Upload folder using huggingface_hub
#6
by
shehzad1
- opened
- checkpoint-3500/config.json +51 -0
- checkpoint-3500/generation_config.json +14 -0
- checkpoint-3500/optimizer.pt +3 -0
- checkpoint-3500/preprocessor_config.json +14 -0
- checkpoint-3500/pytorch_model.bin +3 -0
- checkpoint-3500/rng_state.pth +3 -0
- checkpoint-3500/scaler.pt +3 -0
- checkpoint-3500/scheduler.pt +3 -0
- checkpoint-3500/trainer_state.json +892 -0
- checkpoint-3500/training_args.bin +3 -0
- runs/Sep18_09-16-12_14ab48e85f35/events.out.tfevents.1695029544.14ab48e85f35.24.0 +2 -2
checkpoint-3500/config.json
ADDED
|
@@ -0,0 +1,51 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "shehzad1/whisper_urdu_small",
|
| 3 |
+
"activation_dropout": 0.0,
|
| 4 |
+
"activation_function": "gelu",
|
| 5 |
+
"apply_spec_augment": false,
|
| 6 |
+
"architectures": [
|
| 7 |
+
"WhisperForConditionalGeneration"
|
| 8 |
+
],
|
| 9 |
+
"attention_dropout": 0.0,
|
| 10 |
+
"begin_suppress_tokens": [
|
| 11 |
+
220,
|
| 12 |
+
50257
|
| 13 |
+
],
|
| 14 |
+
"bos_token_id": 50257,
|
| 15 |
+
"classifier_proj_size": 256,
|
| 16 |
+
"d_model": 768,
|
| 17 |
+
"decoder_attention_heads": 12,
|
| 18 |
+
"decoder_ffn_dim": 3072,
|
| 19 |
+
"decoder_layerdrop": 0.0,
|
| 20 |
+
"decoder_layers": 12,
|
| 21 |
+
"decoder_start_token_id": 50258,
|
| 22 |
+
"dropout": 0.0,
|
| 23 |
+
"encoder_attention_heads": 12,
|
| 24 |
+
"encoder_ffn_dim": 3072,
|
| 25 |
+
"encoder_layerdrop": 0.0,
|
| 26 |
+
"encoder_layers": 12,
|
| 27 |
+
"eos_token_id": 50257,
|
| 28 |
+
"forced_decoder_ids": null,
|
| 29 |
+
"init_std": 0.02,
|
| 30 |
+
"is_encoder_decoder": true,
|
| 31 |
+
"mask_feature_length": 10,
|
| 32 |
+
"mask_feature_min_masks": 0,
|
| 33 |
+
"mask_feature_prob": 0.0,
|
| 34 |
+
"mask_time_length": 10,
|
| 35 |
+
"mask_time_min_masks": 2,
|
| 36 |
+
"mask_time_prob": 0.05,
|
| 37 |
+
"max_length": 448,
|
| 38 |
+
"max_source_positions": 1500,
|
| 39 |
+
"max_target_positions": 448,
|
| 40 |
+
"model_type": "whisper",
|
| 41 |
+
"num_hidden_layers": 12,
|
| 42 |
+
"num_mel_bins": 80,
|
| 43 |
+
"pad_token_id": 50257,
|
| 44 |
+
"scale_embedding": false,
|
| 45 |
+
"suppress_tokens": [],
|
| 46 |
+
"torch_dtype": "float32",
|
| 47 |
+
"transformers_version": "4.28.1",
|
| 48 |
+
"use_cache": true,
|
| 49 |
+
"use_weighted_layer_sum": false,
|
| 50 |
+
"vocab_size": 51865
|
| 51 |
+
}
|
checkpoint-3500/generation_config.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"begin_suppress_tokens": [
|
| 3 |
+
220,
|
| 4 |
+
50257
|
| 5 |
+
],
|
| 6 |
+
"bos_token_id": 50257,
|
| 7 |
+
"decoder_start_token_id": 50258,
|
| 8 |
+
"eos_token_id": 50257,
|
| 9 |
+
"max_length": 448,
|
| 10 |
+
"pad_token_id": 50257,
|
| 11 |
+
"return_timestamps": false,
|
| 12 |
+
"suppress_tokens": [],
|
| 13 |
+
"transformers_version": "4.28.1"
|
| 14 |
+
}
|
checkpoint-3500/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2e3f215a2546779deaeff1f9423bf751e175dd60fd5e8dc583b6f20610cbd30d
|
| 3 |
+
size 1934161093
|
checkpoint-3500/preprocessor_config.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"chunk_length": 30,
|
| 3 |
+
"feature_extractor_type": "WhisperFeatureExtractor",
|
| 4 |
+
"feature_size": 80,
|
| 5 |
+
"hop_length": 160,
|
| 6 |
+
"n_fft": 400,
|
| 7 |
+
"n_samples": 480000,
|
| 8 |
+
"nb_max_frames": 3000,
|
| 9 |
+
"padding_side": "right",
|
| 10 |
+
"padding_value": 0.0,
|
| 11 |
+
"processor_class": "WhisperProcessor",
|
| 12 |
+
"return_attention_mask": false,
|
| 13 |
+
"sampling_rate": 16000
|
| 14 |
+
}
|
checkpoint-3500/pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f4198f2423b6432258e6830322befbd3438cc079680f0f8788ebefe63a9bb0a1
