Hello Wav2Vec
Browse files- README.md +237 -0
- config.json +76 -0
- dictionary.py +664 -0
- normalizer.py +203 -0
- predictions.csv +0 -0
- preprocessor_config.json +9 -0
- pytorch_model.bin +3 -0
- sample1.flac +0 -0
- sample2978.flac +0 -0
- sample5168.flac +0 -0
- special_tokens_map.json +1 -0
- tf_model.h5 +3 -0
- tokenizer_config.json +1 -0
- trainer_state.json +3341 -0
- vocab.json +1 -0
README.md
ADDED
@@ -0,0 +1,237 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
language: fa
|
3 |
+
datasets:
|
4 |
+
- common_voice
|
5 |
+
tags:
|
6 |
+
- audio
|
7 |
+
- automatic-speech-recognition
|
8 |
+
- speech
|
9 |
+
- xlsr-fine-tuning-week
|
10 |
+
widget:
|
11 |
+
- label: Common Voice sample 1
|
12 |
+
src: https://huggingface.co/m3hrdadfi/wav2vec2-xlsr-fa/resolve/main/sample1.flac
|
13 |
+
- label: Common Voice sample 2978
|
14 |
+
src: https://huggingface.co/m3hrdadfi/wav2vec2-xlsr-fa/resolve/main/sample2978.flac
|
15 |
+
- label: Common Voice sample 5168
|
16 |
+
src: https://huggingface.co/m3hrdadfi/wav2vec2-xlsr-fa/resolve/main/sample5168.flac
|
17 |
+
model-index:
|
18 |
+
- name: XLSR Wav2Vec2 Persian (Farsi) V3 by Mehrdad Farahani
|
19 |
+
results:
|
20 |
+
- task:
|
21 |
+
name: Speech Recognition
|
22 |
+
type: automatic-speech-recognition
|
23 |
+
dataset:
|
24 |
+
name: Common Voice fa
|
25 |
+
type: common_voice
|
26 |
+
args: fa
|
27 |
+
metrics:
|
28 |
+
- name: Test WER
|
29 |
+
type: wer
|
30 |
+
value: 10.36
|
31 |
+
|
32 |
+
---
|
33 |
+
|
34 |
+
# Wav2Vec2-Large-XLSR-53-Persian V3
|
35 |
+
|
36 |
+
|
37 |
+
## Usage
|
38 |
+
Fine-tuned [facebook/wav2vec2-large-xlsr-53](https://huggingface.co/facebook/wav2vec2-large-xlsr-53) in Persian (Farsi) using [Common Voice](https://huggingface.co/datasets/common_voice). When using this model, make sure that your speech input is sampled at 16kHz.
|
39 |
+
|
40 |
+
|
41 |
+
**Requirements**
|
42 |
+
```bash
|
43 |
+
# requirement packages
|
44 |
+
!pip install git+https://github.com/huggingface/datasets.git
|
45 |
+
!pip install git+https://github.com/huggingface/transformers.git
|
46 |
+
!pip install torchaudio
|
47 |
+
!pip install librosa
|
48 |
+
!pip install jiwer
|
49 |
+
!pip install parsivar
|
50 |
+
!pip install num2fawords
|
51 |
+
```
|
52 |
+
|
53 |
+
**Normalizer**
|
54 |
+
```bash
|
55 |
+
# Normalizer
|
56 |
+
!wget -O normalizer.py https://huggingface.co/m3hrdadfi/"wav2vec2-large-xlsr-persian-v3/raw/main/dictionary.py
|
57 |
+
!wget -O normalizer.py https://huggingface.co/m3hrdadfi/"wav2vec2-large-xlsr-persian-v3/raw/main/normalizer.py
|
58 |
+
```
|
59 |
+
|
60 |
+
**Downloading data**
|
61 |
+
```bash
|
62 |
+
wget https://voice-prod-bundler-ee1969a6ce8178826482b88e843c335139bd3fb4.s3.amazonaws.com/cv-corpus-6.1-2020-12-11/fa.tar.gz
|
63 |
+
|
64 |
+
tar -xzf fa.tar.gz
|
65 |
+
rm -rf fa.tar.gz
|
66 |
+
```
|
67 |
+
|
68 |
+
**Cleaning**
|
69 |
+
```python
|
70 |
+
from normalizer import normalizer
|
71 |
+
|
72 |
+
def cleaning(text):
|
73 |
+
if not isinstance(text, str):
|
74 |
+
return None
|
75 |
+
|
76 |
+
return normalizer({"sentence": text}, return_dict=False)
|
77 |
+
|
78 |
+
data_dir = "/content/cv-corpus-6.1-2020-12-11/fa"
|
79 |
+
|
80 |
+
test = pd.read_csv(f"{data_dir}/test.tsv", sep="\t")
|
81 |
+
test["path"] = data_dir + "/clips/" + test["path"]
|
82 |
+
print(f"Step 0: {len(test)}")
|
83 |
+
|
84 |
+
test["status"] = test["path"].apply(lambda path: True if os.path.exists(path) else None)
|
85 |
+
test = test.dropna(subset=["path"])
|
86 |
+
test = test.drop("status", 1)
|
87 |
+
print(f"Step 1: {len(test)}")
|
88 |
+
|
89 |
+
test["sentence"] = test["sentence"].apply(lambda t: cleaning(t))
|
90 |
+
test = test.dropna(subset=["sentence"])
|
91 |
+
print(f"Step 2: {len(test)}")
|
92 |
+
|
93 |
+
test = test.reset_index(drop=True)
|
94 |
+
print(test.head())
|
95 |
+
|
96 |
+
test = test[["path", "sentence"]]
|
97 |
+
test.to_csv("/content/test.csv", sep="\t", encoding="utf-8", index=False)
|
98 |
+
```
|
99 |
+
|
100 |
+
**Prediction**
|
101 |
+
```python
|
102 |
+
import numpy as np
|
103 |
+
import pandas as pd
|
104 |
+
|
105 |
+
import librosa
|
106 |
+
import torch
|
107 |
+
import torchaudio
|
108 |
+
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
|
109 |
+
from datasets import load_dataset, load_metric
|
110 |
+
|
111 |
+
import IPython.display as ipd
|
112 |
+
|
113 |
+
model_name_or_path = "m3hrdadfi/wav2vec2-large-xlsr-persian-v3"
|
114 |
+
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
115 |
+
print(model_name_or_path, device)
|
116 |
+
|
117 |
+
processor = Wav2Vec2Processor.from_pretrained(model_name_or_path)
|
118 |
+
model = Wav2Vec2ForCTC.from_pretrained(model_name_or_path).to(device)
|
119 |
+
|
120 |
+
|
121 |
+
def speech_file_to_array_fn(batch):
|
122 |
+
speech_array, sampling_rate = torchaudio.load(batch["path"])
|
123 |
+
speech_array = speech_array.squeeze().numpy()
|
124 |
+
speech_array = librosa.resample(np.asarray(speech_array), sampling_rate, processor.feature_extractor.sampling_rate)
|
125 |
+
|
126 |
+
batch["speech"] = speech_array
|
127 |
+
return batch
|
128 |
+
|
129 |
+
|
130 |
+
def predict(batch):
|
131 |
+
features = processor(
|
132 |
+
batch["speech"],
|
133 |
+
sampling_rate=processor.feature_extractor.sampling_rate,
|
134 |
+
return_tensors="pt",
|
135 |
+
padding=True
|
136 |
+
)
|
137 |
+
|
138 |
+
input_values = features.input_values.to(device)
|
139 |
+
attention_mask = features.attention_mask.to(device)
|
140 |
+
|
141 |
+
with torch.no_grad():
|
142 |
+
logits = model(input_values, attention_mask=attention_mask).logits
|
143 |
+
|
144 |
+
pred_ids = torch.argmax(logits, dim=-1)
|
145 |
+
|
146 |
+
batch["predicted"] = processor.batch_decode(pred_ids)
|
147 |
+
return batch
|
148 |
+
|
149 |
+
|
150 |
+
dataset = load_dataset("csv", data_files={"test": "/content/test.csv"}, delimiter="\t")["test"]
|
151 |
+
dataset = dataset.map(speech_file_to_array_fn)
|
152 |
+
result = dataset.map(predict, batched=True, batch_size=4)
|
153 |
+
```
|
154 |
+
|
155 |
+
**WER Score**
|
156 |
+
```python
|
157 |
+
wer = load_metric("wer")
|
158 |
+
print("WER: {:.2f}".format(100 * wer.compute(predictions=result["predicted"], references=result["sentence"])))
|
159 |
+
```
|
160 |
+
|
161 |
+
**Output**
|
162 |
+
```python
|
163 |
+
max_items = np.random.randint(0, len(result), 20).tolist()
|
164 |
+
for i in max_items:
|
165 |
+
reference, predicted = result["sentence"][i], result["predicted"][i]
|
166 |
+
print("reference:", reference)
|
167 |
+
print("predicted:", predicted)
|
168 |
+
print('---')
|
169 |
+
```
|
170 |
+
|
171 |
+
```text
|
172 |
+
reference: ماجرا رو براش تعریف کردم اون گفت مریم اگه میدونی پسر خوبیه خب چه اشکالی داره باهاش بیشتر اشنا بشو
|
173 |
+
predicted: ماجرا رو براش تعریف کردم اون گفت مریم اگه میدونی پسر خوبیه خب چه اشکالی داره باهاش بیشتر اشنا بشو
|
174 |
+
---
|
175 |
+
reference: بیا پایین تو اجازه نداری بری اون بالا
|
176 |
+
predicted: بیا پایین تو اجازه نداری بری اون بالا
|
177 |
+
---
|
178 |
+
reference: هر روز یک دو مداد کش می رفتتم تااین که تا پایان ترم از تمامی دوستانم مداد برداشته بودم
|
179 |
+
predicted: هر روز یک دو مداد کش می رفتم تااین که تا پایین ترم از تمامی دوستان و مداد برداشته بودم
|
180 |
+
---
|
181 |
+
reference: فکر میکنی آروم میشینه
|
182 |
+
predicted: فکر میکنی آروم میشینه
|
183 |
+
---
|
184 |
+
reference: هرکسی با گوشی هوشمند خود میتواند با کایلا متصل گردد در یک محدوده مکانی
|
185 |
+
predicted: هرکسی با گوشی هوشمند خود میتواند با کایلا متصل گردد در یک محدوده مکانی
|
186 |
+
---
|
187 |
+
reference: برو از مهرداد بپرس
|
188 |
+
predicted: برو از مهرداد بپرس
|
189 |
+
---
|
190 |
+
reference: می خواهم شما را با این قدمها آشنا کنم
|
191 |
+
predicted: می خواهم شما را با این قدمها آشنا کنم
|
192 |
+
---
|
193 |
+
reference: میدونم یه روز دوباره می تونم تو رو ببینم
|
194 |
+
predicted: میدونم یه روز دوباره می تونم تو رو ببینم
|
195 |
+
---
|
196 |
+
reference: بسیار خوب خواهد بود دعوت او را بپذیری
|
197 |
+
predicted: بسیار خوب خواهد بود دعوت او را بپذیری
|
198 |
+
---
|
199 |
+
reference: بهت بگن آشغالی خوبه
|
200 |
+
predicted: بهت بگن آشغالی خوبه
|
201 |
+
---
|
202 |
+
reference: چرا معاشرت با هم ایمانان ما را محفوظ نگه میدارد
|
203 |
+
predicted: چرا معاشرت با هم ایمانان آ را م حفوظ نگه میدارد
|
204 |
+
---
|
205 |
+
reference: بولیوی پس از گویان فقیرترین کشور آمریکای جنوبی است
|
206 |
+
predicted: بولیوی پس از گویان فقیرترین کشور آمریکای جنوبی است
|
207 |
+
---
|
208 |
+
reference: بعد از مدتی اینکار برایم عادی شد
|
209 |
+
predicted: بعد از مدتی اینکار برایم عادو شد
|
210 |
+
---
|
211 |
+
reference: به نظر اون هم همینطوره
|
212 |
+
predicted: به نظر اون هم همینطوره
|
213 |
+
---
|
214 |
+
reference: هیچ مایونز ی دارید
|
215 |
+
predicted: هیچ مایونز ی دارید
|
216 |
+
---
|
217 |
+
reference: هیچ یک از انان کاری به سنگ نداشتند
|
218 |
+
predicted: هیچ شک از انان کاری به سنگ نداشتند
|
219 |
+
---
|
220 |
+
reference: می خواهم کمی کتاب شعر ببینم
|
221 |
+
predicted: می خواهم کتاب شعر ببینم
|
222 |
+
---
|
223 |
+
reference: همین شوهر فهیمه مگه نمی گفتی فرمانده بوده کو
|
224 |
+
predicted: همین شوهر فهیمه بینامی گفتی فهمانده بود کو
|
225 |
+
---
|
226 |
+
reference: اون جاها کسی رو نمیبینی که تو دستش کتاب نباشه
|
227 |
+
predicted: اون جاها کسی رو نمیبینی که تو دستش کتاب نباشه
|
228 |
+
---
|
229 |
+
reference: زندان رفتن من در این سالهای اخیر برام شانس بزرگی بود که معما و مشکل چندین سالهام را حل کرد
|
230 |
+
predicted: زندان رفتن من در این سالها اخی براب شانس بزرگی بود که معما و مشکل چندین سالهام را حل کرد
|
231 |
+
---
|
232 |
+
```
|
233 |
+
|
234 |
+
## Evaluation
|
235 |
+
|
236 |
+
**Test Result:**
|
237 |
+
- WER: 10.36%
|
config.json
ADDED
@@ -0,0 +1,76 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "facebook/wav2vec2-large-xlsr-53",
|
3 |
+
"activation_dropout": 0.09216,
|
4 |
+
"apply_spec_augment": true,
|
5 |
+
"architectures": [
|
6 |
+
"Wav2Vec2ForCTC"
|
7 |
+
],
|
8 |
+
"attention_dropout": 0.05316,
|
9 |
+
"bos_token_id": 1,
|
10 |
+
"conv_bias": true,
|
11 |
+
"conv_dim": [
|
12 |
+
512,
|
13 |
+
512,
|
14 |
+
512,
|
15 |
+
512,
|
16 |
+
512,
|
17 |
+
512,
|
18 |
+
512
|
19 |
+
],
|
20 |
+
"conv_kernel": [
|
21 |
+
10,
|
22 |
+
3,
|
23 |
+
3,
|
24 |
+
3,
|
25 |
+
3,
|
26 |
+
2,
|
27 |
+
2
|
28 |
+
],
|
29 |
+
"conv_stride": [
|
30 |
+
5,
|
31 |
+
2,
|
32 |
+
2,
|
33 |
+
2,
|
34 |
+
2,
|
35 |
+
2,
|
36 |
+
2
|
37 |
+
],
|
38 |
+
"ctc_loss_reduction": "mean",
|
39 |
+
"ctc_zero_infinity": true,
|
40 |
+
"do_stable_layer_norm": true,
|
41 |
+
"eos_token_id": 2,
|
42 |
+
"feat_extract_activation": "gelu",
|
43 |
+
"feat_extract_dropout": 0.0,
|
44 |
+
"feat_extract_norm": "layer",
|
45 |
+
"feat_proj_dropout": 0.01249,
|
46 |
+
"final_dropout": 0.0,
|
47 |
+
"gradient_checkpointing": true,
|
48 |
+
"hidden_act": "gelu",
|
49 |
+
"hidden_dropout": 0.01941,
|
50 |
+
"hidden_size": 1024,
|
51 |
+
"initializer_range": 0.02,
|
52 |
+
"intermediate_size": 4096,
|
53 |
+
"layer_norm_eps": 1e-05,
|
54 |
+
"layerdrop": 0.01377,
|
55 |
+
"mask_channel_length": 10,
|
56 |
+
"mask_channel_min_space": 1,
|
57 |
+
"mask_channel_other": 0.0,
|
58 |
+
"mask_channel_prob": 0.0,
|
59 |
+
"mask_channel_selection": "static",
|
60 |
+
"mask_feature_length": 10,
|
61 |
+
"mask_feature_prob": 0.0,
|
62 |
+
"mask_time_length": 10,
|
63 |
+
"mask_time_min_space": 1,
|
64 |
+
"mask_time_other": 0.0,
|
65 |
+
"mask_time_prob": 0.04529,
|
66 |
+
"mask_time_selection": "static",
|
67 |
+
"model_type": "wav2vec2",
|
68 |
+
"num_attention_heads": 16,
|
69 |
+
"num_conv_pos_embedding_groups": 16,
|
70 |
+
"num_conv_pos_embeddings": 128,
|
71 |
+
"num_feat_extract_layers": 7,
|
72 |
+
"num_hidden_layers": 24,
|
73 |
+
"pad_token_id": 0,
|
74 |
+
"transformers_version": "4.6.0.dev0",
|
75 |
+
"vocab_size": 40
|
76 |
+
}
|
dictionary.py
ADDED
@@ -0,0 +1,664 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
dictionary_mapping = {
|
2 |
+
'ك': 'ک', 'دِ': 'د', 'بِ': 'ب', 'زِ': 'ز', 'ذِ': 'ذ', 'شِ': 'ش', 'سِ': 'س', 'ى': 'ی',
|
3 |
+
'ي': 'ی', 'أ': 'ا', 'ؤ': 'و', "ے": "ی", "ۀ": "ه", "ﭘ": "پ", "ﮐ": "ک", "ﯽ": "ی",
|
4 |
+
"ﺎ": "ا", "ﺑ": "ب", "ﺘ": "ت", "ﺧ": "خ", "ﺩ": "د", "ﺱ": "س", "ﻀ": "ض", "ﻌ": "ع",
|
5 |
+
"ﻟ": "ل", "ﻡ": "م", "ﻢ": "م", "ﻪ": "ه", "ﻮ": "و", 'ﺍ': "ا", 'ة': "ه",
|
6 |
+
'ﯾ': "ی", 'ﯿ': "ی", 'ﺒ': "ب", 'ﺖ': "ت", 'ﺪ': "د", 'ﺮ': "ر", 'ﺴ': "س", 'ﺷ': "ش",
|
7 |
+
'ﺸ': "ش", 'ﻋ': "ع", 'ﻤ': "م", 'ﻥ': "ن", 'ﻧ': "ن", 'ﻭ': "و", 'ﺭ': "ر", "ﮔ": "گ",
|
8 |
+
|
9 |
+
"a": "ای", "b": "بی", "c": "سی", "d": "دی", "e": "ایی", "f": "اف",
|
10 |
+
"g": "جی", "h": "اچ", "i": "آی", "j": "جی", "k": "کی", "l": "ال",
|
11 |
+
"m": "ام", "n": "ان", "o": "او", "p": "پی", "q": "کیو", "r": "آر",
|
12 |
+
"s": "اس", "t": "تی", "u": "یو", "v": "وی", "w": "دبلیو", "x": "اکس",
|
13 |
+
"y": "وای", "z": "زد ",
|
14 |
+
"\u200c": " ", "\u200d": " ", "\u200e": " ", "\u200f": " ", "\ufeff": " ",
|
15 |
+
|
16 |
+
"نو آوریمان": "نوآوریمان",
|
17 |
+
"نو آوری مان": "نوآوریمان",
|
18 |
+
"نو آوریمان": "نوآوریمان",
|
19 |
+
" ا م ": "ام ",
|
20 |
+
" م ": "ام ",
|
21 |
+
"کنندهای": "کنندهای",
|
22 |
+
"ارائهای": "ارائهای",
|
23 |
+
"ایدهای": "ایدهای",
|
24 |
+
"ماسهای": "ماسهای",
|
25 |
+
"خامنهای": "خامنهای",
|
26 |
+
"قلهای": "قلهای",
|
27 |
+
"سیارهای": "سیارهای",
|
28 |
+
"کیسهای": "کیسهای",
|
29 |
+
"شانهای": "شانهای",
|
30 |
+
"غریبهای": "غریبهای",
|
31 |
+
"برنامهای": "برنامهای",
|
32 |
+
"سختگیرانهای": "سختگیرانهای",
|
33 |
+
"بهانهای": "بهانهای",
|
34 |
+
"زیرروالهای": "زیر روالهای",
|
35 |
+
"درهای": "درهای",
|
36 |
+
"آمادهای": "آمادهای",
|
37 |
+
"سادهای": "سادهای",
|
38 |
+
"سرمایهگذارهای": "سرمایه گذارهای",
|
39 |
+
"فوقالعادهای": "فوقالعادهای",
|
40 |
+
"حادثهای": "حادثهای",
|
41 |
+
"نویسندههای": "نویسندههای",
|
42 |
+
"علاقهای": "علاقهای",
|
43 |
+
"برجستهای": "برجستهای",
|
44 |
+
"جلگهای": "جلگهای",
|
45 |
+
"زندهای": "زندهای",
|
46 |
+
"فنآوریهای": "فناوریهای",
|
47 |
+
"سایهروشنهای": "سایه روشنهای",
|
48 |
+
"بیسابقهای": "بی سابقهای",
|
49 |
+
"فرضیهای": "فرضیهای",
|
50 |
+
"راهاندازهای": "راه اندازهای",
|
51 |
+
