Upload 9 files
Browse files- checkpoint-4000/config.json +41 -0
- checkpoint-4000/optimizer.pt +3 -0
- checkpoint-4000/preprocessor_config.json +0 -0
- checkpoint-4000/pytorch_model.bin +3 -0
- checkpoint-4000/rng_state.pth +3 -0
- checkpoint-4000/scaler.pt +3 -0
- checkpoint-4000/scheduler.pt +3 -0
- checkpoint-4000/trainer_state.json +1012 -0
- checkpoint-4000/training_args.bin +3 -0
checkpoint-4000/config.json
ADDED
|
@@ -0,0 +1,41 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "openai/whisper-large-v2",
|
| 3 |
+
"activation_dropout": 0.0,
|
| 4 |
+
"activation_function": "gelu",
|
| 5 |
+
"architectures": [
|
| 6 |
+
"WhisperForConditionalGeneration"
|
| 7 |
+
],
|
| 8 |
+
"attention_dropout": 0.0,
|
| 9 |
+
"begin_suppress_tokens": [
|
| 10 |
+
220,
|
| 11 |
+
50257
|
| 12 |
+
],
|
| 13 |
+
"bos_token_id": 50257,
|
| 14 |
+
"d_model": 1280,
|
| 15 |
+
"decoder_attention_heads": 20,
|
| 16 |
+
"decoder_ffn_dim": 5120,
|
| 17 |
+
"decoder_layerdrop": 0.0,
|
| 18 |
+
"decoder_layers": 32,
|
| 19 |
+
"decoder_start_token_id": 50258,
|
| 20 |
+
"dropout": 0.1,
|
| 21 |
+
"encoder_attention_heads": 20,
|
| 22 |
+
"encoder_ffn_dim": 5120,
|
| 23 |
+
"encoder_layerdrop": 0.0,
|
| 24 |
+
"encoder_layers": 32,
|
| 25 |
+
"eos_token_id": 50257,
|
| 26 |
+
"forced_decoder_ids": null,
|
| 27 |
+
"init_std": 0.02,
|
| 28 |
+
"is_encoder_decoder": true,
|
| 29 |
+
"max_length": 448,
|
| 30 |
+
"max_source_positions": 1500,
|
| 31 |
+
"max_target_positions": 448,
|
| 32 |
+
"model_type": "whisper",
|
| 33 |
+
"num_hidden_layers": 32,
|
| 34 |
+
"num_mel_bins": 80,
|
| 35 |
+
"pad_token_id": 50257,
|
| 36 |
+
"scale_embedding": false,
|
| 37 |
+
"torch_dtype": "float32",
|
| 38 |
+
"transformers_version": "4.26.0.dev0",
|
| 39 |
+
"use_cache": false,
|
| 40 |
+
"vocab_size": 51865
|
| 41 |
+
}
|
checkpoint-4000/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0a0d152322c8d2f7150972194de14d9542dfc93a585df317e06848212902c648
|
| 3 |
+
size 12347192855
|
checkpoint-4000/preprocessor_config.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-4000/pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0cdf128692bb55e1e0a36f10d59e31883dc371afcfc15548c79ca8bc7d9a2a93
|
| 3 |
+
size 6173655480
|
checkpoint-4000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d1c04ef3a269adc92bdd7de05298dde08c688a5713a9396193e1a3cbb165f103
|
| 3 |
+
size 14575
|
checkpoint-4000/scaler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:295e39083f91367d10a36e5cc3c25f93cb1a8c52daa0be8993d95483a70b2148
|
| 3 |
+
size 557
|
checkpoint-4000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f8a2b31d699f5e3d503cf4e6d3bec1fda0d53c5e263b7d33234455db9ccd5ba0
|
| 3 |
+
size 627
|
checkpoint-4000/trainer_state.json
ADDED
|
@@ -0,0 +1,1012 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 9.899702823179792,
|
| 3 |
+
"best_model_checkpoint": "./data/finetuningRuns/whisper-lg-el-intlv-xs/checkpoint-4000",
|
| 4 |
+
"epoch": 9.950248756218905,
|
| 5 |
+
"global_step": 4000,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.06,
|
| 12 |
+
"learning_rate": 1.61e-07,
|
| 13 |
+
"loss": 0.5163,
|
| 14 |
+
"step": 25
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.12,
|
| 18 |
+
"learning_rate": 3.36e-07,
|
| 19 |
+
"loss": 0.4817,
|
| 20 |
+
"step": 50
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.19,
|
| 24 |
+
"learning_rate": 5.11e-07,
|
| 25 |
+
"loss": 0.3656,
|
| 26 |
+
"step": 75
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.25,
|
| 30 |
+
"learning_rate": 6.86e-07,
|
| 31 |
+
"loss": 0.2421,
|
| 32 |
+
"step": 100
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.31,
|
| 36 |
+
"learning_rate": 8.61e-07,
|
| 37 |
+
"loss": 0.2035,
|
| 38 |
+
"step": 125
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.37,
|
| 42 |
+
"learning_rate": 1.036e-06,
|
| 43 |
+
