 5688 |
+
{
|
| 5689 |
+
"epoch": 38.19,
|
| 5690 |
+
"learning_rate": 1.3606911447084234e-05,
|
| 5691 |
+
"loss": 0.0377,
|
| 5692 |
+
"step": 8860
|
| 5693 |
+
},
|
| 5694 |
+
{
|
| 5695 |
+
"epoch": 38.23,
|
| 5696 |
+
"learning_rate": 1.3282937365010797e-05,
|
| 5697 |
+
"loss": 0.0458,
|
| 5698 |
+
"step": 8870
|
| 5699 |
+
},
|
| 5700 |
+
{
|
| 5701 |
+
"epoch": 38.28,
|
| 5702 |
+
"learning_rate": 1.2958963282937363e-05,
|
| 5703 |
+
"loss": 0.0417,
|
| 5704 |
+
"step": 8880
|
| 5705 |
+
},
|
| 5706 |
+
{
|
| 5707 |
+
"epoch": 38.32,
|
| 5708 |
+
"learning_rate": 1.263498920086393e-05,
|
| 5709 |
+
"loss": 0.0498,
|
| 5710 |
+
"step": 8890
|
| 5711 |
+
},
|
| 5712 |
+
{
|
| 5713 |
+
"epoch": 38.36,
|
| 5714 |
+
"learning_rate": 1.2311015118790495e-05,
|
| 5715 |
+
"loss": 0.0307,
|
| 5716 |
+
"step": 8900
|
| 5717 |
+
},
|
| 5718 |
+
{
|
| 5719 |
+
"epoch": 38.4,
|
| 5720 |
+
"learning_rate": 1.1987041036717062e-05,
|
| 5721 |
+
"loss": 0.0378,
|
| 5722 |
+
"step": 8910
|
| 5723 |
+
},
|
| 5724 |
+
{
|
| 5725 |
+
"epoch": 38.45,
|
| 5726 |
+
"learning_rate": 1.1663066954643627e-05,
|
| 5727 |
+
"loss": 0.0489,
|
| 5728 |
+
"step": 8920
|
| 5729 |
+
},
|
| 5730 |
+
{
|
| 5731 |
+
"epoch": 38.49,
|
| 5732 |
+
"learning_rate": 1.1339092872570194e-05,
|
| 5733 |
+
"loss": 0.0526,
|
| 5734 |
+
"step": 8930
|
| 5735 |
+
},
|
| 5736 |
+
{
|
| 5737 |
+
"epoch": 38.53,
|
| 5738 |
+
"learning_rate": 1.1015118790496759e-05,
|
| 5739 |
+
"loss": 0.0483,
|
| 5740 |
+
"step": 8940
|
| 5741 |
+
},
|
| 5742 |
+
{
|
| 5743 |
+
"epoch": 38.58,
|
| 5744 |
+
"learning_rate": 1.0691144708423326e-05,
|
| 5745 |
+
"loss": 0.0721,
|
| 5746 |
+
"step": 8950
|
| 5747 |
+
},
|
| 5748 |
+
{
|
| 5749 |
+
"epoch": 38.62,
|
| 5750 |
+
"learning_rate": 1.0367170626349891e-05,
|
| 5751 |
+
"loss": 0.052,
|
| 5752 |
+
"step": 8960
|
| 5753 |
+
},
|
| 5754 |
+
{
|
| 5755 |
+
"epoch": 38.66,
|
| 5756 |
+
"learning_rate": 1.0043196544276458e-05,
|
| 5757 |
+
"loss": 0.0487,
|
| 5758 |
+
"step": 8970
|
| 5759 |
+
},
|
| 5760 |
+
{
|
| 5761 |
+
"epoch": 38.71,
|
| 5762 |
+
"learning_rate": 9.719222462203023e-06,
|
| 5763 |
+
"loss": 0.0369,
|
| 5764 |
+
"step": 8980
|
| 5765 |
+
},
|
| 5766 |
+
{
|
| 5767 |
+
"epoch": 38.75,
|
| 5768 |
+
"learning_rate": 9.395248380129588e-06,
|
| 5769 |
+
"loss": 0.0776,
|
| 5770 |
+
"step": 8990
|
| 5771 |
+
},
|
| 5772 |
+
{
|
| 5773 |
+
"epoch": 38.79,
|
| 5774 |
+
"learning_rate": 9.071274298056155e-06,
|
| 5775 |
+
"loss": 0.0454,
|
| 5776 |
+
"step": 9000
|
| 5777 |
+
},
|
| 5778 |
+
{
|
| 5779 |
+
"epoch": 38.83,
|
| 5780 |
+
"learning_rate": 8.74730021598272e-06,
|
| 5781 |
+
"loss": 0.0452,
|
| 5782 |
+
"step": 9010
|
| 5783 |
+
},
|
| 5784 |
+
{
|
| 5785 |
+
"epoch": 38.88,
|
| 5786 |
+
"learning_rate": 8.423326133909286e-06,