"بیشهای": "بیشهای",
|
52 |
+
"مقالهای": "مقالهای",
|
53 |
+
"دیگهای": "دیگهای",
|
54 |
+
"ماههاست": "ماه هاست",
|
55 |
+
"نرمافزارهای": "نرمافزارهای",
|
56 |
+
"کتابسوزانهای": "کتاب سوزانهای",
|
57 |
+
"سیستمعاملهای": "سیستم عاملهای",
|
58 |
+
"اسلحهای": "اسلحهای",
|
59 |
+
"وقفهای": "وقفهای",
|
60 |
+
"زمینهای": "زمینهای",
|
61 |
+
"حرامزادههای": "حرامزادههای",
|
62 |
+
"هزینهای": "هزینهای",
|
63 |
+
"انداختهای": "انداختهای",
|
64 |
+
"جسورانهای": "جسورانهای",
|
65 |
+
"فاجعهای": "فاجعهای",
|
66 |
+
"جامعهای": "جامعهای",
|
67 |
+
"پدیدهای": "پدیدهای",
|
68 |
+
"اغواگرانهای": "اغواگرانهای",
|
69 |
+
"تکانهای": "تکانهای",
|
70 |
+
"لولهای": "لولهای",
|
71 |
+
"نشانهای": "نشانهای",
|
72 |
+
"وسیلهای": "وسیلهای",
|
73 |
+
"آیندهای": "آیندهای",
|
74 |
+
"بردهای": "بردهای",
|
75 |
+
"سابقهای": "سابقهای",
|
76 |
+
"ناحیهای": "ناحیهای",
|
77 |
+
"تکاندهندهای": "تکان دهندهای",
|
78 |
+
"بودجهای": "بودجهای",
|
79 |
+
"روزانهای": "روزانهای",
|
80 |
+
"چارهای": "چارهای",
|
81 |
+
"انگیزهای": "انگیزهای",
|
82 |
+
"دادهای": "دادهای",
|
83 |
+
"عدهای": "عدهای",
|
84 |
+
"هفتهای": "هفتهای",
|
85 |
+
"منطقهای": "منطقهای",
|
86 |
+
"استارتآپهای": "استارتاپهای",
|
87 |
+
"سازهای": "سازهای",
|
88 |
+
"مجموعهای": "مجموعهای",
|
89 |
+
"فلسفهای": "فلسفهای",
|
90 |
+
"تذکردهندهای": "تذکر دهندهای",
|
91 |
+
"مصاحبهای": "مصابحهای",
|
92 |
+
"نمونهای": "نمونهای",
|
93 |
+
"قلمموهای": "قلم موهای",
|
94 |
+
"شبزندهداری": "شب زندهداری",
|
95 |
+
"خوردهباشد": "خورده باشد",
|
96 |
+
"داشتهباشید": "داشته باشید",
|
97 |
+
"فزایندهای": "فزایندهای",
|
98 |
+
"عمدهای": "عمدهای",
|
99 |
+
"بدیهایی": "بدیهای",
|
100 |
+
"نوشتهایم": "نوشتهایم",
|
101 |
+
"بنتالهدی": "بنت الهدی",
|
102 |
+
"نوشتهام": "نوشتهام",
|
103 |
+
"سرمایهگذاران": "سرمایه گذاران",
|
104 |
+
"خانهی": "خانهی",
|
105 |
+
"گستاخانهی": "گستاخانهی",
|
106 |
+
"گرفتهباشیم": "گرفته باشیم",
|
107 |
+
"خونهی": "خونهی",
|
108 |
+
"داشتهام": "داشتهام",
|
109 |
+
"رشتهام": "رشتهام",
|
110 |
+
"سرمایهگذارانشان": "سرمایه گذارانشان",
|
111 |
+
"ریشهکنی": "ریشهکنی",
|
112 |
+
"مودبانهتری": "مودبانهتری",
|
113 |
+
"برگردانشدهاند": "برگردان شدهاند",
|
114 |
+
"قرمهسبزی": "قرمهسبزی",
|
115 |
+
"راهجویی": "راه جویی",
|
116 |
+
"اماهیچوقت": "اما هیچوقت",
|
117 |
+
"آبوهوای": "آب و هوای",
|
118 |
+
"بقیهاش": "بقیهاش",
|
119 |
+
"طبقهبندی": "طبقهبندی",
|
120 |
+
"مردههان": "مرده هان",
|
121 |
+
"آمادهاند": "آمادهاند",
|
122 |
+
"نشدهاید": "نشدهاید",
|
123 |
+
"آگاهیرسانی": "آگاهی رسانی",
|
124 |
+
"نداشتهاند": "نداشتهاند",
|
125 |
+
"شکنانهترین": "شکنانهترین",
|
126 |
+
"اقدامهایی": "اقدامهایی",
|
127 |
+
"راهآهن": "راه آهن",
|
128 |
+
"شدهاند": "شدهاند",
|
129 |
+
"تازهترین": "تازهترین",
|
130 |
+
"روبهروی": "رو به روی",
|
131 |
+
"منحصربهفرد": "منحصر به فرد",
|
132 |
+
"سیزدهبدر": "سیزده بدر",
|
133 |
+
"برندهی": "برندهی",
|
134 |
+
"خانهاشتراکی": "خانه اشتراکی",
|
135 |
+
"دادههایی": "دادههایی",
|
136 |
+
"استفادهتر": "استفادهتر",
|
137 |
+
"گذرنامهتان": "گذرنامهتان",
|
138 |
+
"کهنترین": "کهنهترین",
|
139 |
+
"فرهنگسرا": "فرهنگسرا",
|
140 |
+
"آمادهاید": "آمادهاید",
|
141 |
+
"ویژهی": "ویژهی",
|
142 |
+
"غریزهات": "غریزهات",
|
143 |
+
"مادرشوهری": "مادر شوهری",
|
144 |
+
"نبودهام": "نبودهام",
|
145 |
+
"بودهاند": "بودهاند",
|
146 |
+
"وتنها": "و تنها",
|
147 |
+
"بداههکاری": "بداههکاری",
|
148 |
+
"سرمایهگذار": "سرمایه گذار",
|
149 |
+
"برنامهنویس": "برنامه نویس",
|
150 |
+
"مهنازخانم": "مهناز خانم",
|
151 |
+
"مواجهاند": "مواجهاند",
|
152 |
+
"توسعهاش": "توسعهاش",
|
153 |
+
"سینهام": "سینهام",
|
154 |
+
"سینهام": "سینهام",
|
155 |
+
"نمیخواهند": "نمیخواهند",
|
156 |
+
"فنآوریها": "فناوریها",
|
157 |
+
"دنبالهرو": "دنبالهرو",
|
158 |
+
"لبهی": "لبهی",
|
159 |
+
"اللهیار": "الله یار",
|
160 |
+
"ارزندهتر": "ارزندهتر",
|
161 |
+
"برههای": "برهای",
|
162 |
+
"پیادهسازی": "پیادهسازی",
|
163 |
+
"دهسالگی": "ده سالگی",
|
164 |
+
"رسانهای": "رسانهای",
|
165 |
+
"ریشسفیدها": "ریش سفیدها",
|
166 |
+
"چهجوری": "چه جوری",
|
167 |
+
"ویژگیهایی": "ویژگیهایی",
|
168 |
+
"میفهمیم": "میفهمیم",
|
169 |
+
"وبهم": "و بهم",
|
170 |
+
"قطرهای": "قطرهای",
|
171 |
+
"ازتنهایی": "از تنهایی",
|
172 |
+
"لطیفهای": "لطیفهای",
|
173 |
+
"باشهاومدم": "باشه اومدم",
|
174 |
+
"منحصربهفردترین": "منحصر به فردترین",
|
175 |
+
"کردهاند": "کردهاند",
|
176 |
+
"اندازهای": "اندازهای",
|
177 |
+
"بهرهبرداری": "بهره برداری",
|
178 |
+
"اماشوهرجان": "اما شوهر جان",
|
179 |
+
"خانوادهاش": "خانوادهاش",
|
180 |
+
"نشدهاند": "نشدهاند",
|
181 |
+
"نکردهایم": "نکردهایم",
|
182 |
+
"تخممرغهایش": "تخم مرغهایش",
|
183 |
+
"وظیفهش": "وظیفهاش",
|
184 |
+
"مشگینشهر": "مشگی شهر",
|
185 |
+
"توسعهدهندگانش": "توسعه دهندگانش",
|
186 |
+
"امینابراهیم": "امین ابراهیم",
|
187 |
+
"دربارهاش": "دربارهاش",
|
188 |
+
"میانافزارها": "میانافزارها",
|
189 |
+
"دیدهاند": "دیدهاند",
|
190 |
+
"خانوادهام": "خانوادهام",
|
191 |
+
"مایهی": "مایهی",
|
192 |
+
"نوشتهشدن": "نوشته شدن",
|
193 |
+
"راهحلهایشان": "راه حلهایشان",
|
194 |
+
"میهماننواز": "میهمان نواز",
|
195 |
+
"زیبندهی": "زیرندهی",
|
196 |
+
"راههایی": "راههایی",
|
197 |
+
"جربزهی": "جربزهی",
|
198 |
+
"بهجا": " به جا",
|
199 |
+
"بطورهمزمان": "به طور همزمان",
|
200 |
+
"فهمیدهبود": "فهمیده بود",
|
201 |
+
"دوربرگردانها": "دور برگردانها",
|
202 |
+
"شالودهی": "شالودهی",
|
203 |
+
"راهکاریی": "راهکاری",
|
204 |
+
"مخالفتهایی": "مخالفتهایی",
|
205 |
+
"چیزهاازشون": "چیزها ازشون",
|
206 |
+
"سکونتگاههای": "سکونت گاههای",
|
207 |
+
"سالهابود": "سالها بود",
|
208 |
+
"نمونهی": "نمونهی",
|
209 |
+
"سرمایهگذاری": "سرمایه گذاری",
|
210 |
+
"شبکهای": "شبکهای",
|
211 |
+
"خواهرشوهر": "خواهر شوهر",
|
212 |
+
"سرگیجهآور": "سرگیجه آور",
|
213 |
+
"آستانهی": "آستانهی",
|
214 |
+
"دادهاست": "داده است",
|
215 |
+
"مجسمهسازی": "مجسمه سازی",
|
216 |
+
"ماهرانهترین": "ماهرانهترین",
|
217 |
+
"پنجشنبههایی": "پنجشنبه شبهایی",
|
218 |
+
"نرفنهام": "نرفتهام",
|
219 |
+
"قورمهسبزی": "قورمه سبزی",
|
220 |
+
"گذارهای": "گذارهای",
|
221 |
+
"بندهخدا": "بنده خدا",
|
222 |
+
"روزنامهنگاران": "روزنامه نگاران",
|
223 |
+
"نقشهی": "نقشهی",
|
224 |
+
"حملهی": "حملهی",
|
225 |
+
"تکنیکهاست": "تکنیک هاست",
|
226 |
+
"نرمافزارهایمان": "نرمافرارهایمان",
|
227 |
+
"مادرشوهرم": "مادر شوهرم",
|
228 |
+
"ماهگیمون": "ماه گیمون",
|
229 |
+
"مادرشوهرمحترم": "مادر شوهر محترم",
|
230 |
+
"شوهرداری": "شوهر داری",
|
231 |
+
"سرمایهگذارها": "سرمایه گذارها",
|
232 |
+
"بهرهمند": "بهرهمند",
|
233 |
+
"درمانهایی": "درمانهایی",
|
234 |
+
"عامدانهتر": "عامدانهتر",
|
235 |
+
"تازهوارد": "تازه وارد",
|
236 |
+
"مونتهویدئو": "مونته ویدئو",
|
237 |
+
"ذائقهاش": "ذائقهاش",
|
238 |
+
"گوشهگیرتر": "گوشهگیرتر",
|
239 |
+
"دنبالهدار": "دنبالهدار",
|
240 |
+
"بیخانمانها": "بیخانمانها",
|
241 |
+
"سرمایهدارها": "سرمایهدارها",
|
242 |
+
"مادرشوهریم": "مادر شوهریم",
|
243 |
+
"صبحانهاش": "صبحانهاش",
|
244 |
+
"جنازهست": "جنازه است",
|
245 |
+
"شمارهات": "شمارهای",
|
246 |
+
"بهقدری": "به قدری",
|
247 |
+
"کیسهی": "کیسهی",
|
248 |
+
"کوششهایی": "کوششهایی",
|
249 |
+
"مادرشوهر": "مادر شوهر",
|
250 |
+
"رابطهی": "رابطهی",
|
251 |
+
"نوشتهاند": "نوشتهاند",
|
252 |
+
"کنجکاوانهی": "کنجکاوانهی",
|
253 |
+
"غیرمتعهد": "غیر متعهد",
|
254 |
+
"کردهای": "کردهای",
|
255 |
+
"وهمکارانم": "و همکارانم",
|
256 |
+
"گردهمآیی": "گردهمایی",
|
257 |
+
"اللهوردی": "الله وردی",
|
258 |
+
"صرفهجویی": "صرفه جویی",
|
259 |
+
"ماندهاند": "ماندهاند",
|
260 |
+
"برنامهنویسی": "برنامهنویسی",
|
261 |
+
"امینمهدی": "امین مهدی",
|
262 |
+
"سهامدارنی": "سهام دارانی",
|
263 |
+
"مسابقهی": "مسابقهی",
|
264 |
+
"ستارهشناسم": "ستار شناسم",
|
265 |
+
"گرفتهاند": "گرفتهاند",
|
266 |
+
"جامعهشان": "جامعهشان",
|
267 |
+
"بچهی": "بچهی",
|
268 |
+
"شیوهی": "شیوهی",
|
269 |
+
"بهکار": "به کار",
|
270 |
+
"بهتراست": "بهتر است",
|
271 |
+
"سروکلهشون": "سر و کلهشون",
|
272 |
+
"رسیدهمسرش": "رسید همسرش",
|
273 |
+
"پسراهل": "پسر اهل",
|
274 |
+
"پروژههای": "پروژههای",
|
275 |
+
"عاقلانهام": "عاقلانهام",
|
276 |
+
"گذاشتهاند": "گذاشتهاند",
|
277 |
+
"کردهام": "کردهام",
|
278 |
+
"اندازهگیری": "اندازه گیری",
|
279 |
+
"یاوهگویی": "یاوه گویی",
|
280 |
+
"سازمانهایی": "سازمانهایی",
|
281 |
+
"نمودهاند": "نمودهاند",
|
282 |
+
"تنهاییآور": "تنهایی آور",
|
283 |
+
"قراردهیم": "قرار دهیم",
|
284 |
+
"ازشوهرجان": "از شوهر جان",
|
285 |
+
"کرهجنوبی": "کره جنوبی",
|
286 |
+
"توهینآمیز": "توهین آمیز",
|
287 |
+
"فنآوریهایی": "فناوریهایی",
|
288 |
+
"داشتهاید": "داشتهاید",
|
289 |
+
"شدهایم": "شدهایم",
|
290 |
+
"نمیفهمم": "نمیفهمم",
|
291 |
+
"مثالهایی": "مثالهایی",
|
292 |
+
"رییسجمهور": "رییس جمهور",
|
293 |
+
"مجموعهی": "مجموعهی",
|
294 |
+
"درندهاند": "درندهاند",
|
295 |
+
"امابهش": "اما بهش",
|
296 |
+
"بازخواهند": "باز خواهند",
|
297 |
+
"برنامههایی": "برنامههایی",
|
298 |
+
"یهجا": "یه جا",
|
299 |
+
"زگیلهایی": "زگیلهایی",
|
300 |
+
"وسیلهی": "وسیلهی",
|
301 |
+
"بهمنیار": "بهمن یار",
|
302 |
+
"دادهام": "دادهام",
|
303 |
+
"بههنگام": "به هنگام",
|
304 |
+
"بهدروغ": "به دروغ",
|
305 |
+
"دورافتادهترین": "دور افتادهترین",
|
306 |
+
"نامهایی": "نامهایی",
|
307 |
+
"سهقسمتی": "سه قسمتی",
|
308 |
+
"توجهازچیدن": "توجه از چیدن",
|
309 |
+
"پیامرسانها": "پیام رسانها",
|
310 |
+
"بهمنزاد": "بهمن زاد",
|
311 |
+
"نشانههایی": "نشانههایی",
|
312 |
+
"راهحلهای": "راه حلهای",
|
313 |
+
"راهحلهایی": "راه حلهایی",
|
314 |
+
"راهحلهای": "راه حلهای",
|
315 |
+
"نظرخواهیها": "نظر خواهیها",
|
316 |
+
"نظرخواهیها": "نظر خواهیها",
|
317 |
+
"کندهی": "کندهی",
|
318 |
+
"حرامزادههای": "حرام زادههای",
|
319 |
+
"شبیهسازیهایی": "شبیه سازیهایی",
|
320 |
+
"مهارتهایی": "مهارتهایی",
|
321 |
+
"روبهرویشان": "رو به رویشان",
|
322 |
+
"برجستهترین": "برجستهترین",
|
323 |
+
"نمیفهمیدم": "نمیفهمیدم",
|
324 |
+
"دستگاههایی": "دستگاههایی",
|
325 |
+
"برادرشوهر": "برادر شوهر",
|
326 |
+
"گرسنهام": "گرستهام",
|
327 |
+
"گرسنههام": "گرستهام",
|
328 |
+
"قهوهخوری": "قهوه خوری",
|
329 |
+
"دادهاید": "دادهاید",
|
330 |
+
"بهآرامی": "به آرمانی",
|
331 |
+
"دانستنیهاست": "دانستنیهاست",
|
332 |
+
"بهراحتی": "به راحتی",
|
333 |
+
"ایدهپردازی": "ایدهپردازی",
|
334 |
+
"ریشسفیدهای": "ریش سفیدهای",
|
335 |
+
"خفهمون": "خفه مون",
|
336 |
+
"بهجای": "به جای",
|
337 |
+
"ریزخشونتها": "ریز خشونتها",
|
338 |
+
"ریزخشونتها": "ریز خشونتها",
|
339 |
+
"حساسیتهایی": "حساسیتهایی",
|
340 |
+
"پشتصحنهی": "پشت صحنهی",
|
341 |
+
"کلهی": "کلهی",
|
342 |
+
"تاشوهرم": "تا شوهرم",
|
343 |
+
"آیندهاش": "آیندهاش",
|
344 |
+
"پروانههایی": "پروانههایی",
|
345 |
+
"خوبیهایی": "خوبیهایی",
|
346 |
+
"نرمافزارها": "نرمافزارها",
|
347 |
+
"رساندهاند": "رساندهاند",
|
348 |
+
"سرمایهگذارنی": "سرمایه گذارانی",
|
349 |
+
"تکهچسبانی": "تکه چسبانی",
|
350 |
+
"بیتوجهی": "بی توجهی",
|
351 |
+
"جاهطلبی": "جاه طلبی",
|
352 |
+
"پرغلغلهتان": "پر غلغلهتان",
|
353 |
+
"خمینیشهر": "خمینی شهر",
|
354 |
+
"رشتهتوییت": "رشته توییت",
|
355 |
+
"موهبتهایی": "موهبتهایی",
|
356 |
+
"برنامهی": "برنامهی",
|
357 |
+
"مادرشوهردارم": "مادر شوهر داردم",
|
358 |
+
"سیاهپوستان": "سیاه پوستان",
|
359 |
+
"شرکتهایی": "شرکتهایی",
|
360 |
+
"نیاوردهاند": "نیاوردهاند",
|
361 |
+
"آنهم": "آن هم",
|
362 |
+
"شوهرداریم": "شوهر داریم",
|
363 |
+
"یکچهارم": "یک چهارم",
|
364 |
+
"پروندههاست": "پرونده هاست",
|
365 |
+
"برنامهت": "برنامهات",
|
366 |
+
"چروکیدهمان": "چروکیدهمان",
|
367 |
+
"زمینهسازی": "زمینه سازی",
|
368 |
+
"زدهاند": "زدهاند",
|
369 |
+
"اظهارنظرپرداختن": "اظهار نظر پرداختن",
|
370 |
+
"صلحطلبانهترین": "صلح طلبانهترین",
|
371 |
+
"بهغلط": "به غلط",
|
372 |
+
"ایدهآلم": "ایده آلم",
|
373 |
+
"سیاهکاران": "سیاه کاران",
|
374 |
+
"امیرابراهیم": "امیر ابراهیم",
|
375 |
+
"توسعهدهندگان": "توسعه دهندگان",
|
376 |
+
"لحظهی": "لحظهی",
|
377 |
+
"امینطاها": "امین طاها",
|
378 |
+
"بینالنهرین": "بین النهرین",
|
379 |
+
"نیمهوقت": "نیمه وقت",
|
380 |
+
"پیادهروی": "پیاده روی",
|
381 |
+
"آلودهاند": "آلودهاند",
|
382 |
+
"گریهکرد": "گره کرد",
|
383 |
+
"نعمتهایی": "نعمتهایی",
|
384 |
+
"مادرشوهرشماهم": "مادر شوهر شما هم",
|
385 |
+
"آشپزخونهاس": "آشپزخونهاس",
|
386 |
+
"مسابقهها": "مسابقهها",
|
387 |
+
"مسابقهای": "مسابقههای",
|
388 |
+
"برنامهریزی": "برنامهریزی",
|
389 |
+
"بازخواهید": "باز خواهید",
|
390 |
+
"جوییما": "جویی ما",
|
391 |
+
"آماده ایم": "آمادهایم",
|
392 |
+
"مدلسازی": "مدلسازی",
|
393 |
+
"درصورتیکه": "در صورتیکه",
|
394 |
+
"آمریکاییات": "آمریکاییات",
|
395 |
+
"مادریاش": "مادریاش",
|
396 |
+
"غافلگیرکننده": "غافلگیر کننده",
|
397 |
+
"پیکرتراشی": "پیکر تراشی",
|
398 |
+
"اذیتوآزار": "اذیت و آزار",
|
399 |
+
"امتیازاورترین": "امتیاز آور",
|
400 |
+
"جیکجیک": "جیک جیک",
|
401 |
+
"تاشب": "تا شب",
|
402 |
+
"کپیرایت": "کپی رایت",
|
403 |
+
"آنتیبادی": "آنتی بادی",
|
404 |
+
"عجیبتر": "عجیبتر",
|
405 |
+
"استانداردسازی": "استاندارد سازی",
|
406 |
+
"هشتادوهشت": "هشتاد و هشت",
|
407 |
+
"متنوعتر": "متنوعتر",
|
408 |
+
"منظورانجام": "منظور انجام",
|
409 |
+
"نگرانکنندهترین": "نگران کنندهترین",
|
410 |
+
"شگفتانگیز": "شگفت انگیز",
|
411 |
+
"رنگینپوست": "رنگین پو��ت",
|
412 |
+
"فارغ التحصیلان": "فارغالتحصیلان",
|
413 |
+
"ترسناکتر": "ترسناکتر",
|
414 |
+
"لا رامبلا": "لارامبلا",
|
415 |
+
"پرجمعیتترین": "پرجمعیتترین",
|
416 |
+
"درمیآیند": "درمیآیند",
|
417 |
+
"باشمالکی": "باشم الکی",
|
418 |
+
"وسیعتر": "وسیعتر",
|
419 |
+
"فاحشهخانه": "فاحشه خانه",
|
420 |
+
"بااحتیاط": "با احتیاط",
|
421 |
+
"قانعکننده": "قانعکننده",
|
422 |
+
"انعطافپذیری": "انعطافپذیری",
|
423 |
+
"بیتالمقدس": "بیتالمقدس",
|
424 |
+
"اوپناستریتمپ": "اوپن استریت مپ",
|
425 |
+
"روزابارونی": "روزا بارونی",
|
426 |
+
"محافظهکارانه": "محافظه کارانه",
|
427 |
+
"فوتبالدستی": "فوتبال دستی",
|
428 |
+
"توسعهدهنده": "توسعه دهنده",
|
429 |
+
"قانونگزاران": "قانون گزاران",
|
430 |
+
"العسریسرا": "العسر یسرا",
|
431 |
+
"خارقالعاده": "خارقالعاده",
|
432 |
+
"بیماریمزمن": "بیماری مزمن",
|
433 |
+
"بادوستانتان": "با دوستانتان",
|
434 |
+
"برابربیشتر": "برابر بیشتر",
|
435 |
+
"ارائهدهنده": "ارائه دهنده",
|
436 |
+
"طوفانزدگان": "طوفان زندگان",
|
437 |
+
"امینمحمد": "امین محمد",
|
438 |
+
"محیطزیست": "محیط زیست",
|
439 |
+
"شقیترینشان": "شقیترینشان",
|
440 |
+
"بودواقعا": "بود واقعا",
|
441 |
+
"نیویورکتایمز": "نیویورک تایمز",
|
442 |
+
"ریودوژانیرو": "ریو دو ژانیرو",
|
443 |
+
"مشترکالمنافع": "مشترکالمنافع",
|
444 |
+
"اسلایدسازم": "اسلاید سازم",
|
445 |
+
"نمیآوریدش": "نمیآوریدش",
|
446 |
+
"بینالملل": "بینالملل",
|
447 |
+