|
| 3 |
+
size 967102601
|
checkpoint-3500/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3db9cc661a6ec71ad391f98800282da2b84ffb373e5ae41256721d0fe18a514a
|
| 3 |
+
size 15597
|
checkpoint-3500/scaler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dfa37875070a3541bbf025714c937240d3152e3279cde26c05dc02b7951ca78c
|
| 3 |
+
size 557
|
checkpoint-3500/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:29cbee5c7182bea2535b60beb56e0e6965f24466dfef2c7104462af9dcc6c02d
|
| 3 |
+
size 627
|
checkpoint-3500/trainer_state.json
ADDED
|
@@ -0,0 +1,892 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 33.334622524752476,
|
| 3 |
+
"best_model_checkpoint": "./whisper_urdu_small/checkpoint-3500",
|
| 4 |
+
"epoch": 54.58687258687259,
|
| 5 |
+
"global_step": 3500,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.05,
|
| 12 |
+
"learning_rate": 4.4e-07,
|
| 13 |
+
"loss": 3.2601,
|
| 14 |
+
"step": 25
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.11,
|
| 18 |
+
"learning_rate": 9.400000000000001e-07,
|
| 19 |
+
"loss": 2.5942,
|
| 20 |
+
"step": 50
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.16,
|
| 24 |
+
"learning_rate": 1.44e-06,
|
| 25 |
+
"loss": 1.8933,
|
| 26 |
+
"step": 75
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.22,
|
| 30 |
+
"learning_rate": 1.94e-06,
|
| 31 |
+
"loss": 1.3779,
|
| 32 |
+
"step": 100
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.27,
|
| 36 |
+
"learning_rate": 2.4400000000000004e-06,
|
| 37 |
+
"loss": 1.1441,
|
| 38 |
+
"step": 125
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.32,
|
| 42 |
+
"learning_rate": 2.9400000000000002e-06,
|
| 43 |
+
"loss": 1.0934,
|
| 44 |
+
"step": 150
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.38,
|
| 48 |
+
"learning_rate": 3.44e-06,
|
| 49 |
+
"loss": 1.0316,
|
| 50 |
+
"step": 175
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.43,
|
| 54 |
+
"learning_rate": 3.94e-06,
|
| 55 |
+
"loss": 0.9975,
|
| 56 |
+
"step": 200
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.48,
|
| 60 |
+
"learning_rate": 4.440000000000001e-06,
|
| 61 |
+
"loss": 0.9251,
|
| 62 |
+
"step": 225
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.54,
|
| 66 |
+
"learning_rate": 4.94e-06,
|
| 67 |
+
"loss": 0.8483,
|
| 68 |
+
"step": 250
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.59,
|
| 72 |
+
"learning_rate": 5.4400000000000004e-06,
|
| 73 |
+
"loss": 0.7772,
|
| 74 |
+
"step": 275
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.65,
|
| 78 |
+
"learning_rate": 5.94e-06,
|
| 79 |
+
"loss": 0.7149,
|
| 80 |
+
"step": 300
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.7,
|
| 84 |
+
"learning_rate": 6.440000000000001e-06,
|
| 85 |
+
"loss": 0.6152,
|
| 86 |
+
"step": 325
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.75,
|
| 90 |
+
"learning_rate": 6.9400000000000005e-06,
|
| 91 |
+
"loss": 0.5333,
|
| 92 |
+
"step": 350
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.81,
|
| 96 |
+
"learning_rate": 7.440000000000001e-06,
|
| 97 |
+
"loss": 0.4789,
|
| 98 |
+
"step": 375
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.86,
|
| 102 |
+
"learning_rate": 7.94e-06,
|
| 103 |
+
"loss": 0.4714,
|
| 104 |
+
"step": 400
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.91,
|
| 108 |
+
"learning_rate": 8.44e-06,
|
| 109 |
+
"loss": 0.4691,
|
| 110 |
+
"step": 425
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.97,
|
| 114 |
+
"learning_rate": 8.94e-06,
|
| 115 |
+
"loss": 0.4856,
|
| 116 |
+
"step": 450
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 1.02,
|
| 120 |
+