"loss": 0.1765,
|
| 44 |
+
"step": 150
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.44,
|
| 48 |
+
"learning_rate": 1.211e-06,
|
| 49 |
+
"loss": 0.1664,
|
| 50 |
+
"step": 175
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.5,
|
| 54 |
+
"learning_rate": 1.379e-06,
|
| 55 |
+
"loss": 0.1451,
|
| 56 |
+
"step": 200
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.56,
|
| 60 |
+
"learning_rate": 1.554e-06,
|
| 61 |
+
"loss": 0.1496,
|
| 62 |
+
"step": 225
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.62,
|
| 66 |
+
"learning_rate": 1.7289999999999999e-06,
|
| 67 |
+
"loss": 0.1385,
|
| 68 |
+
"step": 250
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.68,
|
| 72 |
+
"learning_rate": 1.9040000000000001e-06,
|
| 73 |
+
"loss": 0.1285,
|
| 74 |
+
"step": 275
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.75,
|
| 78 |
+
"learning_rate": 2.079e-06,
|
| 79 |
+
"loss": 0.1227,
|
| 80 |
+
"step": 300
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.81,
|
| 84 |
+
"learning_rate": 2.254e-06,
|
| 85 |
+
"loss": 0.1184,
|
| 86 |
+
"step": 325
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.87,
|
| 90 |
+
"learning_rate": 2.429e-06,
|
| 91 |
+
"loss": 0.113,
|
| 92 |
+
"step": 350
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.93,
|
| 96 |
+
"learning_rate": 2.604e-06,
|
| 97 |
+
"loss": 0.1087,
|
| 98 |
+
"step": 375
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 1.0,
|
| 102 |
+
"learning_rate": 2.779e-06,
|
| 103 |
+
"loss": 0.1128,
|
| 104 |
+
"step": 400
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 1.06,
|
| 108 |
+
"learning_rate": 2.954e-06,
|
| 109 |
+
"loss": 0.0815,
|
| 110 |
+
"step": 425
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 1.12,
|
| 114 |
+
"learning_rate": 3.129e-06,
|
| 115 |
+
"loss": 0.0883,
|
| 116 |
+
"step": 450
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 1.18,
|
| 120 |
+
"learning_rate": 3.3039999999999996e-06,
|
| 121 |
+
"loss": 0.0762,
|
| 122 |
+
"step": 475
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 1.24,
|
| 126 |
+
"learning_rate": 3.479e-06,
|
| 127 |
+
"loss": 0.0804,
|
| 128 |
+
"step": 500
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 1.31,
|
| 132 |
+
"learning_rate": 3.4918947368421053e-06,
|
| 133 |
+
"loss": 0.0724,
|
| 134 |
+
"step": 525
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 1.37,
|
| 138 |
+
"learning_rate": 3.482684210526316e-06,
|
| 139 |
+
"loss": 0.0726,
|
| 140 |
+
"step": 550
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 1.43,
|
| 144 |
+
"learning_rate": 3.4734736842105263e-06,
|
| 145 |
+
"loss": 0.0759,
|
| 146 |
+
"step": 575
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 1.49,
|
| 150 |
+
"learning_rate": 3.464263157894737e-06,
|
| 151 |
+
"loss": 0.0809,
|
| 152 |
+
"step": 600
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 1.55,
|
| 156 |
+
"learning_rate": 3.4550526315789474e-06,
|
| 157 |
+
"loss": 0.0745,
|
| 158 |
+
"step": 625
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 1.62,
|
| 162 |
+
"learning_rate": 3.4458421052631577e-06,
|
| 163 |
+
"loss": 0.0671,
|
| 164 |
+
"step": 650
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 1.68,
|
| 168 |
+
"learning_rate": 3.4366315789473685e-06,
|
| 169 |
+
"loss": 0.0679,
|
| 170 |
+
"step": 675
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 1.74,
|
| 174 |
+
"learning_rate": 3.427421052631579e-06,
|
| 175 |
+
"loss": 0.0728,
|
| 176 |
+
"step": 700
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 1.8,
|
| 180 |
+
"learning_rate": 3.4182105263157896e-06,
|
| 181 |
+
"loss": 0.061,
|
| 182 |
+
"step": 725
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 1.87,
|
| 186 |
+
"learning_rate": 3.409e-06,
|
| 187 |
+
"loss": 0.0619,
|
| 188 |
+
"step": 750
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 1.93,