"مصرفکنندگان": "مصرف کنندگان",
|
448 |
+
"امینالدین": "امین الدین",
|
449 |
+
"امریکااینقدر": "امریکا اینقدر",
|
450 |
+
"بعضیاوقات": "بعضی اوقات",
|
451 |
+
"خاطربچه": "خاطر بچه",
|
452 |
+
"ایناکیلویی": "اینا کیلویی",
|
453 |
+
"ویکیپدیا": "ویکیپدیا",
|
454 |
+
"مافکرمیکنیم": "ما فکر میکنیم",
|
455 |
+
"انگلیسیزبان": "انگلیسی زبان",
|
456 |
+
"کلهشون": "کلهشون",
|
457 |
+
"آدمبزرگی": "آرم بزرگی",
|
458 |
+
"مر آت مر آه": "مرآت مرآت",
|
459 |
+
"آسیبزد": "آسیب زد",
|
460 |
+
"آیآرسی": "آی آرسی",
|
461 |
+
"آسیااقیانوسیه": "آسیا اقیانوسیه",
|
462 |
+
"آیای": "آیا",
|
463 |
+
"میانجنسی": "میان جنسی",
|
464 |
+
"میاننسلی": "میان نسلی",
|
465 |
+
"میانافزارها": "میان افزارها",
|
466 |
+
"آییننامه": "آییننامه",
|
467 |
+
"ارائهشده": "ارائهشده",
|
468 |
+
"اشپزخونه": "آشپزخونه",
|
469 |
+
"اماعلتشونمیپرسه": "اما علتشو نمیپرسه",
|
470 |
+
"امیدوارکننده": "امیدوار کننده",
|
471 |
+
"ایالاتمتحده": "ایالات متحده",
|
472 |
+
"بااینکه": "با اینکه",
|
473 |
+
"بلندپروازانه": "بلند پروازانه",
|
474 |
+
"بهترازاینه": "بهتر از اینه",
|
475 |
+
"بهدستآمده": "به دستآمده",
|
476 |
+
"بهوسیله": "به وسیله",
|
477 |
+
"بیادبانه": "بی ادبانه",
|
478 |
+
"بیاندازه": "بی اندازه",
|
479 |
+
"بیصبرانه": "بی صبرانه",
|
480 |
+
"بیفایده": "بی فایده",
|
481 |
+
"بیمهره": "بی مهره",
|
482 |
+
"بینظیره": "بی نظیره",
|
483 |
+
"تاریخزده": "تاریخ زده",
|
484 |
+
"تهرانزده": "تهران زده",
|
485 |
+
"تولیدشده": "تولید شده",
|
486 |
+
"تولیدکننده": "تولید کننده",
|
487 |
+
"تکمیلشده": "تکمیل شده",
|
488 |
+
"جاافتاده": "جا افتاده",
|
489 |
+
"جمعآوریکننده": "جمع آوری کننده",
|
490 |
+
"جورآدمیه": "جور آدمیه",
|
491 |
+
"حقالزحمه": "حق الزحمه",
|
492 |
+
"دخترونهتره": "دخترونه تره",
|
493 |
+
"دوپنجره": "دو پنجره",
|
494 |
+
"ذاتالریه": "ذاتالریه",
|
495 |
+
"راسالخیمه": "راسالخیمه",
|
496 |
+
"رنگماده": "رنگ ماده",
|
497 |
+
"سوئاستفاده": "سو استفاده",
|
498 |
+
"سواستفاده": "سو استفاده",
|
499 |
+
"شبهجزیره": "شبه جزیره",
|
500 |
+
"صادرکننده": "صادر کننده",
|
501 |
+
"ضررداره": "ضرر داره",
|
502 |
+
"عابرپیاده": "عابر پیاده",
|
503 |
+
"فوقالعاده": "فوقالعاده",
|
504 |
+
"قابلتوجه": "قابل توجه",
|
505 |
+
"قانعکننده": "قانع کننده",
|
506 |
+
"مادربیچاره": "مادر بیچاره",
|
507 |
+
"مشخصشده": "مشخص شده",
|
508 |
+
"مصرفکننده": "مصرف کننده",
|
509 |
+
"مصیبتزده": "مصیب تزده",
|
510 |
+
"ناامیدکننده": "ناامید کننده",
|
511 |
+
"نیمفاصله": "نیمفاصله",
|
512 |
+
"هماهنگکننده": "هماهنگ کننده",
|
513 |
+
"همهجانبه": "همه جانبه",
|
514 |
+
"واردکننده": "وارد کننده",
|
515 |
+
"وخوابگاه": "و خوابگاه",
|
516 |
+
"ودستگاه": "و دستگاه",
|
517 |
+
"وزردچوبه": "و زردچوبه",
|
518 |
+
"وپروانه": "و پروانه",
|
519 |
+
"پدرخوانده": "پدر خوانده",
|
520 |
+
"چاپشده": "چاپ شده",
|
521 |
+
"کردته": "کرد ته",
|
522 |
+
"کردندکه": "کردند که",
|
523 |
+
"یکطرفه": "یک طرفه",
|
524 |
+
"پایینتره": "پایینتره",
|
525 |
+
"اشتراکگذاری": "اشتراک گذاری",
|
526 |
+
"انحصارگراناند": "انحصار گراناند",
|
527 |
+
"خوشحالییییی": "خوشحالی",
|
528 |
+
"همتیمیهایشان": "هم تیمیهایشان",
|
529 |
+
"پایدارامباید": "پایدارام باید",
|
530 |
+
"پرجنبوجوشتر": "پر جنب و جوشتر",
|
531 |
+
"آبمروارید": "آب مروارید",
|
532 |
+
"آتشسوزی": "آتش سوزی",
|
533 |
+
"آتشنشانی": "آتشنشانی",
|
534 |
+
"آتشنشان": "آتشنشان",
|
535 |
+
"آرامشبخش": "آرامش بخش",
|
536 |
+
"آشناداشتن": "آشنا داشتن",
|
537 |
+
"آقاچیزی": "آقا چیزی",
|
538 |
+
"آموختهام": "آموختهام",
|
539 |
+
"آموزششان": "آموزششان",
|
540 |
+
"ازآنجا": "از آنجا",
|
541 |
+
"ازالان": "از الان",
|
542 |
+
"ازاینجا": "از اینجا",
|
543 |
+
"ازجیبش": "از جیبش",
|
544 |
+
"ازدستش": "از دستش",
|
545 |
+
"ازدیوار": "از دیوار",
|
546 |
+
"ازشغلشون": "از شغلشون",
|
547 |
+
"ازوقتی": "از وقتی",
|
548 |
+
"ازکسانی": "از کسانی",
|
549 |
+
"اسباببازی": "اسباب بازی",
|
550 |
+
"اسبسوار": "اسب سوار",
|
551 |
+
"اصیلزاده": "اصیل زاده",
|
552 |
+
"افتادهاید": "افتادهاید",
|
553 |
+
"الهام": "الهام",
|
554 |
+
"امااصلا": "اما اصلا",
|
555 |
+
"امااصلابه": "اما اصلا به",
|
556 |
+
"امااین": "اما این",
|
557 |
+
"امابعد": "اما بعد",
|
558 |
+
"امابعدیکی": "اما بعد یکی",
|
559 |
+
"اماجاذبه": "اما جاذبه",
|
560 |
+
"امرارمعاش": "امرار معاش",
|
561 |
+
"امکانپذیر": "امکان پذیر",
|
562 |
+
"انتهای": "انتهای",
|
563 |
+
"انتهایی": "انتهایی",
|
564 |
+
"ایزدبانوی": "ایزد بانوی",
|
565 |
+
"بااینحال": "با اینحال",
|
566 |
+
"باحتمال": "به احتمال",
|
567 |
+
"باحجاب": "با حجاب",
|
568 |
+
"باخنده": "با خنده",
|
569 |
+
"بادوستاش": "با دوستاش",
|
570 |
+
"بارمان": "بار مان",
|
571 |
+
"بازتر": "باز تر",
|
572 |
+
"باطعنه": "با طعنه",
|
573 |
+
"بافریاد": "با فریاد",
|
574 |
+
"بارگزاری": "بارگذاری",
|
575 |
+
"بالامنم": "بالا منم",
|
576 |
+
"بگیرمامان": "بگیر مامان",
|
577 |
+
"بیاحترامی": "بی احترامی",
|
578 |
+
"بیادبی": "بی ادبی",
|
579 |
+
"بیاعتنا": "بی اعتنا",
|
580 |
+
"بیدارباش": "بیدار باش",
|
581 |
+
"بیشازحد": "بیش از حد",
|
582 |
+
"بیمسئولیت": "بی مسئولیت",
|
583 |
+
"تاسفبار": "تاسف بار",
|
584 |
+
"تامشکلمون": "تا مشکلمون",
|
585 |
+
"تانقشه": "تا نقشه",
|
586 |
+
"تصمیمگیری": "تصمیم گیری",
|
587 |
+
"تقسیمبندی": "تقسیم بندی",
|
588 |
+
"تقصیرارو": "تقصیرا رو",
|
589 |
+
"جدیدابرای": "جدیدا برای",
|
590 |
+
"جعبهابزار": "جعبه ابزار",
|
591 |
+
"جلوتونو": "جلو تو نو",
|
592 |
+
"حاضردر": "حاضر در",
|
593 |
+
"حاضرنیست": "حاضر نیست",
|
594 |
+
"دستنخورده": "دست نخورده",
|
595 |
+
"دوامتیاز": "دو امتیاز",
|
596 |
+
"دوروزتمام": "دو روز تمام",
|
597 |
+
"شخصیسازی": "شخصیسازی",
|
598 |
+
"شدواجناس": "شد و اجناس",
|
599 |
+
"شوهردارم": "شوهر دارم",
|
600 |
+
"شوهرشماهم": "شوهر شما هم",
|
601 |
+
"شوهرمحترم": "شوهر محترم",
|
602 |
+
"شکلگیری": "شکل گیری",
|
603 |
+
"صخرهنوردی": "صخرهنوردی",
|
604 |
+
"صدوبیست": "صد و بیست",
|
605 |
+
"عقبنشینی": "عقب نشینی",
|
606 |
+
"عکسالعمل": "عکسالعمل",
|
607 |
+
"غرغرمیکنم": "غرغر میکنم",
|
608 |
+
"هزاربار": "هزار بار",
|
609 |
+
"هزارتومان": "هزار تومان",
|
610 |
+
"هزارجور": "هزار جور",
|
611 |
+
"هزاروسیصد": "هزار و سیصد",
|
612 |
+
"هممیهنان": "هم میهنان",
|
613 |
+
"هممیهنانش": "هم میهنانش",
|
614 |
+
"همنسلانش": "هم نسلانش",
|
615 |
+
"همهگیری": "همه گیری",
|
616 |
+
"هییییچ": "هیچ",
|
617 |
+
"وقتاخیلی": "وقتا خیلی",
|
618 |
+
"وقتابه": "وقتا به",
|
619 |
+
"وقتگذرانی": "وقت گذرانی",
|
620 |
+
"ومحکوم": "و محکوم",
|
621 |
+
"ومحیطها": "و محیطها",
|
622 |
+
"وکشورتان": "و کشورتان",
|
623 |
+
"ویکیمدیا": "ویکیمدی��",
|
624 |
+
"یهوگفت": "یهو گفت",
|
625 |
+
"اینجااز": "اینجا از",
|
626 |
+
}
|
627 |
+
fixator_dictionary = {
|
628 |
+
"بهای": "بهای",
|
629 |
+
"بهترین": "بهترین",
|
630 |
+
"آستر": "آستر",
|
631 |
+
"ارکستر": "ارکستر",
|
632 |
+
"انتر": "انتر",
|
633 |
+
"بستر": "بستر",
|
634 |
+
"بهتر": "بهتر",
|
635 |
+
"بهترتر": "بهترتر",
|
636 |
+
"توئیتر": "تویتتر",
|
637 |
+
"توییتر": "توییتر",
|
638 |
+
"تیتر": "تیتر",
|
639 |
+
"دختر": "دختر",
|
640 |
+
"دفتر": "دفتر",
|
641 |
+
"دلستر": "دلستر",
|
642 |
+
"دکتر": "دکتر",
|
643 |
+
"شتر": "شتر",
|
644 |
+
"لیتر": "لیتر",
|
645 |
+
"متر": "متر",
|
646 |
+
"هیپستر": "هیپستر",
|
647 |
+
"پیتر": "پیتر",
|
648 |
+
"چتر": "چتر",
|
649 |
+
"کمتر": "کمتر",
|
650 |
+
"گنگستر": "گنگستر",
|
651 |
+
"انگشتر": "انگشتر",
|
652 |
+
"سنتر": "سنتر",
|
653 |
+
"تویتتر": "توییتر",
|
654 |
+
"مادهشتر": "ماده شتر",
|
655 |
+
"ویترین": "ویترین",
|
656 |
+
"کرونومتر": "کرنومتر",
|
657 |
+
"کهتر": "کهتر",
|
658 |
+
"فیلتر": "فیلتر",
|
659 |
+
"الهام": "الهام",
|
660 |
+
"آلمان": "آلمان",
|
661 |
+
"انتهای": "انتهای",
|
662 |
+
"انتهایی": "انتهایی",
|
663 |
+
"آموختهام": "آموختهام",
|
664 |
+
}
|
normalizer.py
ADDED
@@ -0,0 +1,203 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from parsivar import Normalizer
|
2 |
+
|
3 |
+
import num2fawords
|
4 |
+
import re
|
5 |
+
import string
|
6 |
+
|
7 |
+
|
8 |
+
_normalizer = Normalizer(half_space_char="\u200c", statistical_space_correction=True)
|
9 |
+
chars_to_ignore = [
|
10 |
+
",", "?", ".", "!", "-", ";", ":", '""', "%", "'", '"', "�",
|
11 |
+
"#", "!", "؟", "?", "«", "»", "،", "(", ")", "؛", "'ٔ", "٬", 'ٔ', ",", "?",
|
12 |
+
".", "!", "-", ";", ":", '"', "“", "%", "‘", "”", "�", "–", "…", "_", "”", '“', '„',
|
13 |
+
'ā', 'š', 'ّ', 'ْ',
|
14 |
+
]
|
15 |
+
chars_to_ignore = chars_to_ignore + list(string.ascii_lowercase + string.digits)
|
16 |
+
chars_to_ignore = f"""[{"".join(chars_to_ignore)}]"""
|
17 |
+
zwnj = "\u200c"
|
18 |
+
silent_chars = ["ا", "د", "ذ", "ر", "ز", "و", "آ"] + [zwnj] + [" "]
|
19 |
+
|
20 |
+
|
21 |
+
def multiple_replace(text, chars_to_mapping):
|
22 |
+
pattern = "|".join(map(re.escape, chars_to_mapping.keys()))
|
23 |
+
return re.sub(pattern, lambda m: chars_to_mapping[m.group()], str(text))
|
24 |
+
|
25 |
+
|
26 |
+
def remove_special_characters(text, chars_to_ignore_regex):
|
27 |
+
text = re.sub(chars_to_ignore_regex, '', text).lower() + " "
|
28 |
+
return text
|
29 |
+
|
30 |
+
|
31 |
+
def convert_word_nums_to_text(word):
|
32 |
+
try:
|
33 |
+
word = int(word)
|
34 |
+
word = num2fawords.words(word)
|
35 |
+
except:
|
36 |
+
word = word
|
37 |
+
|
38 |
+
return word
|
39 |
+
|
40 |
+
|
41 |
+
def normalizer_at_word_level(text):
|
42 |
+
words = text.split()
|
43 |
+
_text = []
|
44 |
+
|
45 |
+
for word in words:
|
46 |
+
word = convert_word_nums_to_text(word)
|
47 |
+
word = fixator_dictionary.get(word, word)
|
48 |
+
|
49 |
+
_text.append(word)
|
50 |
+
|
51 |
+
return " ".join(_text) + " "
|
52 |
+
|
53 |
+
|
54 |
+
def finder(ss, s, starter=False):
|
55 |
+
found = []
|
56 |
+
for m in re.finditer(ss, s):
|
57 |
+
if starter:
|
58 |
+
found.append(m.start())
|
59 |
+
else:
|
60 |
+
found.append((m.start(), m.end()))
|
61 |
+
|
62 |
+
return found
|
63 |
+
|
64 |
+
|
65 |
+
def substring_replace(ss, s, start, end, stripped=True):
|
66 |
+
s_start = s[:start]
|
67 |
+
s_end = s[end:]
|
68 |
+
|
69 |
+
counter = 0
|
70 |
+
if stripped:
|
71 |
+
counter = 1 if s_start.endswith(" ") else counter
|
72 |
+
s_start = s_start.rstrip()
|
73 |
+
|
74 |
+
return s_start + ss + s_end, counter
|
75 |
+
|
76 |
+
|
77 |
+
def normalizer(
|
78 |
+
batch,
|
79 |
+
is_normalize=True,
|
80 |
+
return_dict=True,
|
81 |
+
filter_trivials=False,
|
82 |
+
remove_extra_space=False
|
83 |
+
):
|
84 |
+
text = batch["sentence"].lower().strip()
|
85 |
+
|
86 |
+
# Parsivar normalizer
|
87 |
+
if is_normalize:
|
88 |
+
text = _normalizer.normalize(text)
|
89 |
+
|
90 |
+
# Dictionary mapping
|
91 |
+
text = multiple_replace(text, dictionary_mapping)
|
92 |
+
text = re.sub(" +", " ", text)
|
93 |
+
|
94 |
+
# Remove specials
|
95 |
+
text = remove_special_characters(text, chars_to_ignore)
|
96 |
+
text = re.sub(" +", " ", text)
|
97 |
+
|
98 |
+
# Replace connected آ
|
99 |
+
special, pointer = "آ", int("0")
|
100 |
+
for f in sorted(finder(special, text, True)):
|
101 |
+
index = f + pointer - 1
|
102 |
+
if len(text) >= index:
|
103 |
+
if text[index] not in silent_chars:
|
104 |
+
new_text, extra_pointer = substring_replace(
|
105 |
+
f"{text[index]}{zwnj}", text, index, index + 1, stripped=True)
|
106 |
+
text = new_text
|
107 |
+
pointer += 1 + 1 - 1 - extra_pointer
|
108 |
+
|
109 |
+
# Replace connected ها
|
110 |
+
pointer = int("0")
|
111 |
+
special_list = [
|
112 |
+
# "ام", "ای", "است", "ایم", "اید", "اند",
|
113 |
+
"هایمان", "هایم", "هایت", "هایش",
|
114 |
+
"هایتان", "هایشان", "هام", "هات",
|
115 |
+
"هاتان", "هامون", "هامان", "هاش",
|
116 |
+
"هاتون", "هاشان", "هاشون",
|
117 |
+
"هایی", "های", "هاس", "ها"
|
118 |
+
]
|
119 |
+
for special in special_list:
|
120 |
+
pointer = 0
|
121 |
+
text = text
|
122 |
+
for f in sorted(finder(special, text, False)):
|
123 |
+
start, end = f[0] + pointer - 1, f[1] + pointer - 1
|
124 |
+
if len(text) >= (end + 1):
|
125 |
+
if len(text) == (end + 1):
|
126 |
+
new_text, extra_pointer = substring_replace(
|
127 |
+
f"{zwnj}{special}",
|
128 |
+
text,
|
129 |
+
start + 1,
|
130 |
+
end + 1,
|
131 |
+
stripped=True)
|
132 |
+
text = new_text
|
133 |
+
pointer += 1 + 1 - 1 - extra_pointer
|
134 |
+
else:
|
135 |
+
if text[end + 1] == " ":
|
136 |
+
new_text, extra_pointer = substring_replace(
|
137 |
+
f"{zwnj}{special}",
|
138 |
+
text,
|
139 |
+
start + 1,
|
140 |
+
end + 1,
|
141 |
+
stripped=True)
|
142 |
+
text = new_text
|
143 |
+
pointer += 1 + 1 - 1 - extra_pointer
|
144 |
+
|
145 |
+
special, pointer = "افزار", int("0")
|
146 |
+
for f in sorted(finder(special, text, False)):
|
147 |
+
start, end = f[0] + pointer - 1, f[1] + pointer - 1
|
148 |
+
|
149 |
+
if len(text) >= (end + 1):
|
150 |
+
new_text, extra_pointer = substring_replace(f"{zwnj}{special}", text, start + 1, end + 1, stripped=True)
|
151 |
+
text = new_text
|
152 |
+
pointer += 1 + 1 - 1 - extra_pointer
|
153 |
+
|
154 |
+
# Replace connected ها
|
155 |
+
pointer = int("0")
|
156 |
+
special_list = [
|
157 |
+
"ترین", "تر"
|
158 |
+
]
|
159 |
+
for special in special_list:
|
160 |
+
pointer = 0
|
161 |
+
text = text
|
162 |
+
for f in sorted(finder(special, text, False)):
|
163 |
+
start, end = f[0] + pointer - 1, f[1] + pointer - 1
|
164 |
+
if len(text) >= (end + 1):
|
165 |
+
if len(text) == (end + 1):
|
166 |
+
new_text, extra_pointer = substring_replace(
|
167 |
+
f"{zwnj}{special}",
|
168 |
+
text,
|
169 |
+
start + 1,
|
170 |
+
end + 1,
|
171 |
+
stripped=True)
|
172 |
+
text = new_text
|
173 |
+
pointer += 1 + 1 - 1 - extra_pointer
|
174 |
+
else:
|
175 |
+
if text[end + 1] == " ":
|
176 |
+
new_text, extra_pointer = substring_replace(
|
177 |
+
f"{zwnj}{special}",
|
178 |
+
text,
|
179 |
+
start + 1,
|
180 |
+
end + 1,
|
181 |
+
stripped=True)
|
182 |
+
text = new_text
|
183 |
+
pointer += 1 + 1 - 1 - extra_pointer
|
184 |
+
|
185 |
+
# Normalizer at word level
|
186 |
+
text = normalizer_at_word_level(text)
|
187 |
+
text = re.sub(" +", " ", text)
|
188 |
+
|
189 |
+
if remove_extra_space:
|
190 |
+
text = text.strip()
|
191 |
+
else:
|
192 |
+
text = text.strip() + " "
|
193 |
+
|
194 |
+
if filter_trivials:
|
195 |
+
if not len(text) > 2:
|
196 |
+
text = None
|
197 |
+
|
198 |
+
if not return_dict:
|
199 |
+
return text
|
200 |
+
|
201 |
+
batch["sentence"] = text
|
202 |
+
return batch
|
203 |
+
|
predictions.csv
ADDED
The diff for this file is too large to render.