"learning_rate": 9.440000000000001e-06,
|
| 121 |
+
"loss": 0.4459,
|
| 122 |
+
"step": 475
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 1.08,
|
| 126 |
+
"learning_rate": 9.940000000000001e-06,
|
| 127 |
+
"loss": 0.399,
|
| 128 |
+
"step": 500
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 1.13,
|
| 132 |
+
"learning_rate": 9.937142857142858e-06,
|
| 133 |
+
"loss": 0.3577,
|
| 134 |
+
"step": 525
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 1.18,
|
| 138 |
+
"learning_rate": 9.865714285714285e-06,
|
| 139 |
+
"loss": 0.3689,
|
| 140 |
+
"step": 550
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 1.24,
|
| 144 |
+
"learning_rate": 9.794285714285714e-06,
|
| 145 |
+
"loss": 0.3339,
|
| 146 |
+
"step": 575
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 1.29,
|
| 150 |
+
"learning_rate": 9.722857142857143e-06,
|
| 151 |
+
"loss": 0.3705,
|
| 152 |
+
"step": 600
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 1.35,
|
| 156 |
+
"learning_rate": 9.651428571428572e-06,
|
| 157 |
+
"loss": 0.3889,
|
| 158 |
+
"step": 625
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 1.4,
|
| 162 |
+
"learning_rate": 9.58e-06,
|
| 163 |
+
"loss": 0.3537,
|
| 164 |
+
"step": 650
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 1.45,
|
| 168 |
+
"learning_rate": 9.508571428571429e-06,
|
| 169 |
+
"loss": 0.3247,
|
| 170 |
+
"step": 675
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 1.51,
|
| 174 |
+
"learning_rate": 9.437142857142858e-06,
|
| 175 |
+
"loss": 0.3638,
|
| 176 |
+
"step": 700
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 1.56,
|
| 180 |
+
"learning_rate": 9.365714285714287e-06,
|
| 181 |
+
"loss": 0.3339,
|
| 182 |
+
"step": 725
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 1.61,
|
| 186 |
+
"learning_rate": 9.294285714285714e-06,
|
| 187 |
+
"loss": 0.3547,
|
| 188 |
+
"step": 750
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 1.67,
|
| 192 |
+
"learning_rate": 9.222857142857143e-06,
|
| 193 |
+
"loss": 0.337,
|
| 194 |
+
"step": 775
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 1.72,
|
| 198 |
+
"learning_rate": 9.151428571428572e-06,
|
| 199 |
+
"loss": 0.3423,
|
| 200 |
+
"step": 800
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 1.78,
|
| 204 |
+
"learning_rate": 9.080000000000001e-06,
|
| 205 |
+
"loss": 0.3435,
|
| 206 |
+
"step": 825
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 1.83,
|
| 210 |
+
"learning_rate": 9.00857142857143e-06,
|
| 211 |
+
"loss": 0.3164,
|
| 212 |
+
"step": 850
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 1.88,
|
| 216 |
+
"learning_rate": 8.937142857142857e-06,
|
| 217 |
+
"loss": 0.3422,
|
| 218 |
+
"step": 875
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 1.94,
|
| 222 |
+
"learning_rate": 8.865714285714287e-06,
|
| 223 |
+
"loss": 0.3239,
|
| 224 |
+
"step": 900
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 1.99,
|
| 228 |
+
"learning_rate": 8.794285714285716e-06,
|
| 229 |
+
"loss": 0.35,
|
| 230 |
+
"step": 925
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 2.05,
|
| 234 |
+
"learning_rate": 8.722857142857145e-06,
|
| 235 |
+
"loss": 0.2176,
|
| 236 |
+
"step": 950
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 2.1,
|
| 240 |
+
"learning_rate": 8.651428571428572e-06,
|
| 241 |
+
"loss": 0.1848,
|
| 242 |
+
"step": 975
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 2.15,
|
| 246 |
+
"learning_rate": 8.580000000000001e-06,
|
| 247 |
+
"loss": 0.2028,
|
| 248 |
+
"step": 1000
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 2.15,
|
| 252 |
+
"eval_loss": 0.5113539695739746,
|