|
| 192 |
+
"learning_rate": 3.39978947368421e-06,
|
| 193 |
+
"loss": 0.0672,
|
| 194 |
+
"step": 775
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 1.99,
|
| 198 |
+
"learning_rate": 3.3909473684210528e-06,
|
| 199 |
+
"loss": 0.0611,
|
| 200 |
+
"step": 800
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 2.05,
|
| 204 |
+
"learning_rate": 3.381736842105263e-06,
|
| 205 |
+
"loss": 0.0409,
|
| 206 |
+
"step": 825
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 2.11,
|
| 210 |
+
"learning_rate": 3.372526315789474e-06,
|
| 211 |
+
"loss": 0.032,
|
| 212 |
+
"step": 850
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 2.18,
|
| 216 |
+
"learning_rate": 3.363315789473684e-06,
|
| 217 |
+
"loss": 0.0308,
|
| 218 |
+
"step": 875
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 2.24,
|
| 222 |
+
"learning_rate": 3.354105263157895e-06,
|
| 223 |
+
"loss": 0.0318,
|
| 224 |
+
"step": 900
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 2.3,
|
| 228 |
+
"learning_rate": 3.3448947368421052e-06,
|
| 229 |
+
"loss": 0.035,
|
| 230 |
+
"step": 925
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 2.36,
|
| 234 |
+
"learning_rate": 3.3356842105263156e-06,
|
| 235 |
+
"loss": 0.0353,
|
| 236 |
+
"step": 950
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 2.43,
|
| 240 |
+
"learning_rate": 3.3264736842105263e-06,
|
| 241 |
+
"loss": 0.0301,
|
| 242 |
+
"step": 975
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 2.49,
|
| 246 |
+
"learning_rate": 3.3172631578947366e-06,
|
| 247 |
+
"loss": 0.0311,
|
| 248 |
+
"step": 1000
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 2.49,
|
| 252 |
+
"eval_loss": 0.1808706372976303,
|
| 253 |
+
"eval_runtime": 1982.2643,
|
| 254 |
+
"eval_samples_per_second": 0.856,
|
| 255 |
+
"eval_steps_per_second": 0.107,
|
| 256 |
+
"eval_wer": 10.549777117384844,
|
| 257 |
+
"step": 1000
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 2.55,
|
| 261 |
+
"learning_rate": 3.3080526315789474e-06,
|
| 262 |
+
"loss": 0.0309,
|
| 263 |
+
"step": 1025
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 2.61,
|
| 267 |
+
"learning_rate": 3.2988421052631577e-06,
|
| 268 |
+
"loss": 0.0364,
|
| 269 |
+
"step": 1050
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 2.67,
|
| 273 |
+
"learning_rate": 3.2896315789473685e-06,
|
| 274 |
+
"loss": 0.0323,
|
| 275 |
+
"step": 1075
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 2.74,
|
| 279 |
+
"learning_rate": 3.2804210526315788e-06,
|
| 280 |
+
"loss": 0.0322,
|
| 281 |
+
"step": 1100
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 2.8,
|
| 285 |
+
"learning_rate": 3.271210526315789e-06,
|
| 286 |
+
"loss": 0.0304,
|
| 287 |
+
"step": 1125
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 2.86,
|
| 291 |
+
"learning_rate": 3.2620000000000003e-06,
|
| 292 |
+
"loss": 0.0299,
|
| 293 |
+
"step": 1150
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 2.92,
|
| 297 |
+
"learning_rate": 3.2527894736842106e-06,
|
| 298 |
+
"loss": 0.0272,
|
| 299 |
+
"step": 1175
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 2.99,
|
| 303 |
+
"learning_rate": 3.243578947368421e-06,
|
| 304 |
+
"loss": 0.0241,
|
| 305 |
+
"step": 1200
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 3.05,
|
| 309 |
+
"learning_rate": 3.2343684210526317e-06,
|
| 310 |
+
"loss": 0.0186,
|
| 311 |
+
"step": 1225
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 3.11,
|
| 315 |
+
"learning_rate": 3.225157894736842e-06,
|
| 316 |
+
"loss": 0.015,
|
| 317 |
+
"step": 1250
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 3.17,
|
| 321 |
+
"learning_rate": 3.2159473684210527e-06,
|
| 322 |
+
"loss": 0.0125,
|
| 323 |
+
"step": 1275
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 3.23,
|
| 327 |
+
"learning_rate": 3.206736842105263e-06,
|