See raw diff
|
|
preprocessor_config.json
ADDED
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"do_normalize": true,
|
3 |
+
"feature_extractor_type": "Wav2Vec2FeatureExtractor",
|
4 |
+
"feature_size": 1,
|
5 |
+
"padding_side": "right",
|
6 |
+
"padding_value": 0.0,
|
7 |
+
"return_attention_mask": true,
|
8 |
+
"sampling_rate": 16000
|
9 |
+
}
|
pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d6d33fc8e01332e2e4cf0230c159f63d02620a0528c19c69c3ce7d58777223e2
|
3 |
+
size 1262097815
|
sample1.flac
ADDED
Binary file (44.5 kB). View file
|
|
sample2978.flac
ADDED
Binary file (193 kB). View file
|
|
sample5168.flac
ADDED
Binary file (70.1 kB). View file
|
|
special_tokens_map.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "<unk>", "pad_token": "<pad>"}
|
tf_model.h5
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:3eb1167b6d301f201da515819ef0bfb697821a4db0e6d5b6dcb1ea2baa3f0aa6
|
3 |
+
size 1262429728
|
tokenizer_config.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"unk_token": "<unk>", "bos_token": "<s>", "eos_token": "</s>", "pad_token": "<pad>", "do_lower_case": false, "word_delimiter_token": "|", "special_tokens_map_file": "/workspace/output_models/fa/xlsr-fa/special_tokens_map.json", "tokenizer_file": null, "name_or_path": "/workspace/output_models/fa/xlsr-fa"}
|
trainer_state.json
ADDED
@@ -0,0 +1,3341 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": null,
|
3 |
+
"best_model_checkpoint": null,
|
4 |
+
"epoch": 30.0,
|
5 |
+
"global_step": 71280,
|
6 |
+
"is_hyper_param_search": false,
|
7 |
+
"is_local_process_zero": true,
|
8 |
+
"is_world_process_zero": true,
|
9 |
+
"log_history": [
|
10 |
+
{
|
11 |
+
"epoch": 0.13,
|
12 |
+
"learning_rate": 0.0002472,
|
13 |
+
"loss": 7.355,
|
14 |
+
"step": 300
|
15 |
+
},
|
16 |
+
{
|
17 |
+
"epoch": 0.13,
|
18 |
+
"eval_loss": 3.007211446762085,
|
19 |
+
"eval_runtime": 459.0118,
|
20 |
+
"eval_samples_per_second": 11.357,
|
21 |
+
"eval_wer": 1.0,
|
22 |
+
"step": 300
|
23 |
+
},
|
24 |
+
{
|
25 |
+
"epoch": 0.25,
|
26 |
+
"learning_rate": 0.0004114179146651596,
|
27 |
+
"loss": 1.4963,
|
28 |
+
"step": 600
|
29 |
+
},
|
30 |
+
{
|
31 |
+
"epoch": 0.25,
|
32 |
+
"eval_loss": 0.6261181235313416,
|
33 |
+
"eval_runtime": 447.0386,
|
34 |
+
"eval_samples_per_second": 11.661,
|
35 |
+
"eval_wer": 0.5529420917507326,
|
36 |
+
"step": 600
|
37 |
+
},
|
38 |
+
{
|
39 |
+
"epoch": 0.38,
|
40 |
+
"learning_rate": 0.0004096716586606386,
|
41 |
+
"loss": 0.5139,
|
42 |
+
"step": 900
|
43 |
+
},
|
44 |
+
{
|
45 |
+
"epoch": 0.38,
|
46 |
+
"eval_loss": 0.47113755345344543,
|
47 |
+
"eval_runtime": 456.1265,
|
48 |
+
"eval_samples_per_second": 11.429,
|
49 |
+
"eval_wer": 0.445372267316719,
|
50 |
+
"step": 900
|
51 |
+
},
|
52 |
+
{
|
53 |
+
"epoch": 0.51,
|
54 |
+
"learning_rate": 0.00040792540265611755,
|
55 |
+
"loss": 0.4189,
|
56 |
+
"step": 1200
|
57 |
+
},
|
58 |
+
{
|
59 |
+
"epoch": 0.51,
|
60 |
+
"eval_loss": 0.4321776032447815,
|
61 |
+
"eval_runtime": 431.3516,
|
62 |
+
"eval_samples_per_second": 12.085,
|
63 |
+
"eval_wer": 0.4174684266486865,
|
64 |
+
"step": 1200
|
65 |
+
},
|
66 |
+
{
|
67 |
+
"epoch": 0.63,
|
68 |
+
"learning_rate": 0.00040617914665159647,
|
69 |
+
"loss": 0.3815,
|
70 |
+
"step": 1500
|
71 |
+
},
|
72 |
+
{
|
73 |
+
"epoch": 0.63,
|
74 |
+
"eval_loss": 0.3827853798866272,
|
75 |
+
"eval_runtime": 459.5456,
|
76 |
+
"eval_samples_per_second": 11.344,
|
77 |
+
"eval_wer": 0.3828998210627318,
|
78 |
+
"step": 1500
|
79 |
+
},
|
80 |
+
{
|
81 |
+
"epoch": 0.76,
|
82 |
+
"learning_rate": 0.00040443289064707544,
|
83 |
+
"loss": 0.3504,
|
84 |
+
"step": 1800
|
85 |
+
},
|
86 |
+
{
|
87 |
+
"epoch": 0.76,
|
88 |
+
"eval_loss": 0.3574122190475464,
|
89 |
+
"eval_runtime": 457.446,
|
90 |
+
"eval_samples_per_second": 11.396,
|
91 |
+
"eval_wer": 0.3792951427608205,
|
92 |
+
"step": 1800
|
93 |
+
},
|
94 |
+
{
|
95 |
+
"epoch": 0.88,
|
96 |
+
"learning_rate": 0.00040268663464255436,
|
97 |
+
"loss": 0.3385,
|
98 |
+
"step": 2100
|
99 |
+
},
|
100 |
+
{
|
101 |
+
"epoch": 0.88,
|
102 |
+
"eval_loss": 0.3647058308124542,
|
103 |
+
"eval_runtime": 432.4987,
|
104 |
+
"eval_samples_per_second": 12.053,
|
105 |
+
"eval_wer": 0.3572002800757242,
|
106 |
+
"step": 2100
|
107 |
+
},
|
108 |
+
{
|
109 |
+
"epoch": 1.01,
|
110 |
+
"learning_rate": 0.00040094037863803333,
|
111 |
+
"loss": 0.3205,
|
112 |
+
"step": 2400
|
113 |
+
},
|
114 |
+
{
|
115 |
+
"epoch": 1.01,
|
116 |
+
"eval_loss": 0.3391036093235016,
|
117 |
+
"eval_runtime": 460.5432,
|
118 |
+
"eval_samples_per_second": 11.319,
|
119 |
+
"eval_wer": 0.34612691579575217,
|
120 |
+
"step": 2400
|
121 |
+
},
|
122 |
+
{
|
123 |
+
"epoch": 1.14,
|
124 |
+
"learning_rate": 0.0003991941226335123,
|
125 |
+
"loss": 0.2697,
|
126 |
+
"step": 2700
|
127 |
+
},
|
128 |
+
{
|
129 |
+
"epoch": 1.14,
|
130 |
+
"eval_loss": 0.34452542662620544,
|
131 |
+
"eval_runtime": 469.56,
|
132 |
+
"eval_samples_per_second": 11.102,
|
133 |
+
"eval_wer": 0.3170042270688001,
|
134 |
+
"step": 2700
|
135 |
+
},
|
136 |
+
{
|
137 |
+
"epoch": 1.26,
|
138 |
+
"learning_rate": 0.0003974478666289912,
|
139 |
+
"loss": 0.2562,
|
140 |
+
"step": 3000
|
141 |
+
},
|
142 |
+
{
|
143 |
+
"epoch": 1.26,
|
144 |
+
"eval_loss": 0.32179856300354004,
|
145 |
+
"eval_runtime": 433.2725,
|
146 |
+
"eval_samples_per_second": 12.032,
|
147 |
+
"eval_wer": 0.31285495708098854,
|
148 |
+
"step": 3000
|
149 |
+
},
|
150 |
+
{
|
151 |
+
"epoch": 1.39,
|
152 |
+
"learning_rate": 0.0003957016106244702,
|
153 |
+
"loss": 0.2514,
|
154 |
+
"step": 3300
|
155 |
+
},
|
156 |
+
{
|
157 |
+
"epoch": 1.39,
|
158 |
+
"eval_loss": 0.3245706558227539,
|
159 |
+
"eval_runtime": 453.9529,
|
160 |
+
"eval_samples_per_second": 11.484,
|
161 |
+
"eval_wer": 0.30668291797411895,
|
162 |
+
"step": 3300
|
163 |
+
},
|
164 |
+
{
|
165 |
+
"epoch": 1.52,
|
166 |
+
"learning_rate": 0.0003939553546199491,
|
167 |
+
"loss": 0.2545,
|
168 |
+
"step": 3600
|
169 |
+
},
|
170 |
+
{
|
171 |
+
"epoch": 1.52,
|
172 |
+
"eval_loss": 0.32298558950424194,
|
173 |
+
"eval_runtime": 460.4237,
|
174 |
+
"eval_samples_per_second": 11.322,
|
175 |
+
"eval_wer": 0.30904281527968674,
|
176 |
+
"step": 3600
|
177 |
+
},
|
178 |
+
{
|
179 |
+
"epoch": 1.64,
|
180 |
+
"learning_rate": 0.0003922090986154281,
|
181 |
+
"loss": 0.2457,
|
182 |
+
"step": 3900
|
183 |
+
},
|
184 |
+
{
|
185 |
+
"epoch": 1.64,
|
186 |
+
"eval_loss": 0.31425100564956665,
|
187 |
+
"eval_runtime": 459.4801,
|
188 |
+
"eval_samples_per_second": 11.345,
|
189 |
+
"eval_wer": 0.2933274552008506,
|
190 |
+
"step": 3900
|
191 |
+
},
|
192 |
+
{
|
193 |
+
"epoch": 1.77,
|
194 |
+
"learning_rate": 0.000390462842610907,
|
195 |
+
"loss": 0.2458,
|
196 |
+
"step": 4200
|
197 |
+
},
|
198 |
+
{
|
199 |
+
"epoch": 1.77,
|
200 |
+
"eval_loss": 0.312289834022522,
|
201 |
+
"eval_runtime": 428.9645,
|
202 |
+
"eval_samples_per_second": 12.153,
|
203 |
+
"eval_wer": 0.28100930992453516,
|
204 |
+
"step": 4200
|
205 |
+
},
|
206 |
+
{
|
207 |
+
"epoch": 1.89,
|
208 |
+
"learning_rate": 0.00038871658660638597,
|
209 |
+
"loss": 0.2353,
|
210 |
+
"step": 4500
|
211 |
+
},
|
212 |
+
{
|
213 |
+
"epoch": 1.89,
|
214 |
+
"eval_loss": 0.29052308201789856,
|
215 |
+
"eval_runtime": 458.3214,
|
216 |
+
"eval_samples_per_second": 11.374,
|
217 |
+
"eval_wer": 0.27597832006431366,
|
218 |
+
"step": 4500
|
219 |
+
},
|
220 |
+
{
|
221 |
+
"epoch": 2.02,
|
222 |
+
"learning_rate": 0.00038697033060186494,
|
223 |
+
"loss": 0.2316,
|
224 |
+
"step": 4800
|
225 |
+
},
|
226 |
+
{
|
227 |
+
"epoch": 2.02,
|
228 |
+
"eval_loss": 0.3132184147834778,
|
229 |
+
"eval_runtime": 450.6205,
|
230 |
+
"eval_samples_per_second": 11.568,
|
231 |
+
"eval_wer": 0.26967661627032496,
|
232 |
+
"step": 4800
|
233 |
+
},
|
234 |
+
{
|
235 |
+
"epoch": 2.15,
|
236 |
+
"learning_rate": 0.00038522407459734386,
|
237 |
+
"loss": 0.1995,
|
238 |
+
"step": 5100
|
239 |
+
},
|
240 |
+
{
|
241 |
+
"epoch": 2.15,
|
242 |
+
"eval_loss": 0.3037310242652893,
|
243 |
+
"eval_runtime": 424.704,
|
244 |
+
"eval_samples_per_second": 12.274,
|
245 |
+
"eval_wer": 0.2598480329866964,
|
246 |
+
"step": 5100
|
247 |
+
},
|
248 |
+
{
|
249 |
+
"epoch": 2.27,
|
250 |
+
"learning_rate": 0.00038347781859282283,
|
251 |
+
"loss": 0.1934,
|
252 |
+
"step": 5400
|
253 |
+
},
|
254 |
+
{
|
255 |
+
"epoch": 2.27,
|
256 |
+
"eval_loss": 0.2947574853897095,
|
257 |
+
"eval_runtime": 450.7971,
|
258 |
+
"eval_samples_per_second": 11.564,
|
259 |
+
"eval_wer": 0.25870698374004825,
|
260 |
+
"step": 5400
|
261 |
+
},
|
262 |
+
{
|
263 |
+
"epoch": 2.4,
|
264 |
+
"learning_rate": 0.00038173156258830174,
|
265 |
+
"loss": 0.1957,
|
266 |
+
"step": 5700
|
267 |
+
},
|
268 |
+
{
|
269 |
+
"epoch": 2.4,
|
270 |
+
"eval_loss": 0.30417683720588684,
|
271 |
+
"eval_runtime": 454.0214,
|
272 |
+
"eval_samples_per_second": 11.482,
|
273 |
+
"eval_wer": 0.2659682062187184,
|
274 |
+
"step": 5700
|
275 |
+
},
|
276 |
+
{
|
277 |
+
"epoch": 2.53,
|
278 |
+
"learning_rate": 0.0003799853065837807,
|
279 |
+
"loss": 0.198,
|
280 |
+
"step": 6000
|
281 |
+
},
|
282 |
+
{
|
283 |
+
"epoch": 2.53,
|
284 |
+
"eval_loss": 0.2877831757068634,
|
285 |
+
"eval_runtime": 423.4049,
|
286 |
+
"eval_samples_per_second": 12.312,
|
287 |
+
"eval_wer": 0.25683981224553304,
|
288 |
+
"step": 6000
|
289 |
+
},
|
290 |
+
{
|
291 |
+
"epoch": 2.65,
|
292 |
+
"learning_rate": 0.00037823905057925963,
|
293 |
+
"loss": 0.2005,
|
294 |
+
"step": 6300
|
295 |
+
},
|
296 |
+
{
|
297 |
+
"epoch": 2.65,
|
298 |
+
"eval_loss": 0.2968423664569855,
|
299 |
+
"eval_runtime": 449.924,
|
300 |
+
"eval_samples_per_second": 11.586,
|
301 |
+
"eval_wer": 0.25951090480018674,
|
302 |
+
"step": 6300
|
303 |
+
},
|
304 |
+
{
|
305 |
+
"epoch": 2.78,
|
306 |
+
"learning_rate": 0.00037649279457473866,
|
307 |
+
"loss": 0.1994,
|
308 |
+
"step": 6600
|
309 |
+
},
|
310 |
+
{
|
311 |
+
"epoch": 2.78,
|
312 |
+
"eval_loss": 0.2824803590774536,
|
313 |
+
"eval_runtime": 452.2888,
|
314 |
+
"eval_samples_per_second": 11.526,
|
315 |
+
"eval_wer": 0.25056404138896815,
|
316 |
+
"step": 6600
|
317 |
+
},
|
318 |
+
{
|
319 |
+
"epoch": 2.9,
|
320 |
+
"learning_rate": 0.0003747465385702176,
|
321 |
+
"loss": 0.1869,
|
322 |
+
"step": 6900
|
323 |
+
},
|
324 |
+
{
|
325 |
+
"epoch": 2.9,
|
326 |
+
"eval_loss": 0.2733488082885742,
|
327 |
+
"eval_runtime": 446.0548,
|
328 |
+
"eval_samples_per_second": 11.687,
|
329 |
+
"eval_wer": 0.24574051502813724,
|
330 |
+
"step": 6900
|
331 |
+
},
|
332 |
+
{
|
333 |
+
"epoch": 3.03,
|
334 |
+
"learning_rate": 0.0003730002825656965,
|
335 |
+
"loss": 0.1916,
|
336 |
+
"step": 7200
|
337 |
+
},
|
338 |
+
{
|
339 |
+
"epoch": 3.03,
|
340 |
+
"eval_loss": 0.2809160351753235,
|
341 |
+
"eval_runtime": 423.4141,
|
342 |
+
"eval_samples_per_second": 12.312,
|
343 |
+
"eval_wer": 0.24529965509193227,
|
344 |
+
"step": 7200
|
345 |
+
},
|
346 |
+
{
|
347 |
+
"epoch": 3.16,
|
348 |
+
"learning_rate": 0.00037125402656117547,
|
349 |
+
"loss": 0.1675,
|
350 |
+
"step": 7500
|
351 |
+
},
|
352 |
+
{
|
353 |
+
"epoch": 3.16,
|
354 |
+
"eval_loss": 0.28469306230545044,
|
355 |
+
"eval_runtime": 443.2959,
|
356 |
+
"eval_samples_per_second": 11.76,
|
357 |
+
"eval_wer": 0.25012318145276313,
|
358 |
+
"step": 7500
|
359 |
+
},
|
360 |
+
{
|
361 |
+
"epoch": 3.28,
|
362 |
+
"learning_rate": 0.0003695077705566544,
|
363 |
+
"loss": 0.162,
|
364 |
+
"step": 7800
|
365 |
+
},
|
366 |
+
{
|
367 |
+
"epoch": 3.28,
|
368 |
+
"eval_loss": 0.2753625810146332,
|
369 |
+
"eval_runtime": 450.2372,
|
370 |
+
"eval_samples_per_second": 11.578,
|
371 |
+
"eval_wer": 0.23324083918985503,
|
372 |
+
"step": 7800
|
373 |
+
},
|
374 |
+
{
|
375 |
+
"epoch": 3.41,
|
376 |
+
"learning_rate": 0.00036776151455213336,
|
377 |
+
"loss": 0.1595,
|
378 |
+
"step": 8100
|
379 |
+
},
|
380 |
+
{
|
381 |
+
"epoch": 3.41,
|
382 |
+
"eval_loss": 0.2853989899158478,
|
383 |
+
"eval_runtime": 425.7608,
|
384 |
+
"eval_samples_per_second": 12.244,
|
385 |
+
"eval_wer": 0.23324083918985503,
|
386 |
+
"step": 8100
|
387 |
+
},
|
388 |
+
{
|
389 |
+
"epoch": 3.54,
|
390 |
+
"learning_rate": 0.0003660152585476123,
|
391 |
+
"loss": 0.1658,
|
392 |
+
"step": 8400
|
393 |
+
},
|
394 |
+
{
|
395 |
+
"epoch": 3.54,
|
396 |
+
"eval_loss": 0.28348785638809204,
|
397 |
+
"eval_runtime": 450.477,
|
398 |
+
"eval_samples_per_second": 11.572,
|
399 |
+
"eval_wer": 0.23103653950883016,
|
400 |
+
"step": 8400
|
401 |
+
},
|
402 |
+
{
|
403 |
+
"epoch": 3.66,
|
404 |
+
"learning_rate": 0.00036426900254309124,
|
405 |
+
"loss": 0.1594,
|
406 |
+
"step": 8700
|
407 |
+
},
|
408 |
+
{
|
409 |
+
"epoch": 3.66,
|
410 |
+
"eval_loss": 0.27261418104171753,
|
411 |
+
"eval_runtime": 459.5467,
|
412 |
+
"eval_samples_per_second": 11.344,
|
413 |
+
"eval_wer": 0.23287777806592153,
|
414 |
+
"step": 8700
|
415 |
+
},
|
416 |
+
{
|
417 |
+
"epoch": 3.79,
|
418 |
+
"learning_rate": 0.0003625227465385702,
|
419 |
+
"loss": 0.1646,
|
420 |
+
"step": 9000
|
421 |
+
},
|
422 |
+
{
|
423 |
+
"epoch": 3.79,
|
424 |
+
"eval_loss": 0.2741488218307495,
|
425 |
+
"eval_runtime": 421.349,
|
426 |
+
"eval_samples_per_second": 12.372,
|
427 |
+
"eval_wer": 0.2319182593812401,
|
428 |
+
"step": 9000
|
429 |
+
},
|
430 |
+
{
|
431 |
+
"epoch": 3.91,
|
432 |
+
"learning_rate": 0.00036077649053404913,
|
433 |
+
"loss": 0.1691,
|
434 |
+
"step": 9300
|
435 |
+
},
|
436 |
+
{
|
437 |
+
"epoch": 3.91,
|
438 |
+
"eval_loss": 0.3101365268230438,
|
439 |
+
"eval_runtime": 444.0677,
|
440 |
+
"eval_samples_per_second": 11.739,
|
441 |
+
"eval_wer": 0.2433546847851456,
|
442 |
+
"step": 9300
|
443 |
+
},
|
444 |
+
{
|
445 |
+
"epoch": 4.04,
|
446 |
+
"learning_rate": 0.0003590302345295281,
|
447 |
+
"loss": 0.1729,
|
448 |
+
"step": 9600
|
449 |
+
},
|
450 |
+
{
|
451 |
+
"epoch": 4.04,
|
452 |
+
"eval_loss": 0.2734999358654022,
|
453 |
+
"eval_runtime": 449.6252,
|
454 |
+
"eval_samples_per_second": 11.594,
|
455 |
+
"eval_wer": 0.21807007079691917,
|
456 |
+
"step": 9600
|
457 |
+
},
|
458 |
+
{
|
459 |
+
"epoch": 4.17,
|
460 |
+
"learning_rate": 0.000357283978525007,
|
461 |
+
"loss": 0.1416,
|
462 |
+
"step": 9900
|
463 |
+
},
|
464 |
+
{
|
465 |
+
"epoch": 4.17,
|
466 |
+
"eval_loss": 0.27035850286483765,
|
467 |
+
"eval_runtime": 421.6286,
|
468 |
+
"eval_samples_per_second": 12.364,
|
469 |
+
"eval_wer": 0.21532117942999404,
|
470 |
+
"step": 9900
|
471 |
+
},
|
472 |
+
{
|
473 |
+
"epoch": 4.29,
|
474 |
+
"learning_rate": 0.000355537722520486,
|
475 |
+
"loss": 0.1334,
|
476 |
+
"step": 10200
|
477 |
+
},
|
478 |
+
{
|
479 |
+
"epoch": 4.29,
|
480 |
+
"eval_loss": 0.2867918908596039,
|
481 |
+
"eval_runtime": 447.7813,
|
482 |
+
"eval_samples_per_second": 11.642,
|
483 |
+
"eval_wer": 0.2140763984336506,
|
484 |
+
"step": 10200
|
485 |
+
},
|
486 |
+
{
|
487 |
+
"epoch": 4.42,
|
488 |
+
"learning_rate": 0.00035379146651596497,
|
489 |
+
"loss": 0.1382,
|
490 |
+
"step": 10500
|
491 |
+
},
|
492 |
+
{
|
493 |
+
"epoch": 4.42,
|
494 |
+
"eval_loss": 0.29414185881614685,
|
495 |
+
"eval_runtime": 447.6999,
|
496 |
+
"eval_samples_per_second": 11.644,
|
497 |
+
"eval_wer": 0.22981769144991054,
|
498 |