| 253 |
+
"eval_runtime": 2118.0914,
|
| 254 |
+
"eval_samples_per_second": 1.559,
|
| 255 |
+
"eval_steps_per_second": 0.195,
|
| 256 |
+
"eval_wer": 95.40616896938737,
|
| 257 |
+
"step": 1000
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 2.21,
|
| 261 |
+
"learning_rate": 8.50857142857143e-06,
|
| 262 |
+
"loss": 0.2092,
|
| 263 |
+
"step": 1025
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 2.26,
|
| 267 |
+
"learning_rate": 8.437142857142859e-06,
|
| 268 |
+
"loss": 0.1918,
|
| 269 |
+
"step": 1050
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 2.31,
|
| 273 |
+
"learning_rate": 8.365714285714286e-06,
|
| 274 |
+
"loss": 0.1733,
|
| 275 |
+
"step": 1075
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 2.37,
|
| 279 |
+
"learning_rate": 8.294285714285715e-06,
|
| 280 |
+
"loss": 0.1949,
|
| 281 |
+
"step": 1100
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 2.42,
|
| 285 |
+
"learning_rate": 8.222857142857144e-06,
|
| 286 |
+
"loss": 0.2219,
|
| 287 |
+
"step": 1125
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 2.48,
|
| 291 |
+
"learning_rate": 8.151428571428572e-06,
|
| 292 |
+
"loss": 0.1785,
|
| 293 |
+
"step": 1150
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 2.53,
|
| 297 |
+
"learning_rate": 8.08e-06,
|
| 298 |
+
"loss": 0.189,
|
| 299 |
+
"step": 1175
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 2.58,
|
| 303 |
+
"learning_rate": 8.00857142857143e-06,
|
| 304 |
+
"loss": 0.1662,
|
| 305 |
+
"step": 1200
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 2.64,
|
| 309 |
+
"learning_rate": 7.937142857142857e-06,
|
| 310 |
+
"loss": 0.2038,
|
| 311 |
+
"step": 1225
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 2.69,
|
| 315 |
+
"learning_rate": 7.865714285714286e-06,
|
| 316 |
+
"loss": 0.2204,
|
| 317 |
+
"step": 1250
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 2.74,
|
| 321 |
+
"learning_rate": 7.794285714285715e-06,
|
| 322 |
+
"loss": 0.1718,
|
| 323 |
+
"step": 1275
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 2.8,
|
| 327 |
+
"learning_rate": 7.722857142857142e-06,
|
| 328 |
+
"loss": 0.1748,
|
| 329 |
+
"step": 1300
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 2.85,
|
| 333 |
+
"learning_rate": 7.651428571428571e-06,
|
| 334 |
+
"loss": 0.1977,
|
| 335 |
+
"step": 1325
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 2.91,
|
| 339 |
+
"learning_rate": 7.58e-06,
|
| 340 |
+
"loss": 0.1849,
|
| 341 |
+
"step": 1350
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 2.96,
|
| 345 |
+
"learning_rate": 7.508571428571429e-06,
|
| 346 |
+
"loss": 0.1972,
|
| 347 |
+
"step": 1375
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 3.01,
|
| 351 |
+
"learning_rate": 7.4371428571428575e-06,
|
| 352 |
+
"loss": 0.1669,
|
| 353 |
+
"step": 1400
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 3.07,
|
| 357 |
+
"learning_rate": 7.365714285714286e-06,
|
| 358 |
+
"loss": 0.0902,
|
| 359 |
+
"step": 1425
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 3.12,
|
| 363 |
+
"learning_rate": 7.294285714285715e-06,
|
| 364 |
+
"loss": 0.099,
|
| 365 |
+
"step": 1450
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 3.18,
|
| 369 |
+
"learning_rate": 7.222857142857144e-06,
|
| 370 |
+
"loss": 0.0856,
|
| 371 |
+
"step": 1475
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 3.23,
|
| 375 |
+
"learning_rate": 7.151428571428573e-06,
|
| 376 |
+
"loss": 0.1049,
|
| 377 |
+
"step": 1500
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 3.28,
|
| 381 |
+
"learning_rate": 7.08e-06,
|
| 382 |
+
"loss": 0.1175,
|
| 383 |
+
"step": 1525