|
| 5787 |
+
"loss": 0.0601,
|
| 5788 |
+
"step": 9020
|
| 5789 |
+
},
|
| 5790 |
+
{
|
| 5791 |
+
"epoch": 38.92,
|
| 5792 |
+
"learning_rate": 8.099352051835853e-06,
|
| 5793 |
+
"loss": 0.0488,
|
| 5794 |
+
"step": 9030
|
| 5795 |
+
},
|
| 5796 |
+
{
|
| 5797 |
+
"epoch": 38.96,
|
| 5798 |
+
"learning_rate": 7.775377969762418e-06,
|
| 5799 |
+
"loss": 0.0614,
|
| 5800 |
+
"step": 9040
|
| 5801 |
+
},
|
| 5802 |
+
{
|
| 5803 |
+
"epoch": 39.0,
|
| 5804 |
+
"eval_loss": 0.31897786259651184,
|
| 5805 |
+
"eval_runtime": 287.468,
|
| 5806 |
+
"eval_samples_per_second": 10.589,
|
| 5807 |
+
"eval_wer": 0.3349651761380842,
|
| 5808 |
+
"step": 9048
|
| 5809 |
+
},
|
| 5810 |
+
{
|
| 5811 |
+
"epoch": 39.01,
|
| 5812 |
+
"learning_rate": 7.451403887688984e-06,
|
| 5813 |
+
"loss": 0.0615,
|
| 5814 |
+
"step": 9050
|
| 5815 |
+
},
|
| 5816 |
+
{
|
| 5817 |
+
"epoch": 39.05,
|
| 5818 |
+
"learning_rate": 7.12742980561555e-06,
|
| 5819 |
+
"loss": 0.0412,
|
| 5820 |
+
"step": 9060
|
| 5821 |
+
},
|
| 5822 |
+
{
|
| 5823 |
+
"epoch": 39.09,
|
| 5824 |
+
"learning_rate": 6.803455723542117e-06,
|
| 5825 |
+
"loss": 0.0496,
|
| 5826 |
+
"step": 9070
|
| 5827 |
+
},
|
| 5828 |
+
{
|
| 5829 |
+
"epoch": 39.14,
|
| 5830 |
+
"learning_rate": 6.479481641468681e-06,
|
| 5831 |
+
"loss": 0.0524,
|
| 5832 |
+
"step": 9080
|
| 5833 |
+
},
|
| 5834 |
+
{
|
| 5835 |
+
"epoch": 39.18,
|
| 5836 |
+
"learning_rate": 6.155507559395247e-06,
|
| 5837 |
+
"loss": 0.0485,
|
| 5838 |
+
"step": 9090
|
| 5839 |
+
},
|
| 5840 |
+
{
|
| 5841 |
+
"epoch": 39.22,
|
| 5842 |
+
"learning_rate": 5.831533477321813e-06,
|
| 5843 |
+
"loss": 0.0646,
|
| 5844 |
+
"step": 9100
|
| 5845 |
+
},
|
| 5846 |
+
{
|
| 5847 |
+
"epoch": 39.27,
|
| 5848 |
+
"learning_rate": 5.5075593952483795e-06,
|
| 5849 |
+
"loss": 0.0602,
|
| 5850 |
+
"step": 9110
|
| 5851 |
+
},
|
| 5852 |
+
{
|
| 5853 |
+
"epoch": 39.31,
|
| 5854 |
+
"learning_rate": 5.1835853131749455e-06,
|
| 5855 |
+
"loss": 0.0623,
|
| 5856 |
+
"step": 9120
|
| 5857 |
+
},
|
| 5858 |
+
{
|
| 5859 |
+
"epoch": 39.35,
|
| 5860 |
+
"learning_rate": 4.859611231101512e-06,
|
| 5861 |
+
"loss": 0.0399,
|
| 5862 |
+
"step": 9130
|
| 5863 |
+
},
|
| 5864 |
+
{
|
| 5865 |
+
"epoch": 39.4,
|
| 5866 |
+
"learning_rate": 4.535637149028078e-06,
|
| 5867 |
+
"loss": 0.0571,
|
| 5868 |
+
"step": 9140
|
| 5869 |
+
},
|
| 5870 |
+
{
|
| 5871 |
+
"epoch": 39.44,
|
| 5872 |
+
"learning_rate": 4.211663066954643e-06,
|
| 5873 |
+
"loss": 0.0668,
|
| 5874 |
+
"step": 9150
|
| 5875 |
+
},
|
| 5876 |
+
{
|
| 5877 |
+
"epoch": 39.48,
|
| 5878 |
+
"learning_rate": 3.887688984881209e-06,
|
| 5879 |
+
"loss": 0.0374,
|
| 5880 |
+
"step": 9160
|
| 5881 |
+
},
|
| 5882 |
+
{
|
| 5883 |
+
"epoch": 39.52,
|
| 5884 |
+
"learning_rate": 3.563714902807775e-06,
|