+
"step": 10500
|
499 |
+
},
|
500 |
+
{
|
501 |
+
"epoch": 4.55,
|
502 |
+
"learning_rate": 0.0003520452105114439,
|
503 |
+
"loss": 0.1563,
|
504 |
+
"step": 10800
|
505 |
+
},
|
506 |
+
{
|
507 |
+
"epoch": 4.55,
|
508 |
+
"eval_loss": 0.2874026298522949,
|
509 |
+
"eval_runtime": 448.1528,
|
510 |
+
"eval_samples_per_second": 11.632,
|
511 |
+
"eval_wer": 0.2264982754596613,
|
512 |
+
"step": 10800
|
513 |
+
},
|
514 |
+
{
|
515 |
+
"epoch": 4.67,
|
516 |
+
"learning_rate": 0.00035029895450692285,
|
517 |
+
"loss": 0.1522,
|
518 |
+
"step": 11100
|
519 |
+
},
|
520 |
+
{
|
521 |
+
"epoch": 4.67,
|
522 |
+
"eval_loss": 0.27748048305511475,
|
523 |
+
"eval_runtime": 422.3545,
|
524 |
+
"eval_samples_per_second": 12.343,
|
525 |
+
"eval_wer": 0.21596950286558958,
|
526 |
+
"step": 11100
|
527 |
+
},
|
528 |
+
{
|
529 |
+
"epoch": 4.8,
|
530 |
+
"learning_rate": 0.00034855269850240177,
|
531 |
+
"loss": 0.152,
|
532 |
+
"step": 11400
|
533 |
+
},
|
534 |
+
{
|
535 |
+
"epoch": 4.8,
|
536 |
+
"eval_loss": 0.2689838707447052,
|
537 |
+
"eval_runtime": 450.681,
|
538 |
+
"eval_samples_per_second": 11.567,
|
539 |
+
"eval_wer": 0.221156090350354,
|
540 |
+
"step": 11400
|
541 |
+
},
|
542 |
+
{
|
543 |
+
"epoch": 4.92,
|
544 |
+
"learning_rate": 0.00034680644249788074,
|
545 |
+
"loss": 0.1537,
|
546 |
+
"step": 11700
|
547 |
+
},
|
548 |
+
{
|
549 |
+
"epoch": 4.92,
|
550 |
+
"eval_loss": 0.26583364605903625,
|
551 |
+
"eval_runtime": 439.0624,
|
552 |
+
"eval_samples_per_second": 11.873,
|
553 |
+
"eval_wer": 0.2123388916262545,
|
554 |
+
"step": 11700
|
555 |
+
},
|
556 |
+
{
|
557 |
+
"epoch": 5.05,
|
558 |
+
"learning_rate": 0.00034506018649335966,
|
559 |
+
"loss": 0.1394,
|
560 |
+
"step": 12000
|
561 |
+
},
|
562 |
+
{
|
563 |
+
"epoch": 5.05,
|
564 |
+
"eval_loss": 0.28286588191986084,
|
565 |
+
"eval_runtime": 420.652,
|
566 |
+
"eval_samples_per_second": 12.393,
|
567 |
+
"eval_wer": 0.20710043826664246,
|
568 |
+
"step": 12000
|
569 |
+
},
|
570 |
+
{
|
571 |
+
"epoch": 5.18,
|
572 |
+
"learning_rate": 0.0003433139304888387,
|
573 |
+
"loss": 0.1318,
|
574 |
+
"step": 12300
|
575 |
+
},
|
576 |
+
{
|
577 |
+
"epoch": 5.18,
|
578 |
+
"eval_loss": 0.2701680362224579,
|
579 |
+
"eval_runtime": 452.9348,
|
580 |
+
"eval_samples_per_second": 11.509,
|
581 |
+
"eval_wer": 0.20406628458805529,
|
582 |
+
"step": 12300
|
583 |
+
},
|
584 |
+
{
|
585 |
+
"epoch": 5.3,
|
586 |
+
"learning_rate": 0.0003415676744843176,
|
587 |
+
"loss": 0.1211,
|
588 |
+
"step": 12600
|
589 |
+
},
|
590 |
+
{
|
591 |
+
"epoch": 5.3,
|
592 |
+
"eval_loss": 0.27643057703971863,
|
593 |
+
"eval_runtime": 455.5802,
|
594 |
+
"eval_samples_per_second": 11.443,
|
595 |
+
"eval_wer": 0.19257799330930214,
|
596 |
+
"step": 12600
|
597 |
+
},
|
598 |
+
{
|
599 |
+
"epoch": 5.43,
|
600 |
+
"learning_rate": 0.0003398214184797965,
|
601 |
+
"loss": 0.1278,
|
602 |
+
"step": 12900
|
603 |
+
},
|
604 |
+
{
|
605 |
+
"epoch": 5.43,
|
606 |
+
"eval_loss": 0.2747470438480377,
|
607 |
+
"eval_runtime": 420.85,
|
608 |
+
"eval_samples_per_second": 12.387,
|
609 |
+
"eval_wer": 0.20979746375871994,
|
610 |
+
"step": 12900
|
611 |
+
},
|
612 |
+
{
|
613 |
+
"epoch": 5.56,
|
614 |
+
"learning_rate": 0.0003380751624752755,
|
615 |
+
"loss": 0.1334,
|
616 |
+
"step": 13200
|
617 |
+
},
|
618 |
+
{
|
619 |
+
"epoch": 5.56,
|
620 |
+
"eval_loss": 0.2526901960372925,
|
621 |
+
"eval_runtime": 447.6704,
|
622 |
+
"eval_samples_per_second": 11.645,
|
623 |
+
"eval_wer": 0.19836103835481445,
|
624 |
+
"step": 13200
|
625 |
+
},
|
626 |
+
{
|
627 |
+
"epoch": 5.68,
|
628 |
+
"learning_rate": 0.0003363289064707544,
|
629 |
+
"loss": 0.1262,
|
630 |
+
"step": 13500
|
631 |
+
},
|
632 |
+
{
|
633 |
+
"epoch": 5.68,
|
634 |
+
"eval_loss": 0.24875697493553162,
|
635 |
+
"eval_runtime": 453.8545,
|
636 |
+
"eval_samples_per_second": 11.486,
|
637 |
+
"eval_wer": 0.2052591997095511,
|
638 |
+
"step": 13500
|
639 |
+
},
|
640 |
+
{
|
641 |
+
"epoch": 5.81,
|
642 |
+
"learning_rate": 0.0003345826504662334,
|
643 |
+
"loss": 0.1366,
|
644 |
+
"step": 13800
|
645 |
+
},
|
646 |
+
{
|
647 |
+
"epoch": 5.81,
|
648 |
+
"eval_loss": 0.2449602484703064,
|
649 |
+
"eval_runtime": 450.7796,
|
650 |
+
"eval_samples_per_second": 11.564,
|
651 |
+
"eval_wer": 0.19501568942714143,
|
652 |
+
"step": 13800
|
653 |
+
},
|
654 |
+
{
|
655 |
+
"epoch": 5.93,
|
656 |
+
"learning_rate": 0.00033283639446171235,
|
657 |
+
"loss": 0.1348,
|
658 |
+
"step": 14100
|
659 |
+
},
|
660 |
+
{
|
661 |
+
"epoch": 5.93,
|
662 |
+
"eval_loss": 0.28218716382980347,
|
663 |
+
"eval_runtime": 422.2899,
|
664 |
+
"eval_samples_per_second": 12.345,
|
665 |
+
"eval_wer": 0.22273799953320714,
|
666 |
+
"step": 14100
|
667 |
+
},
|
668 |
+
{
|
669 |
+
"epoch": 6.06,
|
670 |
+
"learning_rate": 0.00033109013845719127,
|
671 |
+
"loss": 0.1325,
|
672 |
+
"step": 14400
|
673 |
+
},
|
674 |
+
{
|
675 |
+
"epoch": 6.06,
|
676 |
+
"eval_loss": 0.2777564525604248,
|
677 |
+
"eval_runtime": 447.1411,
|
678 |
+
"eval_samples_per_second": 11.659,
|
679 |
+
"eval_wer": 0.20780062757708564,
|
680 |
+
"step": 14400
|
681 |
+
},
|
682 |
+
{
|
683 |
+
"epoch": 6.19,
|
684 |
+
"learning_rate": 0.00032934388245267024,
|
685 |
+
"loss": 0.1261,
|
686 |
+
"step": 14700
|
687 |
+
},
|
688 |
+
{
|
689 |
+
"epoch": 6.19,
|
690 |
+
"eval_loss": 0.28267186880111694,
|
691 |
+
"eval_runtime": 452.8704,
|
692 |
+
"eval_samples_per_second": 11.511,
|
693 |
+
"eval_wer": 0.2153730453048417,
|
694 |
+
"step": 14700
|
695 |
+
},
|
696 |
+
{
|
697 |
+
"epoch": 6.31,
|
698 |
+
"learning_rate": 0.00032759762644814916,
|
699 |
+
"loss": 0.1268,
|
700 |
+
"step": 15000
|
701 |
+
},
|
702 |
+
{
|
703 |
+
"epoch": 6.31,
|
704 |
+
"eval_loss": 0.2655777335166931,
|
705 |
+
"eval_runtime": 421.8213,
|
706 |
+
"eval_samples_per_second": 12.358,
|
707 |
+
"eval_wer": 0.20074686859780608,
|
708 |
+
"step": 15000
|
709 |
+
},
|
710 |
+
{
|
711 |
+
"epoch": 6.44,
|
712 |
+
"learning_rate": 0.00032585137044362813,
|
713 |
+
"loss": 0.1146,
|
714 |
+
"step": 15300
|
715 |
+
},
|
716 |
+
{
|
717 |
+
"epoch": 6.44,
|
718 |
+
"eval_loss": 0.26334822177886963,
|
719 |
+
"eval_runtime": 453.2863,
|
720 |
+
"eval_samples_per_second": 11.5,
|
721 |
+
"eval_wer": 0.19060709006509166,
|
722 |
+
"step": 15300
|
723 |
+
},
|
724 |
+
{
|
725 |
+
"epoch": 6.57,
|
726 |
+
"learning_rate": 0.00032410511443910705,
|
727 |
+
"loss": 0.1196,
|
728 |
+
"step": 15600
|
729 |
+
},
|
730 |
+
{
|
731 |
+
"epoch": 6.57,
|
732 |
+
"eval_loss": 0.2629387676715851,
|
733 |
+
"eval_runtime": 453.8742,
|
734 |
+
"eval_samples_per_second": 11.486,
|
735 |
+
"eval_wer": 0.1959752081118228,
|
736 |
+
"step": 15600
|
737 |
+
},
|
738 |
+
{
|
739 |
+
"epoch": 6.69,
|
740 |
+
"learning_rate": 0.0003223588584345861,
|
741 |
+
"loss": 0.1162,
|
742 |
+
"step": 15900
|
743 |
+
},
|
744 |
+
{
|
745 |
+
"epoch": 6.69,
|
746 |
+
"eval_loss": 0.2686946392059326,
|
747 |
+
"eval_runtime": 423.221,
|
748 |
+
"eval_samples_per_second": 12.317,
|
749 |
+
"eval_wer": 0.19566401286273696,
|
750 |
+
"step": 15900
|
751 |
+
},
|
752 |
+
{
|
753 |
+
"epoch": 6.82,
|
754 |
+
"learning_rate": 0.000320612602430065,
|
755 |
+
"loss": 0.1178,
|
756 |
+
"step": 16200
|
757 |
+
},
|
758 |
+
{
|
759 |
+
"epoch": 6.82,
|
760 |
+
"eval_loss": 0.26208436489105225,
|
761 |
+
"eval_runtime": 455.8818,
|
762 |
+
"eval_samples_per_second": 11.435,
|
763 |
+
"eval_wer": 0.19716812323331864,
|
764 |
+
"step": 16200
|
765 |
+
},
|
766 |
+
{
|
767 |
+
"epoch": 6.94,
|
768 |
+
"learning_rate": 0.0003188663464255439,
|
769 |
+
"loss": 0.1172,
|
770 |
+
"step": 16500
|
771 |
+
},
|
772 |
+
{
|
773 |
+
"epoch": 6.94,
|
774 |
+
"eval_loss": 0.2683062255382538,
|
775 |
+
"eval_runtime": 453.6948,
|
776 |
+
"eval_samples_per_second": 11.49,
|
777 |
+
"eval_wer": 0.18845465625891444,
|
778 |
+
"step": 16500
|
779 |
+
},
|
780 |
+
{
|
781 |
+
"epoch": 7.07,
|
782 |
+
"learning_rate": 0.0003171200904210229,
|
783 |
+
"loss": 0.1137,
|
784 |
+
"step": 16800
|
785 |
+
},
|
786 |
+
{
|
787 |
+
"epoch": 7.07,
|
788 |
+
"eval_loss": 0.246523916721344,
|
789 |
+
"eval_runtime": 446.7642,
|
790 |
+
"eval_samples_per_second": 11.668,
|
791 |
+
"eval_wer": 0.18466844739503643,
|
792 |
+
"step": 16800
|
793 |
+
},
|
794 |
+
{
|
795 |
+
"epoch": 7.2,
|
796 |
+
"learning_rate": 0.0003153738344165018,
|
797 |
+
"loss": 0.1015,
|
798 |
+
"step": 17100
|
799 |
+
},
|
800 |
+
{
|
801 |
+
"epoch": 7.2,
|
802 |
+
"eval_loss": 0.261563241481781,
|
803 |
+
"eval_runtime": 423.2985,
|
804 |
+
"eval_samples_per_second": 12.315,
|
805 |
+
"eval_wer": 0.18783226576074272,
|
806 |
+
"step": 17100
|
807 |
+
},
|
808 |
+
{
|
809 |
+
"epoch": 7.32,
|
810 |
+
"learning_rate": 0.00031362757841198077,
|
811 |
+
"loss": 0.1048,
|
812 |
+
"step": 17400
|
813 |
+
},
|
814 |
+
{
|
815 |
+
"epoch": 7.32,
|
816 |
+
"eval_loss": 0.26347172260284424,
|
817 |
+
"eval_runtime": 455.9994,
|
818 |
+
"eval_samples_per_second": 11.432,
|
819 |
+
"eval_wer": 0.18096003734342989,
|
820 |
+
"step": 17400
|
821 |
+
},
|
822 |
+
{
|
823 |
+
"epoch": 7.45,
|
824 |
+
"learning_rate": 0.0003118813224074597,
|
825 |
+
"loss": 0.1085,
|
826 |
+
"step": 17700
|
827 |
+
},
|
828 |
+
{
|
829 |
+
"epoch": 7.45,
|
830 |
+
"eval_loss": 0.25925150513648987,
|
831 |
+
"eval_runtime": 451.158,
|
832 |
+
"eval_samples_per_second": 11.555,
|
833 |
+
"eval_wer": 0.19657166567257073,
|
834 |
+
"step": 17700
|
835 |
+
},
|
836 |
+
{
|
837 |
+
"epoch": 7.58,
|
838 |
+
"learning_rate": 0.0003101350664029387,
|
839 |
+
"loss": 0.1165,
|
840 |
+
"step": 18000
|
841 |
+
},
|
842 |
+
{
|
843 |
+
"epoch": 7.58,
|
844 |
+
"eval_loss": 0.25639158487319946,
|
845 |
+
"eval_runtime": 422.5184,
|
846 |
+
"eval_samples_per_second": 12.338,
|
847 |
+
"eval_wer": 0.1952490858639558,
|
848 |
+
"step": 18000
|
849 |
+
},
|
850 |
+
{
|
851 |
+
"epoch": 7.7,
|
852 |
+
"learning_rate": 0.00030838881039841763,
|
853 |
+
"loss": 0.1132,
|
854 |
+
"step": 18300
|
855 |
+
},
|
856 |
+
{
|
857 |
+
"epoch": 7.7,
|
858 |
+
"eval_loss": 0.2572135627269745,
|
859 |
+
"eval_runtime": 450.0834,
|
860 |
+
"eval_samples_per_second": 11.582,
|
861 |
+
"eval_wer": 0.19268172505899744,
|
862 |
+
"step": 18300
|
863 |
+
},
|
864 |
+
{
|
865 |
+
"epoch": 7.83,
|
866 |
+
"learning_rate": 0.00030664255439389655,
|
867 |
+
"loss": 0.1105,
|
868 |
+
"step": 18600
|
869 |
+
},
|
870 |
+
{
|
871 |
+
"epoch": 7.83,
|
872 |
+
"eval_loss": 0.26829710602760315,
|
873 |
+
"eval_runtime": 447.2579,
|
874 |
+
"eval_samples_per_second": 11.655,
|
875 |
+
"eval_wer": 0.18083037265631077,
|
876 |
+
"step": 18600
|
877 |
+
},
|
878 |
+
{
|
879 |
+
"epoch": 7.95,
|
880 |
+
"learning_rate": 0.0003048962983893755,
|
881 |
+
"loss": 0.1114,
|
882 |
+
"step": 18900
|
883 |
+
},
|
884 |
+
{
|
885 |
+
"epoch": 7.95,
|
886 |
+
"eval_loss": 0.2628280818462372,
|
887 |
+
"eval_runtime": 420.5057,
|
888 |
+
"eval_samples_per_second": 12.397,
|
889 |
+
"eval_wer": 0.18710614351287572,
|
890 |
+
"step": 18900
|
891 |
+
},
|
892 |
+
{
|
893 |
+
"epoch": 8.08,
|
894 |
+
"learning_rate": 0.00030315004238485444,
|
895 |
+
"loss": 0.1008,
|
896 |
+
"step": 19200
|
897 |
+
},
|
898 |
+
{
|
899 |
+
"epoch": 8.08,
|
900 |
+
"eval_loss": 0.2631722390651703,
|
901 |
+
"eval_runtime": 454.7744,
|
902 |
+
"eval_samples_per_second": 11.463,
|
903 |
+
"eval_wer": 0.1872358081999948,
|
904 |
+
"step": 19200
|
905 |
+
},
|
906 |
+
{
|
907 |
+
"epoch": 8.21,
|
908 |
+
"learning_rate": 0.0003014037863803334,
|
909 |
+
"loss": 0.0916,
|
910 |
+
"step": 19500
|
911 |
+
},
|
912 |
+
{
|
913 |
+
"epoch": 8.21,
|
914 |
+
"eval_loss": 0.2808758616447449,
|
915 |
+
"eval_runtime": 447.7804,
|
916 |
+
"eval_samples_per_second": 11.642,
|
917 |
+
"eval_wer": 0.1823863489017401,
|
918 |
+
"step": 19500
|
919 |
+
},
|
920 |
+
{
|
921 |
+
"epoch": 8.33,
|
922 |
+
"learning_rate": 0.0002996575303758124,
|
923 |
+
"loss": 0.0991,
|
924 |
+
"step": 19800
|
925 |
+
},
|
926 |
+
{
|
927 |
+
"epoch": 8.33,
|
928 |
+
"eval_loss": 0.26910626888275146,
|
929 |
+
"eval_runtime": 418.5271,
|
930 |
+
"eval_samples_per_second": 12.456,
|
931 |
+
"eval_wer": 0.18023391509556289,
|
932 |
+
"step": 19800
|
933 |
+
},
|
934 |
+
{
|
935 |
+
"epoch": 8.46,
|
936 |
+
"learning_rate": 0.0002979112743712913,
|
937 |
+
"loss": 0.0969,
|
938 |
+
"step": 20100
|
939 |
+
},
|
940 |
+
{
|
941 |
+
"epoch": 8.46,
|
942 |
+
"eval_loss": 0.24965764582157135,
|
943 |
+
"eval_runtime": 443.9577,
|
944 |
+
"eval_samples_per_second": 11.742,
|
945 |
+
"eval_wer": 0.1918259381240113,
|
946 |
+
"step": 20100
|
947 |
+
},
|
948 |
+
{
|
949 |
+
"epoch": 8.59,
|
950 |
+
"learning_rate": 0.00029616501836677027,
|
951 |
+
"loss": 0.1008,
|
952 |
+
"step": 20400
|
953 |
+
},
|
954 |
+
{
|
955 |
+
"epoch": 8.59,
|
956 |
+
"eval_loss": 0.2633902132511139,
|
957 |
+
"eval_runtime": 446.6554,
|
958 |
+
"eval_samples_per_second": 11.671,
|
959 |
+
"eval_wer": 0.17963745753481497,
|
960 |
+
"step": 20400
|
961 |
+
},
|
962 |
+
{
|
963 |
+
"epoch": 8.71,
|
964 |
+
"learning_rate": 0.0002944187623622492,
|
965 |
+
"loss": 0.1003,
|
966 |
+
"step": 20700
|
967 |
+
},
|
968 |
+
{
|
969 |
+
"epoch": 8.71,
|
970 |
+
"eval_loss": 0.26895150542259216,
|
971 |
+
"eval_runtime": 447.0422,
|
972 |
+
"eval_samples_per_second": 11.661,
|
973 |
+
"eval_wer": 0.17255776561811156,
|
974 |
+
"step": 20700
|
975 |
+
},
|
976 |
+
{
|
977 |
+
"epoch": 8.84,
|
978 |
+
"learning_rate": 0.00029267250635772816,
|
979 |
+
"loss": 0.1059,
|
980 |
+
"step": 21000
|
981 |
+
},
|
982 |
+
{
|
983 |
+
"epoch": 8.84,
|
984 |
+
"eval_loss": 0.25897544622421265,
|
985 |
+
"eval_runtime": 418.2635,
|
986 |
+
"eval_samples_per_second": 12.463,
|
987 |
+
"eval_wer": 0.18676901532636603,
|
988 |
+
"step": 21000
|
989 |
+
},
|
990 |
+
{
|
991 |
+
"epoch": 8.96,
|
992 |
+
"learning_rate": 0.0002909262503532071,
|
993 |
+
"loss": 0.1011,
|
994 |
+
"step": 21300
|
995 |
+
},
|
996 |
+
{
|
997 |
+
"epoch": 8.96,
|
998 |
+
"eval_loss": 0.27191728353500366,
|
999 |
+
"eval_runtime": 439.9733,
|
1000 |
+
"eval_samples_per_second": 11.848,
|
1001 |
+
"eval_wer": 0.17569565104639404,
|
1002 |
+
"step": 21300
|
1003 |
+
},
|
1004 |
+
{
|
1005 |
+
"epoch": 9.09,
|
1006 |
+
"learning_rate": 0.0002891799943486861,
|
1007 |
+
"loss": 0.086,
|
1008 |
+
"step": 21600
|
1009 |
+
},
|
1010 |
+
{
|
1011 |
+
"epoch": 9.09,
|
1012 |
+
"eval_loss": 0.24117550253868103,
|
1013 |
+
"eval_runtime": 451.9133,
|
1014 |
+
"eval_samples_per_second": 11.535,
|
1015 |
+
"eval_wer": 0.16464821970384586,
|
1016 |
+
"step": 21600
|
1017 |
+
},
|
1018 |
+
{
|
1019 |
+
"epoch": 9.22,
|
1020 |
+
"learning_rate": 0.000287433738344165,
|
1021 |
+
"loss": 0.0896,
|
1022 |
+
"step": 21900
|
1023 |
+
},
|
1024 |
+
{
|
1025 |
+
"epoch": 9.22,
|
1026 |
+
"eval_loss": 0.2518727481365204,
|
1027 |
+
"eval_runtime": 420.864,
|
1028 |
+
"eval_samples_per_second": 12.386,
|
1029 |
+
"eval_wer": 0.1694198801898291,
|
1030 |
+
"step": 21900
|
1031 |
+
},
|
1032 |
+
{
|
1033 |
+
"epoch": 9.34,
|
1034 |
+
"learning_rate": 0.00028568748233964394,
|
1035 |
+
"loss": 0.0853,
|
1036 |
+
"step": 22200
|
1037 |
+
},
|
1038 |
+
{
|
1039 |
+
"epoch": 9.34,