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 3.34,
|
| 387 |
+
"learning_rate": 7.008571428571429e-06,
|
| 388 |
+
"loss": 0.1178,
|
| 389 |
+
"step": 1550
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 3.39,
|
| 393 |
+
"learning_rate": 6.937142857142858e-06,
|
| 394 |
+
"loss": 0.1004,
|
| 395 |
+
"step": 1575
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 3.44,
|
| 399 |
+
"learning_rate": 6.865714285714287e-06,
|
| 400 |
+
"loss": 0.108,
|
| 401 |
+
"step": 1600
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 3.5,
|
| 405 |
+
"learning_rate": 6.794285714285714e-06,
|
| 406 |
+
"loss": 0.1232,
|
| 407 |
+
"step": 1625
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 3.55,
|
| 411 |
+
"learning_rate": 6.722857142857143e-06,
|
| 412 |
+
"loss": 0.1078,
|
| 413 |
+
"step": 1650
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 3.61,
|
| 417 |
+
"learning_rate": 6.651428571428572e-06,
|
| 418 |
+
"loss": 0.1024,
|
| 419 |
+
"step": 1675
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 3.66,
|
| 423 |
+
"learning_rate": 6.5800000000000005e-06,
|
| 424 |
+
"loss": 0.1032,
|
| 425 |
+
"step": 1700
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 3.71,
|
| 429 |
+
"learning_rate": 6.5085714285714295e-06,
|
| 430 |
+
"loss": 0.0919,
|
| 431 |
+
"step": 1725
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 3.77,
|
| 435 |
+
"learning_rate": 6.437142857142858e-06,
|
| 436 |
+
"loss": 0.114,
|
| 437 |
+
"step": 1750
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 3.82,
|
| 441 |
+
"learning_rate": 6.365714285714286e-06,
|
| 442 |
+
"loss": 0.1065,
|
| 443 |
+
"step": 1775
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 3.88,
|
| 447 |
+
"learning_rate": 6.294285714285715e-06,
|
| 448 |
+
"loss": 0.1102,
|
| 449 |
+
"step": 1800
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 3.93,
|
| 453 |
+
"learning_rate": 6.222857142857144e-06,
|
| 454 |
+
"loss": 0.1044,
|
| 455 |
+
"step": 1825
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 3.98,
|
| 459 |
+
"learning_rate": 6.151428571428571e-06,
|
| 460 |
+
"loss": 0.101,
|
| 461 |
+
"step": 1850
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 4.04,
|
| 465 |
+
"learning_rate": 6.08e-06,
|
| 466 |
+
"loss": 0.0729,
|
| 467 |
+
"step": 1875
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 4.09,
|
| 471 |
+
"learning_rate": 6.008571428571429e-06,
|
| 472 |
+
"loss": 0.058,
|
| 473 |
+
"step": 1900
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 4.14,
|
| 477 |
+
"learning_rate": 5.937142857142858e-06,
|
| 478 |
+
"loss": 0.0545,
|
| 479 |
+
"step": 1925
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 4.2,
|
| 483 |
+
"learning_rate": 5.865714285714286e-06,
|
| 484 |
+
"loss": 0.0568,
|
| 485 |
+
"step": 1950
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 4.25,
|
| 489 |
+
"learning_rate": 5.794285714285715e-06,
|
| 490 |
+
"loss": 0.0495,
|
| 491 |
+
"step": 1975
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 4.31,
|
| 495 |
+
"learning_rate": 5.722857142857144e-06,
|
| 496 |
+
"loss": 0.0459,
|
| 497 |
+
"step": 2000
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 4.31,
|
| 501 |
+
"eval_loss": 0.5495272278785706,
|
| 502 |
+
"eval_runtime": 2507.3865,
|
| 503 |
+
"eval_samples_per_second": 1.317,
|
| 504 |
+
"eval_steps_per_second": 0.165,
|
| 505 |
+
"eval_wer": 74.14373621270173,
|
| 506 |
+
"step": 2000
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 4.36,
|
| 510 |
+
"learning_rate": 5.651428571428572e-06,
|
| 511 |
+
"loss": 0.0566,
|
| 512 |
+
"step": 2025
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 4.41,