| 328 |
+
"loss": 0.0127,
|
| 329 |
+
"step": 1300
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 3.3,
|
| 333 |
+
"learning_rate": 3.197526315789474e-06,
|
| 334 |
+
"loss": 0.0144,
|
| 335 |
+
"step": 1325
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 3.36,
|
| 339 |
+
"learning_rate": 3.188315789473684e-06,
|
| 340 |
+
"loss": 0.0148,
|
| 341 |
+
"step": 1350
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 3.42,
|
| 345 |
+
"learning_rate": 3.1791052631578945e-06,
|
| 346 |
+
"loss": 0.0122,
|
| 347 |
+
"step": 1375
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 3.48,
|
| 351 |
+
"learning_rate": 3.1698947368421052e-06,
|
| 352 |
+
"loss": 0.0124,
|
| 353 |
+
"step": 1400
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 3.54,
|
| 357 |
+
"learning_rate": 3.1606842105263155e-06,
|
| 358 |
+
"loss": 0.0165,
|
| 359 |
+
"step": 1425
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 3.61,
|
| 363 |
+
"learning_rate": 3.151842105263158e-06,
|
| 364 |
+
"loss": 0.014,
|
| 365 |
+
"step": 1450
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 3.67,
|
| 369 |
+
"learning_rate": 3.1426315789473684e-06,
|
| 370 |
+
"loss": 0.0138,
|
| 371 |
+
"step": 1475
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 3.73,
|
| 375 |
+
"learning_rate": 3.133421052631579e-06,
|
| 376 |
+
"loss": 0.0142,
|
| 377 |
+
"step": 1500
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 3.79,
|
| 381 |
+
"learning_rate": 3.1242105263157895e-06,
|
| 382 |
+
"loss": 0.0136,
|
| 383 |
+
"step": 1525
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 3.86,
|
| 387 |
+
"learning_rate": 3.115e-06,
|
| 388 |
+
"loss": 0.0123,
|
| 389 |
+
"step": 1550
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 3.92,
|
| 393 |
+
"learning_rate": 3.1057894736842106e-06,
|
| 394 |
+
"loss": 0.0135,
|
| 395 |
+
"step": 1575
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 3.98,
|
| 399 |
+
"learning_rate": 3.096578947368421e-06,
|
| 400 |
+
"loss": 0.0135,
|
| 401 |
+
"step": 1600
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 4.04,
|
| 405 |
+
"learning_rate": 3.0873684210526316e-06,
|
| 406 |
+
"loss": 0.0102,
|
| 407 |
+
"step": 1625
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 4.1,
|
| 411 |
+
"learning_rate": 3.078157894736842e-06,
|
| 412 |
+
"loss": 0.0061,
|
| 413 |
+
"step": 1650
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 4.17,
|
| 417 |
+
"learning_rate": 3.0689473684210523e-06,
|
| 418 |
+
"loss": 0.0064,
|
| 419 |
+
"step": 1675
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 4.23,
|
| 423 |
+
"learning_rate": 3.059736842105263e-06,
|
| 424 |
+
"loss": 0.0068,
|
| 425 |
+
"step": 1700
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 4.29,
|
| 429 |
+
"learning_rate": 3.0505263157894734e-06,
|
| 430 |
+
"loss": 0.0056,
|
| 431 |
+
"step": 1725
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 4.35,
|
| 435 |
+
"learning_rate": 3.041315789473684e-06,
|
| 436 |
+
"loss": 0.0062,
|
| 437 |
+
"step": 1750
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 4.42,
|
| 441 |
+
"learning_rate": 3.032105263157895e-06,
|
| 442 |
+
"loss": 0.0071,
|
| 443 |
+
"step": 1775
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 4.48,
|
| 447 |
+
"learning_rate": 3.022894736842105e-06,
|
| 448 |
+
"loss": 0.0061,
|
| 449 |
+
"step": 1800
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 4.54,
|
| 453 |
+
"learning_rate": 3.013684210526316e-06,
|
| 454 |
+
"loss": 0.0069,
|
| 455 |
+
"step": 1825
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 4.6,
|
| 459 |
+
"learning_rate": 3.0044736842105263e-06,
|
| 460 |
+
"loss": 0.0067,
|
| 461 |
+
"step": 1850
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 4.66,
|
| 465 |
+
"learning_rate": 2.995263157894737e-06,
|
| 466 |
+
"loss": 0.0055,
|
| 467 |
+
"step": 1875