|
1040 |
+
"eval_loss": 0.25565269589424133,
|
1041 |
+
"eval_runtime": 442.9887,
|
1042 |
+
"eval_samples_per_second": 11.768,
|
1043 |
+
"eval_wer": 0.17079432587329166,
|
1044 |
+
"step": 22200
|
1045 |
+
},
|
1046 |
+
{
|
1047 |
+
"epoch": 9.47,
|
1048 |
+
"learning_rate": 0.0002839412263351229,
|
1049 |
+
"loss": 0.0792,
|
1050 |
+
"step": 22500
|
1051 |
+
},
|
1052 |
+
{
|
1053 |
+
"epoch": 9.47,
|
1054 |
+
"eval_loss": 0.24455079436302185,
|
1055 |
+
"eval_runtime": 447.726,
|
1056 |
+
"eval_samples_per_second": 11.643,
|
1057 |
+
"eval_wer": 0.16685251938487072,
|
1058 |
+
"step": 22500
|
1059 |
+
},
|
1060 |
+
{
|
1061 |
+
"epoch": 9.6,
|
1062 |
+
"learning_rate": 0.0002821949703306018,
|
1063 |
+
"loss": 0.0864,
|
1064 |
+
"step": 22800
|
1065 |
+
},
|
1066 |
+
{
|
1067 |
+
"epoch": 9.6,
|
1068 |
+
"eval_loss": 0.2556290626525879,
|
1069 |
+
"eval_runtime": 421.6345,
|
1070 |
+
"eval_samples_per_second": 12.364,
|
1071 |
+
"eval_wer": 0.16667098882290396,
|
1072 |
+
"step": 22800
|
1073 |
+
},
|
1074 |
+
{
|
1075 |
+
"epoch": 9.72,
|
1076 |
+
"learning_rate": 0.0002804487143260808,
|
1077 |
+
"loss": 0.092,
|
1078 |
+
"step": 23100
|
1079 |
+
},
|
1080 |
+
{
|
1081 |
+
"epoch": 9.72,
|
1082 |
+
"eval_loss": 0.2547804117202759,
|
1083 |
+
"eval_runtime": 444.7489,
|
1084 |
+
"eval_samples_per_second": 11.721,
|
1085 |
+
"eval_wer": 0.17476206529913643,
|
1086 |
+
"step": 23100
|
1087 |
+
},
|
1088 |
+
{
|
1089 |
+
"epoch": 9.85,
|
1090 |
+
"learning_rate": 0.00027870245832155977,
|
1091 |
+
"loss": 0.0915,
|
1092 |
+
"step": 23400
|
1093 |
+
},
|
1094 |
+
{
|
1095 |
+
"epoch": 9.85,
|
1096 |
+
"eval_loss": 0.23804427683353424,
|
1097 |
+
"eval_runtime": 440.9531,
|
1098 |
+
"eval_samples_per_second": 11.822,
|
1099 |
+
"eval_wer": 0.16879748969165737,
|
1100 |
+
"step": 23400
|
1101 |
+
},
|
1102 |
+
{
|
1103 |
+
"epoch": 9.97,
|
1104 |
+
"learning_rate": 0.00027695620231703874,
|
1105 |
+
"loss": 0.0926,
|
1106 |
+
"step": 23700
|
1107 |
+
},
|
1108 |
+
{
|
1109 |
+
"epoch": 9.97,
|
1110 |
+
"eval_loss": 0.2537280321121216,
|
1111 |
+
"eval_runtime": 446.3024,
|
1112 |
+
"eval_samples_per_second": 11.68,
|
1113 |
+
"eval_wer": 0.1642332927050647,
|
1114 |
+
"step": 23700
|
1115 |
+
},
|
1116 |
+
{
|
1117 |
+
"epoch": 10.1,
|
1118 |
+
"learning_rate": 0.00027520994631251766,
|
1119 |
+
"loss": 0.0819,
|
1120 |
+
"step": 24000
|
1121 |
+
},
|
1122 |
+
{
|
1123 |
+
"epoch": 10.1,
|
1124 |
+
"eval_loss": 0.24399155378341675,
|
1125 |
+
"eval_runtime": 420.6608,
|
1126 |
+
"eval_samples_per_second": 12.392,
|
1127 |
+
"eval_wer": 0.16405176214309794,
|
1128 |
+
"step": 24000
|
1129 |
+
},
|
1130 |
+
{
|
1131 |
+
"epoch": 10.23,
|
1132 |
+
"learning_rate": 0.0002734636903079966,
|
1133 |
+
"loss": 0.08,
|
1134 |
+
"step": 24300
|
1135 |
+
},
|
1136 |
+
{
|
1137 |
+
"epoch": 10.23,
|
1138 |
+
"eval_loss": 0.24013200402259827,
|
1139 |
+
"eval_runtime": 448.905,
|
1140 |
+
"eval_samples_per_second": 11.613,
|
1141 |
+
"eval_wer": 0.15940976634423382,
|
1142 |
+
"step": 24300
|
1143 |
+
},
|
1144 |
+
{
|
1145 |
+
"epoch": 10.35,
|
1146 |
+
"learning_rate": 0.00027171743430347555,
|
1147 |
+
"loss": 0.0775,
|
1148 |
+
"step": 24600
|
1149 |
+
},
|
1150 |
+
{
|
1151 |
+
"epoch": 10.35,
|
1152 |
+
"eval_loss": 0.2440200299024582,
|
1153 |
+
"eval_runtime": 449.5925,
|
1154 |
+
"eval_samples_per_second": 11.595,
|
1155 |
+
"eval_wer": 0.1610435414019346,
|
1156 |
+
"step": 24600
|
1157 |
+
},
|
1158 |
+
{
|
1159 |
+
"epoch": 10.48,
|
1160 |
+
"learning_rate": 0.00026997117829895447,
|
1161 |
+
"loss": 0.0772,
|
1162 |
+
"step": 24900
|
1163 |
+
},
|
1164 |
+
{
|
1165 |
+
"epoch": 10.48,
|
1166 |
+
"eval_loss": 0.27414917945861816,
|
1167 |
+
"eval_runtime": 419.8275,
|
1168 |
+
"eval_samples_per_second": 12.417,
|
1169 |
+
"eval_wer": 0.166230128886699,
|
1170 |
+
"step": 24900
|
1171 |
+
},
|
1172 |
+
{
|
1173 |
+
"epoch": 10.61,
|
1174 |
+
"learning_rate": 0.00026822492229443344,
|
1175 |
+
"loss": 0.0808,
|
1176 |
+
"step": 25200
|
1177 |
+
},
|
1178 |
+
{
|
1179 |
+
"epoch": 10.61,
|
1180 |
+
"eval_loss": 0.2687024176120758,
|
1181 |
+
"eval_runtime": 443.5669,
|
1182 |
+
"eval_samples_per_second": 11.752,
|
1183 |
+
"eval_wer": 0.1630663105209927,
|
1184 |
+
"step": 25200
|
1185 |
+
},
|
1186 |
+
{
|
1187 |
+
"epoch": 10.73,
|
1188 |
+
"learning_rate": 0.0002664786662899124,
|
1189 |
+
"loss": 0.0801,
|
1190 |
+
"step": 25500
|
1191 |
+
},
|
1192 |
+
{
|
1193 |
+
"epoch": 10.73,
|
1194 |
+
"eval_loss": 0.25520530343055725,
|
1195 |
+
"eval_runtime": 450.4913,
|
1196 |
+
"eval_samples_per_second": 11.572,
|
1197 |
+
"eval_wer": 0.1645704208915744,
|
1198 |
+
"step": 25500
|
1199 |
+
},
|
1200 |
+
{
|
1201 |
+
"epoch": 10.86,
|
1202 |
+
"learning_rate": 0.0002647324102853913,
|
1203 |
+
"loss": 0.0852,
|
1204 |
+
"step": 25800
|
1205 |
+
},
|
1206 |
+
{
|
1207 |
+
"epoch": 10.86,
|
1208 |
+
"eval_loss": 0.24987785518169403,
|
1209 |
+
"eval_runtime": 423.7767,
|
1210 |
+
"eval_samples_per_second": 12.301,
|
1211 |
+
"eval_wer": 0.16527061020201758,
|
1212 |
+
"step": 25800
|
1213 |
+
},
|
1214 |
+
{
|
1215 |
+
"epoch": 10.98,
|
1216 |
+
"learning_rate": 0.0002629861542808703,
|
1217 |
+
"loss": 0.0822,
|
1218 |
+
"step": 26100
|
1219 |
+
},
|
1220 |
+
{
|
1221 |
+
"epoch": 10.98,
|
1222 |
+
"eval_loss": 0.24487553536891937,
|
1223 |
+
"eval_runtime": 453.0579,
|
1224 |
+
"eval_samples_per_second": 11.506,
|
1225 |
+
"eval_wer": 0.1553901610435414,
|
1226 |
+
"step": 26100
|
1227 |
+
},
|
1228 |
+
{
|
1229 |
+
"epoch": 11.11,
|
1230 |
+
"learning_rate": 0.0002612398982763492,
|
1231 |
+
"loss": 0.0693,
|
1232 |
+
"step": 26400
|
1233 |
+
},
|
1234 |
+
{
|
1235 |
+
"epoch": 11.11,
|
1236 |
+
"eval_loss": 0.2533997893333435,
|
1237 |
+
"eval_runtime": 453.0359,
|
1238 |
+
"eval_samples_per_second": 11.507,
|
1239 |
+
"eval_wer": 0.14846606675138094,
|
1240 |
+
"step": 26400
|
1241 |
+
},
|
1242 |
+
{
|
1243 |
+
"epoch": 11.24,
|
1244 |
+
"learning_rate": 0.0002594936422718282,
|
1245 |
+
"loss": 0.0694,
|
1246 |
+
"step": 26700
|
1247 |
+
},
|
1248 |
+
{
|
1249 |
+
"epoch": 11.24,
|
1250 |
+
"eval_loss": 0.2577211260795593,
|
1251 |
+
"eval_runtime": 448.0425,
|
1252 |
+
"eval_samples_per_second": 11.635,
|
1253 |
+
"eval_wer": 0.15048883587043904,
|
1254 |
+
"step": 26700
|
1255 |
+
},
|
1256 |
+
{
|
1257 |
+
"epoch": 11.36,
|
1258 |
+
"learning_rate": 0.0002577473862673071,
|
1259 |
+
"loss": 0.0718,
|
1260 |
+
"step": 27000
|
1261 |
+
},
|
1262 |
+
{
|
1263 |
+
"epoch": 11.36,
|
1264 |
+
"eval_loss": 0.24945984780788422,
|
1265 |
+
"eval_runtime": 422.4019,
|
1266 |
+
"eval_samples_per_second": 12.341,
|
1267 |
+
"eval_wer": 0.15331552604963564,
|
1268 |
+
"step": 27000
|
1269 |
+
},
|
1270 |
+
{
|
1271 |
+
"epoch": 11.49,
|
1272 |
+
"learning_rate": 0.00025600113026278613,
|
1273 |
+
"loss": 0.0712,
|
1274 |
+
"step": 27300
|
1275 |
+
},
|
1276 |
+
{
|
1277 |
+
"epoch": 11.49,
|
1278 |
+
"eval_loss": 0.25240784883499146,
|
1279 |
+
"eval_runtime": 445.1796,
|
1280 |
+
"eval_samples_per_second": 11.71,
|
1281 |
+
"eval_wer": 0.15487150229506497,
|
1282 |
+
"step": 27300
|
1283 |
+
},
|
1284 |
+
{
|
1285 |
+
"epoch": 11.62,
|
1286 |
+
"learning_rate": 0.00025425487425826505,
|
1287 |
+
"loss": 0.0724,
|
1288 |
+
"step": 27600
|
1289 |
+
},
|
1290 |
+
{
|
1291 |
+
"epoch": 11.62,
|
1292 |
+
"eval_loss": 0.2555846869945526,
|
1293 |
+
"eval_runtime": 442.017,
|
1294 |
+
"eval_samples_per_second": 11.794,
|
1295 |
+
"eval_wer": 0.15507896579445554,
|
1296 |
+
"step": 27600
|
1297 |
+
},
|
1298 |
+
{
|
1299 |
+
"epoch": 11.74,
|
1300 |
+
"learning_rate": 0.00025250861825374397,
|
1301 |
+
"loss": 0.0733,
|
1302 |
+
"step": 27900
|
1303 |
+
},
|
1304 |
+
{
|
1305 |
+
"epoch": 11.74,
|
1306 |
+
"eval_loss": 0.2552262246608734,
|
1307 |
+
"eval_runtime": 422.3705,
|
1308 |
+
"eval_samples_per_second": 12.342,
|
1309 |
+
"eval_wer": 0.15331552604963564,
|
1310 |
+
"step": 27900
|
1311 |
+
},
|
1312 |
+
{
|
1313 |
+
"epoch": 11.87,
|
1314 |
+
"learning_rate": 0.00025076236224922294,
|
1315 |
+
"loss": 0.0739,
|
1316 |
+
"step": 28200
|
1317 |
+
},
|
1318 |
+
{
|
1319 |
+
"epoch": 11.87,
|
1320 |
+
"eval_loss": 0.25745663046836853,
|
1321 |
+
"eval_runtime": 446.6268,
|
1322 |
+
"eval_samples_per_second": 11.672,
|
1323 |
+
"eval_wer": 0.15181141567905396,
|
1324 |
+
"step": 28200
|
1325 |
+
},
|
1326 |
+
{
|
1327 |
+
"epoch": 11.99,
|
1328 |
+
"learning_rate": 0.00024901610624470185,
|
1329 |
+
"loss": 0.0756,
|
1330 |
+
"step": 28500
|
1331 |
+
},
|
1332 |
+
{
|
1333 |
+
"epoch": 11.99,
|
1334 |
+
"eval_loss": 0.25365495681762695,
|
1335 |
+
"eval_runtime": 445.0479,
|
1336 |
+
"eval_samples_per_second": 11.713,
|
1337 |
+
"eval_wer": 0.15199294624102072,
|
1338 |
+
"step": 28500
|
1339 |
+
},
|
1340 |
+
{
|
1341 |
+
"epoch": 12.12,
|
1342 |
+
"learning_rate": 0.0002472698502401808,
|
1343 |
+
"loss": 0.0623,
|
1344 |
+
"step": 28800
|
1345 |
+
},
|
1346 |
+
{
|
1347 |
+
"epoch": 12.12,
|
1348 |
+
"eval_loss": 0.25725752115249634,
|
1349 |
+
"eval_runtime": 422.5366,
|
1350 |
+
"eval_samples_per_second": 12.337,
|
1351 |
+
"eval_wer": 0.15450844117113147,
|
1352 |
+
"step": 28800
|
1353 |
+
},
|
1354 |
+
{
|
1355 |
+
"epoch": 12.25,
|
1356 |
+
"learning_rate": 0.0002455235942356598,
|
1357 |
+
"loss": 0.0642,
|
1358 |
+
"step": 29100
|
1359 |
+
},
|
1360 |
+
{
|
1361 |
+
"epoch": 12.25,
|
1362 |
+
"eval_loss": 0.2661702334880829,
|
1363 |
+
"eval_runtime": 452.9004,
|
1364 |
+
"eval_samples_per_second": 11.51,
|
1365 |
+
"eval_wer": 0.1533673919244833,
|
1366 |
+
"step": 29100
|
1367 |
+
},
|
1368 |
+
{
|
1369 |
+
"epoch": 12.37,
|
1370 |
+
"learning_rate": 0.00024377733823113874,
|
1371 |
+
"loss": 0.0656,
|
1372 |
+
"step": 29400
|
1373 |
+
},
|
1374 |
+
{
|
1375 |
+
"epoch": 12.37,
|
1376 |
+
"eval_loss": 0.27368617057800293,
|
1377 |
+
"eval_runtime": 444.5703,
|
1378 |
+
"eval_samples_per_second": 11.726,
|
1379 |
+
"eval_wer": 0.15121495811830607,
|
1380 |
+
"step": 29400
|
1381 |
+
},
|
1382 |
+
{
|
1383 |
+
"epoch": 12.5,
|
1384 |
+
"learning_rate": 0.0002420310822266177,
|
1385 |
+
"loss": 0.0713,
|
1386 |
+
"step": 29700
|
1387 |
+
},
|
1388 |
+
{
|
1389 |
+
"epoch": 12.5,
|
1390 |
+
"eval_loss": 0.27105453610420227,
|
1391 |
+
"eval_runtime": 422.4149,
|
1392 |
+
"eval_samples_per_second": 12.341,
|
1393 |
+
"eval_wer": 0.1492699878115194,
|
1394 |
+
"step": 29700
|
1395 |
+
},
|
1396 |
+
{
|
1397 |
+
"epoch": 12.63,
|
1398 |
+
"learning_rate": 0.00024028482622209663,
|
1399 |
+
"loss": 0.0667,
|
1400 |
+
"step": 30000
|
1401 |
+
},
|
1402 |
+
{
|
1403 |
+
"epoch": 12.63,
|
1404 |
+
"eval_loss": 0.24904416501522064,
|
1405 |
+
"eval_runtime": 450.1403,
|
1406 |
+
"eval_samples_per_second": 11.581,
|
1407 |
+
"eval_wer": 0.150203573558777,
|
1408 |
+
"step": 30000
|
1409 |
+
},
|
1410 |
+
{
|
1411 |
+
"epoch": 12.75,
|
1412 |
+
"learning_rate": 0.00023853857021757558,
|
1413 |
+
"loss": 0.0629,
|
1414 |
+
"step": 30300
|
1415 |
+
},
|
1416 |
+
{
|
1417 |
+
"epoch": 12.75,
|
1418 |
+
"eval_loss": 0.26200103759765625,
|
1419 |
+
"eval_runtime": 443.2234,
|
1420 |
+
"eval_samples_per_second": 11.762,
|
1421 |
+
"eval_wer": 0.1519670133035969,
|
1422 |
+
"step": 30300
|
1423 |
+
},
|
1424 |
+
{
|
1425 |
+
"epoch": 12.88,
|
1426 |
+
"learning_rate": 0.00023679231421305452,
|
1427 |
+
"loss": 0.0657,
|
1428 |
+
"step": 30600
|
1429 |
+
},
|
1430 |
+
{
|
1431 |
+
"epoch": 12.88,
|
1432 |
+
"eval_loss": 0.2610265016555786,
|
1433 |
+
"eval_runtime": 438.2352,
|
1434 |
+
"eval_samples_per_second": 11.895,
|
1435 |
+
"eval_wer": 0.1481030056274474,
|
1436 |
+
"step": 30600
|
1437 |
+
},
|
1438 |
+
{
|
1439 |
+
"epoch": 13.01,
|
1440 |
+
"learning_rate": 0.0002350460582085335,
|
1441 |
+
"loss": 0.0712,
|
1442 |
+
"step": 30900
|
1443 |
+
},
|
1444 |
+
{
|
1445 |
+
"epoch": 13.01,
|
1446 |
+
"eval_loss": 0.2637769281864166,
|
1447 |
+
"eval_runtime": 423.4731,
|
1448 |
+
"eval_samples_per_second": 12.31,
|
1449 |
+
"eval_wer": 0.1468841575685278,
|
1450 |
+
"step": 30900
|
1451 |
+
},
|
1452 |
+
{
|
1453 |
+
"epoch": 13.13,
|
1454 |
+
"learning_rate": 0.00023329980220401244,
|
1455 |
+
"loss": 0.0624,
|
1456 |
+
"step": 31200
|
1457 |
+
},
|
1458 |
+
{
|
1459 |
+
"epoch": 13.13,
|
1460 |
+
"eval_loss": 0.25238868594169617,
|
1461 |
+
"eval_runtime": 449.8496,
|
1462 |
+
"eval_samples_per_second": 11.588,
|
1463 |
+
"eval_wer": 0.14999611005938643,
|
1464 |
+
"step": 31200
|
1465 |
+
},
|
1466 |
+
{
|
1467 |
+
"epoch": 13.26,
|
1468 |
+
"learning_rate": 0.00023155354619949138,
|
1469 |
+
"loss": 0.0592,
|
1470 |
+
"step": 31500
|
1471 |
+
},
|
1472 |
+
{
|
1473 |
+
"epoch": 13.26,
|
1474 |
+
"eval_loss": 0.25298765301704407,
|
1475 |
+
"eval_runtime": 451.2139,
|
1476 |
+
"eval_samples_per_second": 11.553,
|
1477 |
+
"eval_wer": 0.1419309665205778,
|
1478 |
+
"step": 31500
|
1479 |
+
},
|
1480 |
+
{
|
1481 |
+
"epoch": 13.38,
|
1482 |
+
"learning_rate": 0.00022980729019497033,
|
1483 |
+
"loss": 0.0634,
|
1484 |
+
"step": 31800
|
1485 |
+
},
|
1486 |
+
{
|
1487 |
+
"epoch": 13.38,
|
1488 |
+
"eval_loss": 0.2578509449958801,
|
1489 |
+
"eval_runtime": 421.4953,
|
1490 |
+
"eval_samples_per_second": 12.368,
|
1491 |
+
"eval_wer": 0.15080003111952492,
|
1492 |
+
"step": 31800
|
1493 |
+
},
|
1494 |
+
{
|
1495 |
+
"epoch": 13.51,
|
1496 |
+
"learning_rate": 0.00022806103419044927,
|
1497 |
+
"loss": 0.0575,
|
1498 |
+
"step": 32100
|
1499 |
+
},
|
1500 |
+
{
|
1501 |
+
"epoch": 13.51,
|
1502 |
+
"eval_loss": 0.24078154563903809,
|
1503 |
+
"eval_runtime": 445.6242,
|
1504 |
+
"eval_samples_per_second": 11.698,
|
1505 |
+
"eval_wer": 0.1440574673893312,
|
1506 |
+
"step": 32100
|
1507 |
+
},
|
1508 |
+
{
|
1509 |
+
"epoch": 13.64,
|
1510 |
+
"learning_rate": 0.00022631477818592821,
|
1511 |
+
"loss": 0.0571,
|
1512 |
+
"step": 32400
|
1513 |
+
},
|
1514 |
+
{
|
1515 |
+
"epoch": 13.64,
|
1516 |
+
"eval_loss": 0.24118457734584808,
|
1517 |
+
"eval_runtime": 441.1228,
|
1518 |
+
"eval_samples_per_second": 11.818,
|
1519 |
+
"eval_wer": 0.13801509296958067,
|
1520 |
+
"step": 32400
|
1521 |
+
},
|
1522 |
+
{
|
1523 |
+
"epoch": 13.76,
|
1524 |
+
"learning_rate": 0.00022456852218140716,
|
1525 |
+
"loss": 0.0583,
|
1526 |
+
"step": 32700
|
1527 |
+
},
|
1528 |
+
{
|
1529 |
+
"epoch": 13.76,
|
1530 |
+
"eval_loss": 0.2555965483188629,
|
1531 |
+
"eval_runtime": 420.9676,
|
1532 |
+
"eval_samples_per_second": 12.383,
|
1533 |
+
"eval_wer": 0.1451207178237079,
|
1534 |
+
"step": 32700
|
1535 |
+
},
|
1536 |
+
{
|
1537 |
+
"epoch": 13.89,
|
1538 |
+
"learning_rate": 0.00022282226617688613,
|
1539 |
+
"loss": 0.0564,
|
1540 |
+
"step": 33000
|
1541 |
+
},
|
1542 |
+
{
|
1543 |
+
"epoch": 13.89,
|
1544 |
+
"eval_loss": 0.25013622641563416,
|
1545 |
+
"eval_runtime": 445.1032,
|
1546 |
+
"eval_samples_per_second": 11.712,
|
1547 |
+
"eval_wer": 0.1420606312076969,
|
1548 |
+
"step": 33000
|
1549 |
+
},
|
1550 |
+
{
|
1551 |
+
"epoch": 14.02,
|
1552 |
+
"learning_rate": 0.00022107601017236508,
|
1553 |
+
"loss": 0.0628,
|
1554 |
+
"step": 33300
|
1555 |
+
},
|
1556 |
+
{
|
1557 |
+
"epoch": 14.02,
|
1558 |
+
"eval_loss": 0.2687903344631195,
|
1559 |