|
| 516 |
+
"learning_rate": 5.580000000000001e-06,
|
| 517 |
+
"loss": 0.0596,
|
| 518 |
+
"step": 2050
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 4.47,
|
| 522 |
+
"learning_rate": 5.508571428571429e-06,
|
| 523 |
+
"loss": 0.0554,
|
| 524 |
+
"step": 2075
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 4.52,
|
| 528 |
+
"learning_rate": 5.437142857142857e-06,
|
| 529 |
+
"loss": 0.0469,
|
| 530 |
+
"step": 2100
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 4.57,
|
| 534 |
+
"learning_rate": 5.365714285714286e-06,
|
| 535 |
+
"loss": 0.0548,
|
| 536 |
+
"step": 2125
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 4.63,
|
| 540 |
+
"learning_rate": 5.294285714285715e-06,
|
| 541 |
+
"loss": 0.0558,
|
| 542 |
+
"step": 2150
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 4.68,
|
| 546 |
+
"learning_rate": 5.2228571428571425e-06,
|
| 547 |
+
"loss": 0.0644,
|
| 548 |
+
"step": 2175
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 4.74,
|
| 552 |
+
"learning_rate": 5.1514285714285715e-06,
|
| 553 |
+
"loss": 0.0536,
|
| 554 |
+
"step": 2200
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 4.79,
|
| 558 |
+
"learning_rate": 5.0800000000000005e-06,
|
| 559 |
+
"loss": 0.0553,
|
| 560 |
+
"step": 2225
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 4.84,
|
| 564 |
+
"learning_rate": 5.0085714285714295e-06,
|
| 565 |
+
"loss": 0.0528,
|
| 566 |
+
"step": 2250
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 4.9,
|
| 570 |
+
"learning_rate": 4.937142857142858e-06,
|
| 571 |
+
"loss": 0.0685,
|
| 572 |
+
"step": 2275
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 4.95,
|
| 576 |
+
"learning_rate": 4.865714285714287e-06,
|
| 577 |
+
"loss": 0.0459,
|
| 578 |
+
"step": 2300
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 5.01,
|
| 582 |
+
"learning_rate": 4.794285714285715e-06,
|
| 583 |
+
"loss": 0.0524,
|
| 584 |
+
"step": 2325
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 5.06,
|
| 588 |
+
"learning_rate": 4.722857142857144e-06,
|
| 589 |
+
"loss": 0.0306,
|
| 590 |
+
"step": 2350
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 5.11,
|
| 594 |
+
"learning_rate": 4.651428571428572e-06,
|
| 595 |
+
"loss": 0.0265,
|
| 596 |
+
"step": 2375
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 5.17,
|
| 600 |
+
"learning_rate": 4.58e-06,
|
| 601 |
+
"loss": 0.0359,
|
| 602 |
+
"step": 2400
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 5.22,
|
| 606 |
+
"learning_rate": 4.508571428571429e-06,
|
| 607 |
+
"loss": 0.0234,
|
| 608 |
+
"step": 2425
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 5.27,
|
| 612 |
+
"learning_rate": 4.437142857142857e-06,
|
| 613 |
+
"loss": 0.0249,
|
| 614 |
+
"step": 2450
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 5.33,
|
| 618 |
+
"learning_rate": 4.3657142857142855e-06,
|
| 619 |
+
"loss": 0.0252,
|
| 620 |
+
"step": 2475
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 5.38,
|
| 624 |
+
"learning_rate": 4.2942857142857146e-06,
|
| 625 |
+
"loss": 0.0258,
|
| 626 |
+
"step": 2500
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 5.44,
|
| 630 |
+
"learning_rate": 4.222857142857143e-06,
|
| 631 |
+
"loss": 0.0226,
|
| 632 |
+
"step": 2525
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 5.49,
|
| 636 |
+
"learning_rate": 4.151428571428572e-06,
|
| 637 |
+
"loss": 0.0313,
|
| 638 |
+
"step": 2550
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 5.54,
|
| 642 |
+
"learning_rate": 4.08e-06,
|
| 643 |
+
"loss": 0.0223,
|
| 644 |
+
"step": 2575
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 5.6,