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 4.73,
|
| 471 |
+
"learning_rate": 2.9860526315789473e-06,
|
| 472 |
+
"loss": 0.0067,
|
| 473 |
+
"step": 1900
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 4.79,
|
| 477 |
+
"learning_rate": 2.9768421052631577e-06,
|
| 478 |
+
"loss": 0.0061,
|
| 479 |
+
"step": 1925
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 4.85,
|
| 483 |
+
"learning_rate": 2.9676315789473684e-06,
|
| 484 |
+
"loss": 0.0074,
|
| 485 |
+
"step": 1950
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 4.91,
|
| 489 |
+
"learning_rate": 2.9584210526315787e-06,
|
| 490 |
+
"loss": 0.0087,
|
| 491 |
+
"step": 1975
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 4.98,
|
| 495 |
+
"learning_rate": 2.9492105263157895e-06,
|
| 496 |
+
"loss": 0.0074,
|
| 497 |
+
"step": 2000
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 4.98,
|
| 501 |
+
"eval_loss": 0.2470209300518036,
|
| 502 |
+
"eval_runtime": 1972.083,
|
| 503 |
+
"eval_samples_per_second": 0.86,
|
| 504 |
+
"eval_steps_per_second": 0.108,
|
| 505 |
+
"eval_wer": 10.280460624071322,
|
| 506 |
+
"step": 2000
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 5.04,
|
| 510 |
+
"learning_rate": 2.94e-06,
|
| 511 |
+
"loss": 0.0049,
|
| 512 |
+
"step": 2025
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 5.1,
|
| 516 |
+
"learning_rate": 2.9307894736842106e-06,
|
| 517 |
+
"loss": 0.0035,
|
| 518 |
+
"step": 2050
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 5.16,
|
| 522 |
+
"learning_rate": 2.921578947368421e-06,
|
| 523 |
+
"loss": 0.0045,
|
| 524 |
+
"step": 2075
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 5.22,
|
| 528 |
+
"learning_rate": 2.9123684210526316e-06,
|
| 529 |
+
"loss": 0.0038,
|
| 530 |
+
"step": 2100
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 5.29,
|
| 534 |
+
"learning_rate": 2.9031578947368424e-06,
|
| 535 |
+
"loss": 0.0033,
|
| 536 |
+
"step": 2125
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 5.35,
|
| 540 |
+
"learning_rate": 2.8939473684210527e-06,
|
| 541 |
+
"loss": 0.0036,
|
| 542 |
+
"step": 2150
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 5.41,
|
| 546 |
+
"learning_rate": 2.884736842105263e-06,
|
| 547 |
+
"loss": 0.0039,
|
| 548 |
+
"step": 2175
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 5.47,
|
| 552 |
+
"learning_rate": 2.8755263157894738e-06,
|
| 553 |
+
"loss": 0.0035,
|
| 554 |
+
"step": 2200
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 5.53,
|
| 558 |
+
"learning_rate": 2.866315789473684e-06,
|
| 559 |
+
"loss": 0.0038,
|
| 560 |
+
"step": 2225
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 5.6,
|
| 564 |
+
"learning_rate": 2.857105263157895e-06,
|
| 565 |
+
"loss": 0.0039,
|
| 566 |
+
"step": 2250
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 5.66,
|
| 570 |
+
"learning_rate": 2.847894736842105e-06,
|
| 571 |
+
"loss": 0.0036,
|
| 572 |
+
"step": 2275
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 5.72,
|
| 576 |
+
"learning_rate": 2.838684210526316e-06,
|
| 577 |
+
"loss": 0.0043,
|
| 578 |
+
"step": 2300
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 5.78,
|
| 582 |
+
"learning_rate": 2.8294736842105262e-06,
|
| 583 |
+
"loss": 0.0049,
|
| 584 |
+
"step": 2325
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 5.85,
|
| 588 |
+
"learning_rate": 2.8202631578947366e-06,
|
| 589 |
+
"loss": 0.0041,
|
| 590 |
+
"step": 2350
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 5.91,
|
| 594 |
+
"learning_rate": 2.8110526315789473e-06,
|
| 595 |
+
"loss": 0.004,
|
| 596 |
+
"step": 2375
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 5.97,
|
| 600 |
+
"learning_rate": 2.8018421052631576e-06,
|
| 601 |
+
"loss": 0.0036,
|
| 602 |
+
"step": 2400
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 6.03,
|
| 606 |