+
"eval_runtime": 448.3298,
|
1560 |
+
"eval_samples_per_second": 11.628,
|
1561 |
+
"eval_wer": 0.14968491481030055,
|
1562 |
+
"step": 33300
|
1563 |
+
},
|
1564 |
+
{
|
1565 |
+
"epoch": 14.14,
|
1566 |
+
"learning_rate": 0.00021932975416784402,
|
1567 |
+
"loss": 0.0565,
|
1568 |
+
"step": 33600
|
1569 |
+
},
|
1570 |
+
{
|
1571 |
+
"epoch": 14.14,
|
1572 |
+
"eval_loss": 0.2639634609222412,
|
1573 |
+
"eval_runtime": 444.628,
|
1574 |
+
"eval_samples_per_second": 11.724,
|
1575 |
+
"eval_wer": 0.14154197245922046,
|
1576 |
+
"step": 33600
|
1577 |
+
},
|
1578 |
+
{
|
1579 |
+
"epoch": 14.27,
|
1580 |
+
"learning_rate": 0.00021758349816332296,
|
1581 |
+
"loss": 0.0539,
|
1582 |
+
"step": 33900
|
1583 |
+
},
|
1584 |
+
{
|
1585 |
+
"epoch": 14.27,
|
1586 |
+
"eval_loss": 0.25850602984428406,
|
1587 |
+
"eval_runtime": 422.0625,
|
1588 |
+
"eval_samples_per_second": 12.351,
|
1589 |
+
"eval_wer": 0.14177536889603484,
|
1590 |
+
"step": 33900
|
1591 |
+
},
|
1592 |
+
{
|
1593 |
+
"epoch": 14.39,
|
1594 |
+
"learning_rate": 0.0002158372421588019,
|
1595 |
+
"loss": 0.0515,
|
1596 |
+
"step": 34200
|
1597 |
+
},
|
1598 |
+
{
|
1599 |
+
"epoch": 14.39,
|
1600 |
+
"eval_loss": 0.2456846386194229,
|
1601 |
+
"eval_runtime": 439.8595,
|
1602 |
+
"eval_samples_per_second": 11.852,
|
1603 |
+
"eval_wer": 0.13741863540883276,
|
1604 |
+
"step": 34200
|
1605 |
+
},
|
1606 |
+
{
|
1607 |
+
"epoch": 14.52,
|
1608 |
+
"learning_rate": 0.00021409098615428085,
|
1609 |
+
"loss": 0.0535,
|
1610 |
+
"step": 34500
|
1611 |
+
},
|
1612 |
+
{
|
1613 |
+
"epoch": 14.52,
|
1614 |
+
"eval_loss": 0.2575737535953522,
|
1615 |
+
"eval_runtime": 450.7209,
|
1616 |
+
"eval_samples_per_second": 11.566,
|
1617 |
+
"eval_wer": 0.13684811078550868,
|
1618 |
+
"step": 34500
|
1619 |
+
},
|
1620 |
+
{
|
1621 |
+
"epoch": 14.65,
|
1622 |
+
"learning_rate": 0.00021234473014975982,
|
1623 |
+
"loss": 0.0528,
|
1624 |
+
"step": 34800
|
1625 |
+
},
|
1626 |
+
{
|
1627 |
+
"epoch": 14.65,
|
1628 |
+
"eval_loss": 0.24703530967235565,
|
1629 |
+
"eval_runtime": 422.3035,
|
1630 |
+
"eval_samples_per_second": 12.344,
|
1631 |
+
"eval_wer": 0.13889681284199062,
|
1632 |
+
"step": 34800
|
1633 |
+
},
|
1634 |
+
{
|
1635 |
+
"epoch": 14.77,
|
1636 |
+
"learning_rate": 0.00021059847414523877,
|
1637 |
+
"loss": 0.0556,
|
1638 |
+
"step": 35100
|
1639 |
+
},
|
1640 |
+
{
|
1641 |
+
"epoch": 14.77,
|
1642 |
+
"eval_loss": 0.2394973188638687,
|
1643 |
+
"eval_runtime": 452.3855,
|
1644 |
+
"eval_samples_per_second": 11.523,
|
1645 |
+
"eval_wer": 0.13962293508985762,
|
1646 |
+
"step": 35100
|
1647 |
+
},
|
1648 |
+
{
|
1649 |
+
"epoch": 14.9,
|
1650 |
+
"learning_rate": 0.00020885221814071771,
|
1651 |
+
"loss": 0.0562,
|
1652 |
+
"step": 35400
|
1653 |
+
},
|
1654 |
+
{
|
1655 |
+
"epoch": 14.9,
|
1656 |
+
"eval_loss": 0.2461041510105133,
|
1657 |
+
"eval_runtime": 443.6609,
|
1658 |
+
"eval_samples_per_second": 11.75,
|
1659 |
+
"eval_wer": 0.13835222115609036,
|
1660 |
+
"step": 35400
|
1661 |
+
},
|
1662 |
+
{
|
1663 |
+
"epoch": 15.03,
|
1664 |
+
"learning_rate": 0.00020710596213619666,
|
1665 |
+
"loss": 0.055,
|
1666 |
+
"step": 35700
|
1667 |
+
},
|
1668 |
+
{
|
1669 |
+
"epoch": 15.03,
|
1670 |
+
"eval_loss": 0.261261522769928,
|
1671 |
+
"eval_runtime": 421.1886,
|
1672 |
+
"eval_samples_per_second": 12.377,
|
1673 |
+
"eval_wer": 0.14125671014755842,
|
1674 |
+
"step": 35700
|
1675 |
+
},
|
1676 |
+
{
|
1677 |
+
"epoch": 15.15,
|
1678 |
+
"learning_rate": 0.0002053597061316756,
|
1679 |
+
"loss": 0.0504,
|
1680 |
+
"step": 36000
|
1681 |
+
},
|
1682 |
+
{
|
1683 |
+
"epoch": 15.15,
|
1684 |
+
"eval_loss": 0.2512105405330658,
|
1685 |
+
"eval_runtime": 449.4571,
|
1686 |
+
"eval_samples_per_second": 11.598,
|
1687 |
+
"eval_wer": 0.14104924664816784,
|
1688 |
+
"step": 36000
|
1689 |
+
},
|
1690 |
+
{
|
1691 |
+
"epoch": 15.28,
|
1692 |
+
"learning_rate": 0.00020361345012715457,
|
1693 |
+
"loss": 0.0492,
|
1694 |
+
"step": 36300
|
1695 |
+
},
|
1696 |
+
{
|
1697 |
+
"epoch": 15.28,
|
1698 |
+
"eval_loss": 0.2652667164802551,
|
1699 |
+
"eval_runtime": 447.4649,
|
1700 |
+
"eval_samples_per_second": 11.65,
|
1701 |
+
"eval_wer": 0.1455875106973367,
|
1702 |
+
"step": 36300
|
1703 |
+
},
|
1704 |
+
{
|
1705 |
+
"epoch": 15.4,
|
1706 |
+
"learning_rate": 0.0002018671941226335,
|
1707 |
+
"loss": 0.0495,
|
1708 |
+
"step": 36600
|
1709 |
+
},
|
1710 |
+
{
|
1711 |
+
"epoch": 15.4,
|
1712 |
+
"eval_loss": 0.25162389874458313,
|
1713 |
+
"eval_runtime": 420.3357,
|
1714 |
+
"eval_samples_per_second": 12.402,
|
1715 |
+
"eval_wer": 0.1370037084100516,
|
1716 |
+
"step": 36600
|
1717 |
+
},
|
1718 |
+
{
|
1719 |
+
"epoch": 15.53,
|
1720 |
+
"learning_rate": 0.00020012093811811244,
|
1721 |
+
"loss": 0.0549,
|
1722 |
+
"step": 36900
|
1723 |
+
},
|
1724 |
+
{
|
1725 |
+
"epoch": 15.53,
|
1726 |
+
"eval_loss": 0.2618192136287689,
|
1727 |
+
"eval_runtime": 445.4016,
|
1728 |
+
"eval_samples_per_second": 11.704,
|
1729 |
+
"eval_wer": 0.14042685614999612,
|
1730 |
+
"step": 36900
|
1731 |
+
},
|
1732 |
+
{
|
1733 |
+
"epoch": 15.66,
|
1734 |
+
"learning_rate": 0.0001983746821135914,
|
1735 |
+
"loss": 0.0497,
|
1736 |
+
"step": 37200
|
1737 |
+
},
|
1738 |
+
{
|
1739 |
+
"epoch": 15.66,
|
1740 |
+
"eval_loss": 0.254682332277298,
|
1741 |
+
"eval_runtime": 449.7454,
|
1742 |
+
"eval_samples_per_second": 11.591,
|
1743 |
+
"eval_wer": 0.13679624491066103,
|
1744 |
+
"step": 37200
|
1745 |
+
},
|
1746 |
+
{
|
1747 |
+
"epoch": 15.78,
|
1748 |
+
"learning_rate": 0.00019662842610907035,
|
1749 |
+
"loss": 0.0517,
|
1750 |
+
"step": 37500
|
1751 |
+
},
|
1752 |
+
{
|
1753 |
+
"epoch": 15.78,
|
1754 |
+
"eval_loss": 0.2562556862831116,
|
1755 |
+
"eval_runtime": 452.5481,
|
1756 |
+
"eval_samples_per_second": 11.519,
|
1757 |
+
"eval_wer": 0.13721117190944218,
|
1758 |
+
"step": 37500
|
1759 |
+
},
|
1760 |
+
{
|
1761 |
+
"epoch": 15.91,
|
1762 |
+
"learning_rate": 0.0001948821701045493,
|
1763 |
+
"loss": 0.0541,
|
1764 |
+
"step": 37800
|
1765 |
+
},
|
1766 |
+
{
|
1767 |
+
"epoch": 15.91,
|
1768 |
+
"eval_loss": 0.25422725081443787,
|
1769 |
+
"eval_runtime": 421.3988,
|
1770 |
+
"eval_samples_per_second": 12.371,
|
1771 |
+
"eval_wer": 0.13549959803946993,
|
1772 |
+
"step": 37800
|
1773 |
+
},
|
1774 |
+
{
|
1775 |
+
"epoch": 16.04,
|
1776 |
+
"learning_rate": 0.00019313591410002827,
|
1777 |
+
"loss": 0.0494,
|
1778 |
+
"step": 38100
|
1779 |
+
},
|
1780 |
+
{
|
1781 |
+
"epoch": 16.04,
|
1782 |
+
"eval_loss": 0.2502134144306183,
|
1783 |
+
"eval_runtime": 457.8032,
|
1784 |
+
"eval_samples_per_second": 11.387,
|
1785 |
+
"eval_wer": 0.13267290786027333,
|
1786 |
+
"step": 38100
|
1787 |
+
},
|
1788 |
+
{
|
1789 |
+
"epoch": 16.16,
|
1790 |
+
"learning_rate": 0.00019138965809550719,
|
1791 |
+
"loss": 0.0447,
|
1792 |
+
"step": 38400
|
1793 |
+
},
|
1794 |
+
{
|
1795 |
+
"epoch": 16.16,
|
1796 |
+
"eval_loss": 0.2713734805583954,
|
1797 |
+
"eval_runtime": 453.8402,
|
1798 |
+
"eval_samples_per_second": 11.486,
|
1799 |
+
"eval_wer": 0.13220611498664453,
|
1800 |
+
"step": 38400
|
1801 |
+
},
|
1802 |
+
{
|
1803 |
+
"epoch": 16.29,
|
1804 |
+
"learning_rate": 0.00018964340209098613,
|
1805 |
+
"loss": 0.0441,
|
1806 |
+
"step": 38700
|
1807 |
+
},
|
1808 |
+
{
|
1809 |
+
"epoch": 16.29,
|
1810 |
+
"eval_loss": 0.25368762016296387,
|
1811 |
+
"eval_runtime": 421.2775,
|
1812 |
+
"eval_samples_per_second": 12.374,
|
1813 |
+
"eval_wer": 0.13114286455226784,
|
1814 |
+
"step": 38700
|
1815 |
+
},
|
1816 |
+
{
|
1817 |
+
"epoch": 16.41,
|
1818 |
+
"learning_rate": 0.0001878971460864651,
|
1819 |
+
"loss": 0.0457,
|
1820 |
+
"step": 39000
|
1821 |
+
},
|
1822 |
+
{
|
1823 |
+
"epoch": 16.41,
|
1824 |
+
"eval_loss": 0.24593111872673035,
|
1825 |
+
"eval_runtime": 450.5495,
|
1826 |
+
"eval_samples_per_second": 11.57,
|
1827 |
+
"eval_wer": 0.13430668291797412,
|
1828 |
+
"step": 39000
|
1829 |
+
},
|
1830 |
+
{
|
1831 |
+
"epoch": 16.54,
|
1832 |
+
"learning_rate": 0.00018615089008194405,
|
1833 |
+
"loss": 0.0484,
|
1834 |
+
"step": 39300
|
1835 |
+
},
|
1836 |
+
{
|
1837 |
+
"epoch": 16.54,
|
1838 |
+
"eval_loss": 0.24097564816474915,
|
1839 |
+
"eval_runtime": 447.1986,
|
1840 |
+
"eval_samples_per_second": 11.657,
|
1841 |
+
"eval_wer": 0.1310909986774202,
|
1842 |
+
"step": 39300
|
1843 |
+
},
|
1844 |
+
{
|
1845 |
+
"epoch": 16.67,
|
1846 |
+
"learning_rate": 0.000184404634077423,
|
1847 |
+
"loss": 0.0435,
|
1848 |
+
"step": 39600
|
1849 |
+
},
|
1850 |
+
{
|
1851 |
+
"epoch": 16.67,
|
1852 |
+
"eval_loss": 0.24710460007190704,
|
1853 |
+
"eval_runtime": 420.3947,
|
1854 |
+
"eval_samples_per_second": 12.4,
|
1855 |
+
"eval_wer": 0.13150592567620134,
|
1856 |
+
"step": 39600
|
1857 |
+
},
|
1858 |
+
{
|
1859 |
+
"epoch": 16.79,
|
1860 |
+
"learning_rate": 0.00018265837807290196,
|
1861 |
+
"loss": 0.045,
|
1862 |
+
"step": 39900
|
1863 |
+
},
|
1864 |
+
{
|
1865 |
+
"epoch": 16.79,
|
1866 |
+
"eval_loss": 0.24996981024742126,
|
1867 |
+
"eval_runtime": 445.0069,
|
1868 |
+
"eval_samples_per_second": 11.714,
|
1869 |
+
"eval_wer": 0.12870516843442856,
|
1870 |
+
"step": 39900
|
1871 |
+
},
|
1872 |
+
{
|
1873 |
+
"epoch": 16.92,
|
1874 |
+
"learning_rate": 0.00018091212206838088,
|
1875 |
+
"loss": 0.0476,
|
1876 |
+
"step": 40200
|
1877 |
+
},
|
1878 |
+
{
|
1879 |
+
"epoch": 16.92,
|
1880 |
+
"eval_loss": 0.24158866703510284,
|
1881 |
+
"eval_runtime": 450.2986,
|
1882 |
+
"eval_samples_per_second": 11.577,
|
1883 |
+
"eval_wer": 0.12753818625035657,
|
1884 |
+
"step": 40200
|
1885 |
+
},
|
1886 |
+
{
|
1887 |
+
"epoch": 17.05,
|
1888 |
+
"learning_rate": 0.00017916586606385983,
|
1889 |
+
"loss": 0.0421,
|
1890 |
+
"step": 40500
|
1891 |
+
},
|
1892 |
+
{
|
1893 |
+
"epoch": 17.05,
|
1894 |
+
"eval_loss": 0.25376275181770325,
|
1895 |
+
"eval_runtime": 439.9451,
|
1896 |
+
"eval_samples_per_second": 11.849,
|
1897 |
+
"eval_wer": 0.12727885687611837,
|
1898 |
+
"step": 40500
|
1899 |
+
},
|
1900 |
+
{
|
1901 |
+
"epoch": 17.17,
|
1902 |
+
"learning_rate": 0.0001774196100593388,
|
1903 |
+
"loss": 0.043,
|
1904 |
+
"step": 40800
|
1905 |
+
},
|
1906 |
+
{
|
1907 |
+
"epoch": 17.17,
|
1908 |
+
"eval_loss": 0.2612881064414978,
|
1909 |
+
"eval_runtime": 420.2448,
|
1910 |
+
"eval_samples_per_second": 12.405,
|
1911 |
+
"eval_wer": 0.13638131791187988,
|
1912 |
+
"step": 40800
|
1913 |
+
},
|
1914 |
+
{
|
1915 |
+
"epoch": 17.3,
|
1916 |
+
"learning_rate": 0.00017567335405481774,
|
1917 |
+
"loss": 0.0421,
|
1918 |
+
"step": 41100
|
1919 |
+
},
|
1920 |
+
{
|
1921 |
+
"epoch": 17.3,
|
1922 |
+
"eval_loss": 0.25012627243995667,
|
1923 |
+
"eval_runtime": 442.3555,
|
1924 |
+
"eval_samples_per_second": 11.785,
|
1925 |
+
"eval_wer": 0.129949949430772,
|
1926 |
+
"step": 41100
|
1927 |
+
},
|
1928 |
+
{
|
1929 |
+
"epoch": 17.42,
|
1930 |
+
"learning_rate": 0.00017392709805029669,
|
1931 |
+
"loss": 0.0455,
|
1932 |
+
"step": 41400
|
1933 |
+
},
|
1934 |
+
{
|
1935 |
+
"epoch": 17.42,
|
1936 |
+
"eval_loss": 0.27013105154037476,
|
1937 |
+
"eval_runtime": 439.1226,
|
1938 |
+
"eval_samples_per_second": 11.871,
|
1939 |
+
"eval_wer": 0.13282850548481626,
|
1940 |
+
"step": 41400
|
1941 |
+
},
|
1942 |
+
{
|
1943 |
+
"epoch": 17.55,
|
1944 |
+
"learning_rate": 0.00017218084204577566,
|
1945 |
+
"loss": 0.0447,
|
1946 |
+
"step": 41700
|
1947 |
+
},
|
1948 |
+
{
|
1949 |
+
"epoch": 17.55,
|
1950 |
+
"eval_loss": 0.25483304262161255,
|
1951 |
+
"eval_runtime": 422.116,
|
1952 |
+
"eval_samples_per_second": 12.35,
|
1953 |
+
"eval_wer": 0.13098726692772492,
|
1954 |
+
"step": 41700
|
1955 |
+
},
|
1956 |
+
{
|
1957 |
+
"epoch": 17.68,
|
1958 |
+
"learning_rate": 0.0001704345860412546,
|
1959 |
+
"loss": 0.0422,
|
1960 |
+
"step": 42000
|
1961 |
+
},
|
1962 |
+
{
|
1963 |
+
"epoch": 17.68,
|
1964 |
+
"eval_loss": 0.25715696811676025,
|
1965 |
+
"eval_runtime": 455.1849,
|
1966 |
+
"eval_samples_per_second": 11.452,
|
1967 |
+
"eval_wer": 0.12795311324913772,
|
1968 |
+
"step": 42000
|
1969 |
+
},
|
1970 |
+
{
|
1971 |
+
"epoch": 17.8,
|
1972 |
+
"learning_rate": 0.00016868833003673352,
|
1973 |
+
"loss": 0.0421,
|
1974 |
+
"step": 42300
|
1975 |
+
},
|
1976 |
+
{
|
1977 |
+
"epoch": 17.8,
|
1978 |
+
"eval_loss": 0.25144657492637634,
|
1979 |
+
"eval_runtime": 438.538,
|
1980 |
+
"eval_samples_per_second": 11.887,
|
1981 |
+
"eval_wer": 0.13508467104068878,
|
1982 |
+
"step": 42300
|
1983 |
+
},
|
1984 |
+
{
|
1985 |
+
"epoch": 17.93,
|
1986 |
+
"learning_rate": 0.00016694207403221246,
|
1987 |
+
"loss": 0.0421,
|
1988 |
+
"step": 42600
|
1989 |
+
},
|
1990 |
+
{
|
1991 |
+
"epoch": 17.93,
|
1992 |
+
"eval_loss": 0.25014957785606384,
|
1993 |
+
"eval_runtime": 419.3857,
|
1994 |
+
"eval_samples_per_second": 12.43,
|
1995 |
+
"eval_wer": 0.12906822955836209,
|
1996 |
+
"step": 42600
|
1997 |
+
},
|
1998 |
+
{
|
1999 |
+
"epoch": 18.06,
|
2000 |
+
"learning_rate": 0.00016519581802769144,
|
2001 |
+
"loss": 0.0412,
|
2002 |
+
"step": 42900
|
2003 |
+
},
|
2004 |
+
{
|
2005 |
+
"epoch": 18.06,
|
2006 |
+
"eval_loss": 0.25013405084609985,
|
2007 |
+
"eval_runtime": 447.7509,
|
2008 |
+
"eval_samples_per_second": 11.643,
|
2009 |
+
"eval_wer": 0.12891263193381913,
|
2010 |
+
"step": 42900
|
2011 |
+
},
|
2012 |
+
{
|
2013 |
+
"epoch": 18.18,
|
2014 |
+
"learning_rate": 0.00016344956202317038,
|
2015 |
+
"loss": 0.0365,
|
2016 |
+
"step": 43200
|
2017 |
+
},
|
2018 |
+
{
|
2019 |
+
"epoch": 18.18,
|
2020 |
+
"eval_loss": 0.24244625866413116,
|
2021 |
+
"eval_runtime": 452.5128,
|
2022 |
+
"eval_samples_per_second": 11.52,
|
2023 |
+
"eval_wer": 0.13249137729830657,
|
2024 |
+
"step": 43200
|
2025 |
+
},
|
2026 |
+
{
|
2027 |
+
"epoch": 18.31,
|
2028 |
+
"learning_rate": 0.00016170330601864932,
|
2029 |
+
"loss": 0.037,
|
2030 |
+
"step": 43500
|
2031 |
+
},
|
2032 |
+
{
|
2033 |
+
"epoch": 18.31,
|
2034 |
+
"eval_loss": 0.24644562602043152,
|
2035 |
+
"eval_runtime": 447.1204,
|
2036 |
+
"eval_samples_per_second": 11.659,
|
2037 |
+
"eval_wer": 0.12963875418168616,
|
2038 |
+
"step": 43500
|
2039 |
+
},
|
2040 |
+
{
|
2041 |
+
"epoch": 18.43,
|
2042 |
+
"learning_rate": 0.0001599570500141283,
|
2043 |
+
"loss": 0.0398,
|
2044 |
+
"step": 43800
|
2045 |
+
},
|
2046 |
+
{
|
2047 |
+
"epoch": 18.43,
|
2048 |
+
"eval_loss": 0.24118588864803314,
|
2049 |
+
"eval_runtime": 420.9169,
|
2050 |
+
"eval_samples_per_second": 12.385,
|
2051 |
+
"eval_wer": 0.12551541713129846,
|
2052 |
+
"step": 43800
|
2053 |
+
},
|
2054 |
+
{
|
2055 |
+
"epoch": 18.56,
|
2056 |
+
"learning_rate": 0.00015821079400960721,
|
2057 |
+
"loss": 0.0398,
|
2058 |
+
"step": 44100
|
2059 |
+
},
|
2060 |
+
{
|
2061 |
+
"epoch": 18.56,
|
2062 |
+
"eval_loss": 0.2401241511106491,
|
2063 |
+
"eval_runtime": 443.2738,
|
2064 |
+
"eval_samples_per_second": 11.76,
|
2065 |
+
"eval_wer": 0.12642306994113223,
|
2066 |
+
"step": 44100
|
2067 |
+
},
|
2068 |
+
{
|
2069 |
+
"epoch": 18.69,
|
2070 |
+
"learning_rate": 0.00015646453800508616,
|
2071 |
+
"loss": 0.0404,
|
2072 |
+
"step": 44400
|
2073 |
+
},
|
2074 |
+
{
|
2075 |
+
"epoch": 18.69,
|
2076 |
+
"eval_loss": 0.24249568581581116,
|
2077 |
+
"eval_runtime": 454.0524,
|
2078 |