|
| 648 |
+
"learning_rate": 4.008571428571429e-06,
|
| 649 |
+
"loss": 0.0253,
|
| 650 |
+
"step": 2600
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 5.65,
|
| 654 |
+
"learning_rate": 3.937142857142858e-06,
|
| 655 |
+
"loss": 0.0205,
|
| 656 |
+
"step": 2625
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 5.71,
|
| 660 |
+
"learning_rate": 3.865714285714286e-06,
|
| 661 |
+
"loss": 0.0257,
|
| 662 |
+
"step": 2650
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 5.76,
|
| 666 |
+
"learning_rate": 3.7942857142857147e-06,
|
| 667 |
+
"loss": 0.0244,
|
| 668 |
+
"step": 2675
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 5.81,
|
| 672 |
+
"learning_rate": 3.722857142857143e-06,
|
| 673 |
+
"loss": 0.0293,
|
| 674 |
+
"step": 2700
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 5.87,
|
| 678 |
+
"learning_rate": 3.651428571428572e-06,
|
| 679 |
+
"loss": 0.0262,
|
| 680 |
+
"step": 2725
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 5.92,
|
| 684 |
+
"learning_rate": 3.58e-06,
|
| 685 |
+
"loss": 0.0257,
|
| 686 |
+
"step": 2750
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 5.97,
|
| 690 |
+
"learning_rate": 3.508571428571429e-06,
|
| 691 |
+
"loss": 0.0191,
|
| 692 |
+
"step": 2775
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 6.03,
|
| 696 |
+
"learning_rate": 3.437142857142857e-06,
|
| 697 |
+
"loss": 0.0134,
|
| 698 |
+
"step": 2800
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 6.08,
|
| 702 |
+
"learning_rate": 3.3657142857142862e-06,
|
| 703 |
+
"loss": 0.0129,
|
| 704 |
+
"step": 2825
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 6.14,
|
| 708 |
+
"learning_rate": 3.2942857142857144e-06,
|
| 709 |
+
"loss": 0.0124,
|
| 710 |
+
"step": 2850
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 6.19,
|
| 714 |
+
"learning_rate": 3.222857142857143e-06,
|
| 715 |
+
"loss": 0.0108,
|
| 716 |
+
"step": 2875
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 6.24,
|
| 720 |
+
"learning_rate": 3.151428571428572e-06,
|
| 721 |
+
"loss": 0.0161,
|
| 722 |
+
"step": 2900
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 6.3,
|
| 726 |
+
"learning_rate": 3.08e-06,
|
| 727 |
+
"loss": 0.01,
|
| 728 |
+
"step": 2925
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 6.35,
|
| 732 |
+
"learning_rate": 3.008571428571429e-06,
|
| 733 |
+
"loss": 0.0257,
|
| 734 |
+
"step": 2950
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 6.4,
|
| 738 |
+
"learning_rate": 2.9371428571428573e-06,
|
| 739 |
+
"loss": 0.0124,
|
| 740 |
+
"step": 2975
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 6.46,
|
| 744 |
+
"learning_rate": 2.865714285714286e-06,
|
| 745 |
+
"loss": 0.0112,
|
| 746 |
+
"step": 3000
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 6.46,
|
| 750 |
+
"eval_loss": 0.6222674250602722,
|
| 751 |
+
"eval_runtime": 2409.126,
|
| 752 |
+
"eval_samples_per_second": 1.371,
|
| 753 |
+
"eval_steps_per_second": 0.171,
|
| 754 |
+
"eval_wer": 68.96164712256666,
|
| 755 |
+
"step": 3000
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 47.25,
|
| 759 |
+
"learning_rate": 6.120000000000001e-06,
|
| 760 |
+
"loss": 0.0095,
|
| 761 |
+
"step": 3025
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 47.64,
|
| 765 |
+
"learning_rate": 6.081538461538462e-06,
|
| 766 |
+
"loss": 0.0121,
|
| 767 |
+
"step": 3050
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 48.02,
|
| 771 |
+
"learning_rate": 6.043076923076923e-06,
|
| 772 |
+
"loss": 0.0089,
|
| 773 |
+
"step": 3075
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 48.41,
|
| 777 |
+
"learning_rate": 6.0046153846153855e-06,