+
"learning_rate": 2.7926315789473684e-06,
|
| 607 |
+
"loss": 0.003,
|
| 608 |
+
"step": 2425
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 6.09,
|
| 612 |
+
"learning_rate": 2.783421052631579e-06,
|
| 613 |
+
"loss": 0.0026,
|
| 614 |
+
"step": 2450
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 6.16,
|
| 618 |
+
"learning_rate": 2.7742105263157895e-06,
|
| 619 |
+
"loss": 0.0019,
|
| 620 |
+
"step": 2475
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 6.22,
|
| 624 |
+
"learning_rate": 2.765e-06,
|
| 625 |
+
"loss": 0.0023,
|
| 626 |
+
"step": 2500
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 6.28,
|
| 630 |
+
"learning_rate": 2.7557894736842105e-06,
|
| 631 |
+
"loss": 0.0029,
|
| 632 |
+
"step": 2525
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 6.34,
|
| 636 |
+
"learning_rate": 2.7465789473684213e-06,
|
| 637 |
+
"loss": 0.0032,
|
| 638 |
+
"step": 2550
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 6.41,
|
| 642 |
+
"learning_rate": 2.7373684210526316e-06,
|
| 643 |
+
"loss": 0.0028,
|
| 644 |
+
"step": 2575
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 6.47,
|
| 648 |
+
"learning_rate": 2.728157894736842e-06,
|
| 649 |
+
"loss": 0.0025,
|
| 650 |
+
"step": 2600
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 6.53,
|
| 654 |
+
"learning_rate": 2.7189473684210527e-06,
|
| 655 |
+
"loss": 0.0028,
|
| 656 |
+
"step": 2625
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 6.59,
|
| 660 |
+
"learning_rate": 2.709736842105263e-06,
|
| 661 |
+
"loss": 0.0025,
|
| 662 |
+
"step": 2650
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 6.65,
|
| 666 |
+
"learning_rate": 2.7005263157894738e-06,
|
| 667 |
+
"loss": 0.0024,
|
| 668 |
+
"step": 2675
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 6.72,
|
| 672 |
+
"learning_rate": 2.691315789473684e-06,
|
| 673 |
+
"loss": 0.0032,
|
| 674 |
+
"step": 2700
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 6.78,
|
| 678 |
+
"learning_rate": 2.6821052631578944e-06,
|
| 679 |
+
"loss": 0.0031,
|
| 680 |
+
"step": 2725
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 6.84,
|
| 684 |
+
"learning_rate": 2.672894736842105e-06,
|
| 685 |
+
"loss": 0.0032,
|
| 686 |
+
"step": 2750
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 6.9,
|
| 690 |
+
"learning_rate": 2.6636842105263155e-06,
|
| 691 |
+
"loss": 0.0034,
|
| 692 |
+
"step": 2775
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 6.97,
|
| 696 |
+
"learning_rate": 2.6544736842105266e-06,
|
| 697 |
+
"loss": 0.0026,
|
| 698 |
+
"step": 2800
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 7.03,
|
| 702 |
+
"learning_rate": 2.645263157894737e-06,
|
| 703 |
+
"loss": 0.003,
|
| 704 |
+
"step": 2825
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 7.09,
|
| 708 |
+
"learning_rate": 2.6360526315789473e-06,
|
| 709 |
+
"loss": 0.003,
|
| 710 |
+
"step": 2850
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 7.15,
|
| 714 |
+
"learning_rate": 2.626842105263158e-06,
|
| 715 |
+
"loss": 0.0023,
|
| 716 |
+
"step": 2875
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 7.21,
|
| 720 |
+
"learning_rate": 2.6176315789473684e-06,
|
| 721 |
+
"loss": 0.0024,
|
| 722 |
+
"step": 2900
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 7.28,
|
| 726 |
+
"learning_rate": 2.608421052631579e-06,
|
| 727 |
+
"loss": 0.0019,
|
| 728 |
+
"step": 2925
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 7.34,
|
| 732 |
+
"learning_rate": 2.5992105263157894e-06,
|
| 733 |
+
"loss": 0.0021,
|
| 734 |
+
"step": 2950
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 7.4,
|
| 738 |
+
"learning_rate": 2.5899999999999998e-06,
|
| 739 |
+
"loss": 0.0019,
|
| 740 |
+
"step": 2975
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 7.46,
|
| 744 |
+
"learning_rate": 2.5807894736842105e-06,