| 5885 |
+
"loss": 0.0424,
|
| 5886 |
+
"step": 9170
|
| 5887 |
+
},
|
| 5888 |
+
{
|
| 5889 |
+
"epoch": 39.57,
|
| 5890 |
+
"learning_rate": 3.2397408207343406e-06,
|
| 5891 |
+
"loss": 0.047,
|
| 5892 |
+
"step": 9180
|
| 5893 |
+
},
|
| 5894 |
+
{
|
| 5895 |
+
"epoch": 39.61,
|
| 5896 |
+
"learning_rate": 2.9157667386609067e-06,
|
| 5897 |
+
"loss": 0.0562,
|
| 5898 |
+
"step": 9190
|
| 5899 |
+
},
|
| 5900 |
+
{
|
| 5901 |
+
"epoch": 39.65,
|
| 5902 |
+
"learning_rate": 2.5917926565874728e-06,
|
| 5903 |
+
"loss": 0.0789,
|
| 5904 |
+
"step": 9200
|
| 5905 |
+
},
|
| 5906 |
+
{
|
| 5907 |
+
"epoch": 39.7,
|
| 5908 |
+
"learning_rate": 2.267818574514039e-06,
|
| 5909 |
+
"loss": 0.0485,
|
| 5910 |
+
"step": 9210
|
| 5911 |
+
},
|
| 5912 |
+
{
|
| 5913 |
+
"epoch": 39.74,
|
| 5914 |
+
"learning_rate": 1.9438444924406045e-06,
|
| 5915 |
+
"loss": 0.0513,
|
| 5916 |
+
"step": 9220
|
| 5917 |
+
},
|
| 5918 |
+
{
|
| 5919 |
+
"epoch": 39.78,
|
| 5920 |
+
"learning_rate": 1.6198704103671703e-06,
|
| 5921 |
+
"loss": 0.0518,
|
| 5922 |
+
"step": 9230
|
| 5923 |
+
},
|
| 5924 |
+
{
|
| 5925 |
+
"epoch": 39.83,
|
| 5926 |
+
"learning_rate": 1.2958963282937364e-06,
|
| 5927 |
+
"loss": 0.0334,
|
| 5928 |
+
"step": 9240
|
| 5929 |
+
},
|
| 5930 |
+
{
|
| 5931 |
+
"epoch": 39.87,
|
| 5932 |
+
"learning_rate": 9.719222462203022e-07,
|
| 5933 |
+
"loss": 0.0257,
|
| 5934 |
+
"step": 9250
|
| 5935 |
+
},
|
| 5936 |
+
{
|
| 5937 |
+
"epoch": 39.91,
|
| 5938 |
+
"learning_rate": 6.479481641468682e-07,
|
| 5939 |
+
"loss": 0.055,
|
| 5940 |
+
"step": 9260
|
| 5941 |
+
},
|
| 5942 |
+
{
|
| 5943 |
+
"epoch": 39.95,
|
| 5944 |
+
"learning_rate": 3.239740820734341e-07,
|
| 5945 |
+
"loss": 0.0444,
|
| 5946 |
+
"step": 9270
|
| 5947 |
+
},
|
| 5948 |
+
{
|
| 5949 |
+
"epoch": 40.0,
|
| 5950 |
+
"learning_rate": 0.0,
|
| 5951 |
+
"loss": 0.0732,
|
| 5952 |
+
"step": 9280
|
| 5953 |
+
},
|
| 5954 |
+
{
|
| 5955 |
+
"epoch": 40.0,
|
| 5956 |
+
"eval_loss": 0.31733959913253784,
|
| 5957 |
+
"eval_runtime": 284.3146,
|
| 5958 |
+
"eval_samples_per_second": 10.706,
|
| 5959 |
+
"eval_wer": 0.3336529726456041,
|
| 5960 |
+
"step": 9280
|
| 5961 |
+
},
|
| 5962 |
+
{
|
| 5963 |
+
"epoch": 40.0,
|
| 5964 |
+
"step": 9280,
|
| 5965 |
+
"total_flos": 5.602791262432019e+19,
|
| 5966 |
+
"train_runtime": 9410.6193,
|
| 5967 |
+
"train_samples_per_second": 0.986
|
| 5968 |
}
|
| 5969 |
],
|
| 5970 |
+
"max_steps": 9280,
|
| 5971 |
+
"num_train_epochs": 40,
|
| 5972 |
+
"total_flos": 5.602791262432019e+19,
|
| 5973 |
"trial_name": null,
|
| 5974 |
"trial_params": null
|
| 5975 |
}
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 2351
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bfe3500aa78573c5ffcb7645c8b988c8aff05806fb8cbd1ecaaf65db5f3c2601
|
| 3 |
size 2351
|