|
| 778 |
+
"loss": 0.0057,
|
| 779 |
+
"step": 3100
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 48.8,
|
| 783 |
+
"learning_rate": 5.966153846153847e-06,
|
| 784 |
+
"loss": 0.0044,
|
| 785 |
+
"step": 3125
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 49.18,
|
| 789 |
+
"learning_rate": 5.927692307692307e-06,
|
| 790 |
+
"loss": 0.0053,
|
| 791 |
+
"step": 3150
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 49.57,
|
| 795 |
+
"learning_rate": 5.8892307692307695e-06,
|
| 796 |
+
"loss": 0.0036,
|
| 797 |
+
"step": 3175
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 49.95,
|
| 801 |
+
"learning_rate": 5.850769230769231e-06,
|
| 802 |
+
"loss": 0.0033,
|
| 803 |
+
"step": 3200
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 50.34,
|
| 807 |
+
"learning_rate": 5.812307692307693e-06,
|
| 808 |
+
"loss": 0.0035,
|
| 809 |
+
"step": 3225
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 50.73,
|
| 813 |
+
"learning_rate": 5.773846153846154e-06,
|
| 814 |
+
"loss": 0.0031,
|
| 815 |
+
"step": 3250
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 51.11,
|
| 819 |
+
"learning_rate": 5.735384615384616e-06,
|
| 820 |
+
"loss": 0.0023,
|
| 821 |
+
"step": 3275
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 51.5,
|
| 825 |
+
"learning_rate": 5.696923076923078e-06,
|
| 826 |
+
"loss": 0.0029,
|
| 827 |
+
"step": 3300
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 51.88,
|
| 831 |
+
"learning_rate": 5.658461538461539e-06,
|
| 832 |
+
"loss": 0.002,
|
| 833 |
+
"step": 3325
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 52.27,
|
| 837 |
+
"learning_rate": 5.620000000000001e-06,
|
| 838 |
+
"loss": 0.0023,
|
| 839 |
+
"step": 3350
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 52.66,
|
| 843 |
+
"learning_rate": 5.581538461538462e-06,
|
| 844 |
+
"loss": 0.0018,
|
| 845 |
+
"step": 3375
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 53.04,
|
| 849 |
+
"learning_rate": 5.543076923076923e-06,
|
| 850 |
+
"loss": 0.0017,
|
| 851 |
+
"step": 3400
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 53.43,
|
| 855 |
+
"learning_rate": 5.504615384615385e-06,
|
| 856 |
+
"loss": 0.0016,
|
| 857 |
+
"step": 3425
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 53.81,
|
| 861 |
+
"learning_rate": 5.4661538461538465e-06,
|
| 862 |
+
"loss": 0.0015,
|
| 863 |
+
"step": 3450
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 54.2,
|
| 867 |
+
"learning_rate": 5.427692307692308e-06,
|
| 868 |
+
"loss": 0.0018,
|
| 869 |
+
"step": 3475
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 54.59,
|
| 873 |
+
"learning_rate": 5.38923076923077e-06,
|
| 874 |
+
"loss": 0.0014,
|
| 875 |
+
"step": 3500
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 54.59,
|
| 879 |
+
"eval_loss": 0.6847506165504456,
|
| 880 |
+
"eval_runtime": 1692.6277,
|
| 881 |
+
"eval_samples_per_second": 1.952,
|
| 882 |
+
"eval_steps_per_second": 0.122,
|
| 883 |
+
"eval_wer": 33.334622524752476,
|
| 884 |
+
"step": 3500
|
| 885 |
+
}
|
| 886 |
+
],
|
| 887 |
+
"max_steps": 7000,
|
| 888 |
+
"num_train_epochs": 110,
|
| 889 |
+
"total_flos": 2.30522017775616e+19,
|
| 890 |
+
"trial_name": null,
|
| 891 |
+
"trial_params": null
|
| 892 |
+
}
|
checkpoint-3500/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d96bffa365dd9b279d59e578669e8c209f13b24443502000b7d54d12b313be97
|
| 3 |
+
size 3771
|
runs/Sep18_09-16-12_14ab48e85f35/events.out.tfevents.1695029544.14ab48e85f35.24.0
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7cac8bc399f7d1da92e028ba6f544ed37efdbad34d36633c6ff6614a5ebbe5e7
|
| 3 |
+
size 9837
|