|
| 745 |
+
"loss": 0.0019,
|
| 746 |
+
"step": 3000
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 7.46,
|
| 750 |
+
"eval_loss": 0.30083394050598145,
|
| 751 |
+
"eval_runtime": 1980.8887,
|
| 752 |
+
"eval_samples_per_second": 0.856,
|
| 753 |
+
"eval_steps_per_second": 0.107,
|
| 754 |
+
"eval_wer": 10.029717682020802,
|
| 755 |
+
"step": 3000
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 7.52,
|
| 759 |
+
"learning_rate": 2.571578947368421e-06,
|
| 760 |
+
"loss": 0.0022,
|
| 761 |
+
"step": 3025
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 7.59,
|
| 765 |
+
"learning_rate": 2.5623684210526316e-06,
|
| 766 |
+
"loss": 0.0022,
|
| 767 |
+
"step": 3050
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 7.65,
|
| 771 |
+
"learning_rate": 2.553157894736842e-06,
|
| 772 |
+
"loss": 0.0021,
|
| 773 |
+
"step": 3075
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 7.71,
|
| 777 |
+
"learning_rate": 2.5439473684210522e-06,
|
| 778 |
+
"loss": 0.0027,
|
| 779 |
+
"step": 3100
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 7.77,
|
| 783 |
+
"learning_rate": 2.5347368421052634e-06,
|
| 784 |
+
"loss": 0.0023,
|
| 785 |
+
"step": 3125
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 7.84,
|
| 789 |
+
"learning_rate": 2.5255263157894737e-06,
|
| 790 |
+
"loss": 0.0026,
|
| 791 |
+
"step": 3150
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 7.9,
|
| 795 |
+
"learning_rate": 2.5163157894736845e-06,
|
| 796 |
+
"loss": 0.0022,
|
| 797 |
+
"step": 3175
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 7.96,
|
| 801 |
+
"learning_rate": 2.507105263157895e-06,
|
| 802 |
+
"loss": 0.0023,
|
| 803 |
+
"step": 3200
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 8.02,
|
| 807 |
+
"learning_rate": 2.497894736842105e-06,
|
| 808 |
+
"loss": 0.0023,
|
| 809 |
+
"step": 3225
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 8.08,
|
| 813 |
+
"learning_rate": 2.488684210526316e-06,
|
| 814 |
+
"loss": 0.0018,
|
| 815 |
+
"step": 3250
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 8.15,
|
| 819 |
+
"learning_rate": 2.479473684210526e-06,
|
| 820 |
+
"loss": 0.0015,
|
| 821 |
+
"step": 3275
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 8.21,
|
| 825 |
+
"learning_rate": 2.470263157894737e-06,
|
| 826 |
+
"loss": 0.0016,
|
| 827 |
+
"step": 3300
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 8.27,
|
| 831 |
+
"learning_rate": 2.4610526315789473e-06,
|
| 832 |
+
"loss": 0.0016,
|
| 833 |
+
"step": 3325
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 8.33,
|
| 837 |
+
"learning_rate": 2.4518421052631576e-06,
|
| 838 |
+
"loss": 0.0016,
|
| 839 |
+
"step": 3350
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 8.4,
|
| 843 |
+
"learning_rate": 2.4426315789473684e-06,
|
| 844 |
+
"loss": 0.0014,
|
| 845 |
+
"step": 3375
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 8.46,
|
| 849 |
+
"learning_rate": 2.4334210526315787e-06,
|
| 850 |
+
"loss": 0.0015,
|
| 851 |
+
"step": 3400
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 8.52,
|
| 855 |
+
"learning_rate": 2.4242105263157894e-06,
|
| 856 |
+
"loss": 0.0016,
|
| 857 |
+
"step": 3425
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 8.58,
|
| 861 |
+
"learning_rate": 2.4149999999999997e-06,
|
| 862 |
+
"loss": 0.0019,
|
| 863 |
+
"step": 3450
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 8.64,
|
| 867 |
+
"learning_rate": 2.4057894736842105e-06,
|
| 868 |
+
"loss": 0.0015,
|
| 869 |
+
"step": 3475
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 8.71,
|
| 873 |
+
"learning_rate": 2.3965789473684212e-06,
|
| 874 |
+
"loss": 0.0022,
|
| 875 |
+
"step": 3500
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 8.77,
|
| 879 |
+
"learning_rate": 2.3873684210526316e-06,
|
| 880 |
+
"loss": 0.0018,
|
| 881 |
+
"step": 3525
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 8.83,
|
| 885 |
+
"learning_rate": 2.3781578947368423e-06,
|
| 886 |
+
"loss": 0.0027,
|
| 887 |
+
"step": 3550
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"epoch": 8.89,
|
| 891 |
+
"learning_rate": 2.3689473684210526e-06,
|
| 892 |
+
"loss": 0.0016,
|
| 893 |
+
"step": 3575
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 8.96,
|
| 897 |
+
"learning_rate": 2.359736842105263e-06,
|
| 898 |
+
"loss": 0.0016,
|
| 899 |
+
"step": 3600
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 9.02,
|
| 903 |
+
"learning_rate": 2.3505263157894737e-06,
|
| 904 |
+
"loss": 0.0015,
|
| 905 |
+
"step": 3625
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 9.08,
|
| 909 |
+
"learning_rate": 2.341315789473684e-06,
|
| 910 |
+
"loss": 0.001,
|
| 911 |
+
"step": 3650
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 9.14,
|
| 915 |
+
"learning_rate": 2.3321052631578948e-06,
|
| 916 |
+
"loss": 0.0011,
|
| 917 |
+
"step": 3675
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 9.2,
|
| 921 |
+
"learning_rate": 2.322894736842105e-06,
|
| 922 |
+
"loss": 0.0011,
|
| 923 |
+
"step": 3700
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 9.27,
|
| 927 |
+
"learning_rate": 2.3136842105263154e-06,
|
| 928 |
+
"loss": 0.0012,
|
| 929 |
+
"step": 3725
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 9.33,
|
| 933 |
+
"learning_rate": 2.304473684210526e-06,
|
| 934 |
+
"loss": 0.0012,
|
| 935 |
+
"step": 3750
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 9.39,
|
| 939 |
+
"learning_rate": 2.2952631578947365e-06,
|
| 940 |
+
"loss": 0.0012,
|
| 941 |
+
"step": 3775
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 9.45,
|
| 945 |
+
"learning_rate": 2.2860526315789477e-06,
|
| 946 |
+
"loss": 0.0011,
|
| 947 |
+
"step": 3800
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 9.51,
|
| 951 |
+
"learning_rate": 2.276842105263158e-06,
|
| 952 |
+
"loss": 0.0009,
|
| 953 |
+
"step": 3825
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 9.58,
|
| 957 |
+
"learning_rate": 2.2676315789473683e-06,
|
| 958 |
+
"loss": 0.0009,
|
| 959 |
+
"step": 3850
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 9.64,
|
| 963 |
+
"learning_rate": 2.258421052631579e-06,
|
| 964 |
+
"loss": 0.0014,
|
| 965 |
+
"step": 3875
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 9.7,
|
| 969 |
+
"learning_rate": 2.2492105263157894e-06,
|
| 970 |
+
"loss": 0.0008,
|
| 971 |
+
"step": 3900
|
| 972 |
+
},
|
| 973 |
+
{
|
| 974 |
+
"epoch": 9.76,
|
| 975 |
+
"learning_rate": 2.24e-06,
|
| 976 |
+
"loss": 0.0015,
|
| 977 |
+
"step": 3925
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 9.83,
|
| 981 |
+
"learning_rate": 2.2307894736842105e-06,
|
| 982 |
+
"loss": 0.0009,
|
| 983 |
+
"step": 3950
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 9.89,
|
| 987 |
+
"learning_rate": 2.221578947368421e-06,
|
| 988 |
+
"loss": 0.0012,
|
| 989 |
+
"step": 3975
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 9.95,
|
| 993 |
+
"learning_rate": 2.2123684210526316e-06,
|
| 994 |
+
"loss": 0.0011,
|
| 995 |
+
"step": 4000
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 9.95,
|
| 999 |
+
"eval_loss": 0.29131078720092773,
|
| 1000 |
+
"eval_runtime": 1980.7982,
|
| 1001 |
+
"eval_samples_per_second": 0.856,
|
| 1002 |
+
"eval_steps_per_second": 0.107,
|
| 1003 |
+
"eval_wer": 9.899702823179792,
|
| 1004 |
+
"step": 4000
|
| 1005 |
+
}
|
| 1006 |
+
],
|
| 1007 |
+
"max_steps": 10000,
|
| 1008 |
+
"num_train_epochs": 25,
|
| 1009 |
+
"total_flos": 2.716512394272768e+20,
|
| 1010 |
+
"trial_name": null,
|
| 1011 |
+
"trial_params": null
|
| 1012 |
+
}
|
checkpoint-4000/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9e8126c9bd977f023d6d9036acfa695a88d0f035afe6d272bc812673284fa236
|
| 3 |
+
size 3707
|