+
"eval_samples_per_second": 11.481,
|
2079 |
+
"eval_wer": 0.12416690438525972,
|
2080 |
+
"step": 44400
|
2081 |
+
},
|
2082 |
+
{
|
2083 |
+
"epoch": 18.81,
|
2084 |
+
"learning_rate": 0.00015471828200056513,
|
2085 |
+
"loss": 0.0385,
|
2086 |
+
"step": 44700
|
2087 |
+
},
|
2088 |
+
{
|
2089 |
+
"epoch": 18.81,
|
2090 |
+
"eval_loss": 0.2565571069717407,
|
2091 |
+
"eval_runtime": 421.1105,
|
2092 |
+
"eval_samples_per_second": 12.379,
|
2093 |
+
"eval_wer": 0.13007961411789112,
|
2094 |
+
"step": 44700
|
2095 |
+
},
|
2096 |
+
{
|
2097 |
+
"epoch": 18.94,
|
2098 |
+
"learning_rate": 0.00015297202599604407,
|
2099 |
+
"loss": 0.0409,
|
2100 |
+
"step": 45000
|
2101 |
+
},
|
2102 |
+
{
|
2103 |
+
"epoch": 18.94,
|
2104 |
+
"eval_loss": 0.24367791414260864,
|
2105 |
+
"eval_runtime": 440.1807,
|
2106 |
+
"eval_samples_per_second": 11.843,
|
2107 |
+
"eval_wer": 0.12823837556079978,
|
2108 |
+
"step": 45000
|
2109 |
+
},
|
2110 |
+
{
|
2111 |
+
"epoch": 19.07,
|
2112 |
+
"learning_rate": 0.00015122576999152302,
|
2113 |
+
"loss": 0.0386,
|
2114 |
+
"step": 45300
|
2115 |
+
},
|
2116 |
+
{
|
2117 |
+
"epoch": 19.07,
|
2118 |
+
"eval_loss": 0.2512024939060211,
|
2119 |
+
"eval_runtime": 448.55,
|
2120 |
+
"eval_samples_per_second": 11.622,
|
2121 |
+
"eval_wer": 0.1251004901325173,
|
2122 |
+
"step": 45300
|
2123 |
+
},
|
2124 |
+
{
|
2125 |
+
"epoch": 19.19,
|
2126 |
+
"learning_rate": 0.000149479513987002,
|
2127 |
+
"loss": 0.0382,
|
2128 |
+
"step": 45600
|
2129 |
+
},
|
2130 |
+
{
|
2131 |
+
"epoch": 19.19,
|
2132 |
+
"eval_loss": 0.26347675919532776,
|
2133 |
+
"eval_runtime": 420.744,
|
2134 |
+
"eval_samples_per_second": 12.39,
|
2135 |
+
"eval_wer": 0.1246077643214647,
|
2136 |
+
"step": 45600
|
2137 |
+
},
|
2138 |
+
{
|
2139 |
+
"epoch": 19.32,
|
2140 |
+
"learning_rate": 0.00014773325798248094,
|
2141 |
+
"loss": 0.0356,
|
2142 |
+
"step": 45900
|
2143 |
+
},
|
2144 |
+
{
|
2145 |
+
"epoch": 19.32,
|
2146 |
+
"eval_loss": 0.24348150193691254,
|
2147 |
+
"eval_runtime": 444.991,
|
2148 |
+
"eval_samples_per_second": 11.715,
|
2149 |
+
"eval_wer": 0.11929151214958118,
|
2150 |
+
"step": 45900
|
2151 |
+
},
|
2152 |
+
{
|
2153 |
+
"epoch": 19.44,
|
2154 |
+
"learning_rate": 0.00014598700197795985,
|
2155 |
+
"loss": 0.0347,
|
2156 |
+
"step": 46200
|
2157 |
+
},
|
2158 |
+
{
|
2159 |
+
"epoch": 19.44,
|
2160 |
+
"eval_loss": 0.24703486263751984,
|
2161 |
+
"eval_runtime": 450.7542,
|
2162 |
+
"eval_samples_per_second": 11.565,
|
2163 |
+
"eval_wer": 0.11866912165140946,
|
2164 |
+
"step": 46200
|
2165 |
+
},
|
2166 |
+
{
|
2167 |
+
"epoch": 19.57,
|
2168 |
+
"learning_rate": 0.00014424074597343882,
|
2169 |
+
"loss": 0.0351,
|
2170 |
+
"step": 46500
|
2171 |
+
},
|
2172 |
+
{
|
2173 |
+
"epoch": 19.57,
|
2174 |
+
"eval_loss": 0.25269144773483276,
|
2175 |
+
"eval_runtime": 420.9044,
|
2176 |
+
"eval_samples_per_second": 12.385,
|
2177 |
+
"eval_wer": 0.12250719639013512,
|
2178 |
+
"step": 46500
|
2179 |
+
},
|
2180 |
+
{
|
2181 |
+
"epoch": 19.7,
|
2182 |
+
"learning_rate": 0.00014249448996891777,
|
2183 |
+
"loss": 0.0387,
|
2184 |
+
"step": 46800
|
2185 |
+
},
|
2186 |
+
{
|
2187 |
+
"epoch": 19.7,
|
2188 |
+
"eval_loss": 0.2696446180343628,
|
2189 |
+
"eval_runtime": 447.6183,
|
2190 |
+
"eval_samples_per_second": 11.646,
|
2191 |
+
"eval_wer": 0.12899043074609062,
|
2192 |
+
"step": 46800
|
2193 |
+
},
|
2194 |
+
{
|
2195 |
+
"epoch": 19.82,
|
2196 |
+
"learning_rate": 0.0001407482339643967,
|
2197 |
+
"loss": 0.0381,
|
2198 |
+
"step": 47100
|
2199 |
+
},
|
2200 |
+
{
|
2201 |
+
"epoch": 19.82,
|
2202 |
+
"eval_loss": 0.24737927317619324,
|
2203 |
+
"eval_runtime": 454.8982,
|
2204 |
+
"eval_samples_per_second": 11.46,
|
2205 |
+
"eval_wer": 0.123544513887088,
|
2206 |
+
"step": 47100
|
2207 |
+
},
|
2208 |
+
{
|
2209 |
+
"epoch": 19.95,
|
2210 |
+
"learning_rate": 0.00013900197795987568,
|
2211 |
+
"loss": 0.0353,
|
2212 |
+
"step": 47400
|
2213 |
+
},
|
2214 |
+
{
|
2215 |
+
"epoch": 19.95,
|
2216 |
+
"eval_loss": 0.24795196950435638,
|
2217 |
+
"eval_runtime": 453.6312,
|
2218 |
+
"eval_samples_per_second": 11.492,
|
2219 |
+
"eval_wer": 0.12318145276315448,
|
2220 |
+
"step": 47400
|
2221 |
+
},
|
2222 |
+
{
|
2223 |
+
"epoch": 20.08,
|
2224 |
+
"learning_rate": 0.00013725572195535463,
|
2225 |
+
"loss": 0.0339,
|
2226 |
+
"step": 47700
|
2227 |
+
},
|
2228 |
+
{
|
2229 |
+
"epoch": 20.08,
|
2230 |
+
"eval_loss": 0.25215083360671997,
|
2231 |
+
"eval_runtime": 420.9875,
|
2232 |
+
"eval_samples_per_second": 12.383,
|
2233 |
+
"eval_wer": 0.12250719639013512,
|
2234 |
+
"step": 47700
|
2235 |
+
},
|
2236 |
+
{
|
2237 |
+
"epoch": 20.2,
|
2238 |
+
"learning_rate": 0.00013550946595083355,
|
2239 |
+
"loss": 0.0329,
|
2240 |
+
"step": 48000
|
2241 |
+
},
|
2242 |
+
{
|
2243 |
+
"epoch": 20.2,
|
2244 |
+
"eval_loss": 0.25695204734802246,
|
2245 |
+
"eval_runtime": 443.8662,
|
2246 |
+
"eval_samples_per_second": 11.745,
|
2247 |
+
"eval_wer": 0.11952490858639558,
|
2248 |
+
"step": 48000
|
2249 |
+
},
|
2250 |
+
{
|
2251 |
+
"epoch": 20.33,
|
2252 |
+
"learning_rate": 0.00013376320994631252,
|
2253 |
+
"loss": 0.0316,
|
2254 |
+
"step": 48300
|
2255 |
+
},
|
2256 |
+
{
|
2257 |
+
"epoch": 20.33,
|
2258 |
+
"eval_loss": 0.25657811760902405,
|
2259 |
+
"eval_runtime": 443.4461,
|
2260 |
+
"eval_samples_per_second": 11.756,
|
2261 |
+
"eval_wer": 0.11999170146002437,
|
2262 |
+
"step": 48300
|
2263 |
+
},
|
2264 |
+
{
|
2265 |
+
"epoch": 20.45,
|
2266 |
+
"learning_rate": 0.00013201695394179146,
|
2267 |
+
"loss": 0.0339,
|
2268 |
+
"step": 48600
|
2269 |
+
},
|
2270 |
+
{
|
2271 |
+
"epoch": 20.45,
|
2272 |
+
"eval_loss": 0.2606305181980133,
|
2273 |
+
"eval_runtime": 420.9039,
|
2274 |
+
"eval_samples_per_second": 12.385,
|
2275 |
+
"eval_wer": 0.11809859702808537,
|
2276 |
+
"step": 48600
|
2277 |
+
},
|
2278 |
+
{
|
2279 |
+
"epoch": 20.58,
|
2280 |
+
"learning_rate": 0.0001302706979372704,
|
2281 |
+
"loss": 0.0324,
|
2282 |
+
"step": 48900
|
2283 |
+
},
|
2284 |
+
{
|
2285 |
+
"epoch": 20.58,
|
2286 |
+
"eval_loss": 0.2572144865989685,
|
2287 |
+
"eval_runtime": 456.9807,
|
2288 |
+
"eval_samples_per_second": 11.407,
|
2289 |
+
"eval_wer": 0.11488291278753145,
|
2290 |
+
"step": 48900
|
2291 |
+
},
|
2292 |
+
{
|
2293 |
+
"epoch": 20.71,
|
2294 |
+
"learning_rate": 0.00012852444193274935,
|
2295 |
+
"loss": 0.0306,
|
2296 |
+
"step": 49200
|
2297 |
+
},
|
2298 |
+
{
|
2299 |
+
"epoch": 20.71,
|
2300 |
+
"eval_loss": 0.25449544191360474,
|
2301 |
+
"eval_runtime": 449.6003,
|
2302 |
+
"eval_samples_per_second": 11.595,
|
2303 |
+
"eval_wer": 0.11783926765384715,
|
2304 |
+
"step": 49200
|
2305 |
+
},
|
2306 |
+
{
|
2307 |
+
"epoch": 20.83,
|
2308 |
+
"learning_rate": 0.00012677818592822832,
|
2309 |
+
"loss": 0.0339,
|
2310 |
+
"step": 49500
|
2311 |
+
},
|
2312 |
+
{
|
2313 |
+
"epoch": 20.83,
|
2314 |
+
"eval_loss": 0.24496378004550934,
|
2315 |
+
"eval_runtime": 420.0292,
|
2316 |
+
"eval_samples_per_second": 12.411,
|
2317 |
+
"eval_wer": 0.1147013822255647,
|
2318 |
+
"step": 49500
|
2319 |
+
},
|
2320 |
+
{
|
2321 |
+
"epoch": 20.96,
|
2322 |
+
"learning_rate": 0.00012503192992370724,
|
2323 |
+
"loss": 0.0344,
|
2324 |
+
"step": 49800
|
2325 |
+
},
|
2326 |
+
{
|
2327 |
+
"epoch": 20.96,
|
2328 |
+
"eval_loss": 0.23759284615516663,
|
2329 |
+
"eval_runtime": 450.494,
|
2330 |
+
"eval_samples_per_second": 11.572,
|
2331 |
+
"eval_wer": 0.12004356733487202,
|
2332 |
+
"step": 49800
|
2333 |
+
},
|
2334 |
+
{
|
2335 |
+
"epoch": 21.09,
|
2336 |
+
"learning_rate": 0.00012328567391918619,
|
2337 |
+
"loss": 0.0302,
|
2338 |
+
"step": 50100
|
2339 |
+
},
|
2340 |
+
{
|
2341 |
+
"epoch": 21.09,
|
2342 |
+
"eval_loss": 0.2617592215538025,
|
2343 |
+
"eval_runtime": 450.3195,
|
2344 |
+
"eval_samples_per_second": 11.576,
|
2345 |
+
"eval_wer": 0.11584243147221286,
|
2346 |
+
"step": 50100
|
2347 |
+
},
|
2348 |
+
{
|
2349 |
+
"epoch": 21.21,
|
2350 |
+
"learning_rate": 0.00012153941791466516,
|
2351 |
+
"loss": 0.0312,
|
2352 |
+
"step": 50400
|
2353 |
+
},
|
2354 |
+
{
|
2355 |
+
"epoch": 21.21,
|
2356 |
+
"eval_loss": 0.25956034660339355,
|
2357 |
+
"eval_runtime": 440.9223,
|
2358 |
+
"eval_samples_per_second": 11.823,
|
2359 |
+
"eval_wer": 0.11462358341329322,
|
2360 |
+
"step": 50400
|
2361 |
+
},
|
2362 |
+
{
|
2363 |
+
"epoch": 21.34,
|
2364 |
+
"learning_rate": 0.0001197931619101441,
|
2365 |
+
"loss": 0.0296,
|
2366 |
+
"step": 50700
|
2367 |
+
},
|
2368 |
+
{
|
2369 |
+
"epoch": 21.34,
|
2370 |
+
"eval_loss": 0.25406599044799805,
|
2371 |
+
"eval_runtime": 422.3783,
|
2372 |
+
"eval_samples_per_second": 12.342,
|
2373 |
+
"eval_wer": 0.11480511397525998,
|
2374 |
+
"step": 50700
|
2375 |
+
},
|
2376 |
+
{
|
2377 |
+
"epoch": 21.46,
|
2378 |
+
"learning_rate": 0.00011804690590562305,
|
2379 |
+
"loss": 0.0288,
|
2380 |
+
"step": 51000
|
2381 |
+
},
|
2382 |
+
{
|
2383 |
+
"epoch": 21.46,
|
2384 |
+
"eval_loss": 0.25333333015441895,
|
2385 |
+
"eval_runtime": 451.8621,
|
2386 |
+
"eval_samples_per_second": 11.537,
|
2387 |
+
"eval_wer": 0.11270454604393039,
|
2388 |
+
"step": 51000
|
2389 |
+
},
|
2390 |
+
{
|
2391 |
+
"epoch": 21.59,
|
2392 |
+
"learning_rate": 0.000116300649901102,
|
2393 |
+
"loss": 0.0305,
|
2394 |
+
"step": 51300
|
2395 |
+
},
|
2396 |
+
{
|
2397 |
+
"epoch": 21.59,
|
2398 |
+
"eval_loss": 0.2583317756652832,
|
2399 |
+
"eval_runtime": 448.0342,
|
2400 |
+
"eval_samples_per_second": 11.635,
|
2401 |
+
"eval_wer": 0.11542750447343171,
|
2402 |
+
"step": 51300
|
2403 |
+
},
|
2404 |
+
{
|
2405 |
+
"epoch": 21.72,
|
2406 |
+
"learning_rate": 0.00011455439389658095,
|
2407 |
+
"loss": 0.0298,
|
2408 |
+
"step": 51600
|
2409 |
+
},
|
2410 |
+
{
|
2411 |
+
"epoch": 21.72,
|
2412 |
+
"eval_loss": 0.23815418779850006,
|
2413 |
+
"eval_runtime": 423.0616,
|
2414 |
+
"eval_samples_per_second": 12.322,
|
2415 |
+
"eval_wer": 0.11524597391146495,
|
2416 |
+
"step": 51600
|
2417 |
+
},
|
2418 |
+
{
|
2419 |
+
"epoch": 21.84,
|
2420 |
+
"learning_rate": 0.0001128081378920599,
|
2421 |
+
"loss": 0.0291,
|
2422 |
+
"step": 51900
|
2423 |
+
},
|
2424 |
+
{
|
2425 |
+
"epoch": 21.84,
|
2426 |
+
"eval_loss": 0.2578916549682617,
|
2427 |
+
"eval_runtime": 448.7674,
|
2428 |
+
"eval_samples_per_second": 11.616,
|
2429 |
+
"eval_wer": 0.1138974611654262,
|
2430 |
+
"step": 51900
|
2431 |
+
},
|
2432 |
+
{
|
2433 |
+
"epoch": 21.97,
|
2434 |
+
"learning_rate": 0.00011106188188753885,
|
2435 |
+
"loss": 0.0295,
|
2436 |
+
"step": 52200
|
2437 |
+
},
|
2438 |
+
{
|
2439 |
+
"epoch": 21.97,
|
2440 |
+
"eval_loss": 0.26773908734321594,
|
2441 |
+
"eval_runtime": 447.9798,
|
2442 |
+
"eval_samples_per_second": 11.637,
|
2443 |
+
"eval_wer": 0.1136640647286118,
|
2444 |
+
"step": 52200
|
2445 |
+
},
|
2446 |
+
{
|
2447 |
+
"epoch": 22.1,
|
2448 |
+
"learning_rate": 0.0001093156258830178,
|
2449 |
+
"loss": 0.0279,
|
2450 |
+
"step": 52500
|
2451 |
+
},
|
2452 |
+
{
|
2453 |
+
"epoch": 22.1,
|
2454 |
+
"eval_loss": 0.24904872477054596,
|
2455 |
+
"eval_runtime": 420.4369,
|
2456 |
+
"eval_samples_per_second": 12.399,
|
2457 |
+
"eval_wer": 0.1128342107310495,
|
2458 |
+
"step": 52500
|
2459 |
+
},
|
2460 |
+
{
|
2461 |
+
"epoch": 22.22,
|
2462 |
+
"learning_rate": 0.00010756936987849674,
|
2463 |
+
"loss": 0.0283,
|
2464 |
+
"step": 52800
|
2465 |
+
},
|
2466 |
+
{
|
2467 |
+
"epoch": 22.22,
|
2468 |
+
"eval_loss": 0.2628434896469116,
|
2469 |
+
"eval_runtime": 449.8543,
|
2470 |
+
"eval_samples_per_second": 11.588,
|
2471 |
+
"eval_wer": 0.11433832110163118,
|
2472 |
+
"step": 52800
|
2473 |
+
},
|
2474 |
+
{
|
2475 |
+
"epoch": 22.35,
|
2476 |
+
"learning_rate": 0.0001058231138739757,
|
2477 |
+
"loss": 0.0277,
|
2478 |
+
"step": 53100
|
2479 |
+
},
|
2480 |
+
{
|
2481 |
+
"epoch": 22.35,
|
2482 |
+
"eval_loss": 0.26064106822013855,
|
2483 |
+
"eval_runtime": 447.314,
|
2484 |
+
"eval_samples_per_second": 11.654,
|
2485 |
+
"eval_wer": 0.11280827779362568,
|
2486 |
+
"step": 53100
|
2487 |
+
},
|
2488 |
+
{
|
2489 |
+
"epoch": 22.47,
|
2490 |
+
"learning_rate": 0.00010407685786945464,
|
2491 |
+
"loss": 0.0263,
|
2492 |
+
"step": 53400
|
2493 |
+
},
|
2494 |
+
{
|
2495 |
+
"epoch": 22.47,
|
2496 |
+
"eval_loss": 0.264713853597641,
|
2497 |
+
"eval_runtime": 446.9042,
|
2498 |
+
"eval_samples_per_second": 11.665,
|
2499 |
+
"eval_wer": 0.11210808848318249,
|
2500 |
+
"step": 53400
|
2501 |
+
},
|
2502 |
+
{
|
2503 |
+
"epoch": 22.6,
|
2504 |
+
"learning_rate": 0.0001023306018649336,
|
2505 |
+
"loss": 0.0278,
|
2506 |
+
"step": 53700
|
2507 |
+
},
|
2508 |
+
{
|
2509 |
+
"epoch": 22.6,
|
2510 |
+
"eval_loss": 0.2553078830242157,
|
2511 |
+
"eval_runtime": 421.9865,
|
2512 |
+
"eval_samples_per_second": 12.353,
|
2513 |
+
"eval_wer": 0.1103446487383626,
|
2514 |
+
"step": 53700
|
2515 |
+
},
|
2516 |
+
{
|
2517 |
+
"epoch": 22.73,
|
2518 |
+
"learning_rate": 0.00010058434586041255,
|
2519 |
+
"loss": 0.028,
|
2520 |
+
"step": 54000
|
2521 |
+
},
|
2522 |
+
{
|
2523 |
+
"epoch": 22.73,
|
2524 |
+
"eval_loss": 0.25996631383895874,
|
2525 |
+
"eval_runtime": 448.115,
|
2526 |
+
"eval_samples_per_second": 11.633,
|
2527 |
+
"eval_wer": 0.11226368610772543,
|
2528 |
+
"step": 54000
|
2529 |
+
},
|
2530 |
+
{
|
2531 |
+
"epoch": 22.85,
|
2532 |
+
"learning_rate": 9.883808985589149e-05,
|
2533 |
+
"loss": 0.0289,
|
2534 |
+
"step": 54300
|
2535 |
+
},
|
2536 |
+
{
|
2537 |
+
"epoch": 22.85,
|
2538 |
+
"eval_loss": 0.2588217258453369,
|
2539 |
+
"eval_runtime": 457.0011,
|
2540 |
+
"eval_samples_per_second": 11.407,
|
2541 |
+
"eval_wer": 0.11221182023287778,
|
2542 |
+
"step": 54300
|
2543 |
+
},
|
2544 |
+
{
|
2545 |
+
"epoch": 22.98,
|
2546 |
+
"learning_rate": 9.709183385137043e-05,
|
2547 |
+
"loss": 0.0288,
|
2548 |
+
"step": 54600
|
2549 |
+
},
|
2550 |
+
{
|
2551 |
+
"epoch": 22.98,
|
2552 |
+
"eval_loss": 0.2563655972480774,
|
2553 |
+
"eval_runtime": 421.5067,
|
2554 |
+
"eval_samples_per_second": 12.368,
|
2555 |
+
"eval_wer": 0.1130676071678639,
|
2556 |
+
"step": 54600
|
2557 |
+
},
|
2558 |
+
{
|
2559 |
+
"epoch": 23.11,
|
2560 |
+
"learning_rate": 9.534557784684939e-05,
|
2561 |
+
"loss": 0.0259,
|
2562 |
+
"step": 54900
|
2563 |
+
},
|
2564 |
+
{
|
2565 |
+
"epoch": 23.11,
|
2566 |
+
"eval_loss": 0.25361210107803345,
|
2567 |
+
"eval_runtime": 454.0477,
|
2568 |
+
"eval_samples_per_second": 11.481,
|
2569 |
+
"eval_wer": 0.10935919711625736,
|
2570 |
+
"step": 54900
|
2571 |
+
},
|
2572 |
+
{
|
2573 |
+
"epoch": 23.23,
|
2574 |
+
"learning_rate": 9.359932184232834e-05,
|
2575 |
+
"loss": 0.0278,
|
2576 |
+
"step": 55200
|
2577 |
+
},
|
2578 |
+
{
|
2579 |
+
"epoch": 23.23,
|
2580 |
+
"eval_loss": 0.26548078656196594,
|
2581 |
+
"eval_runtime": 453.6451,
|
2582 |
+
"eval_samples_per_second": 11.491,
|
2583 |
+
"eval_wer": 0.10974819117761468,
|
2584 |
+
"step": 55200
|
2585 |
+
},
|
2586 |
+
{
|
2587 |
+
"epoch": 23.36,
|
2588 |
+
"learning_rate": 9.185306583780728e-05,
|
2589 |
+
"loss": 0.0254,
|
2590 |
+
"step": 55500
|
2591 |
+
},
|
2592 |
+
{
|
2593 |
+
"epoch": 23.36,
|
2594 |
+
"eval_loss": 0.27383095026016235,
|
2595 |
+
"eval_runtime": 421.4102,
|
2596 |
+
"eval_samples_per_second": 12.37,
|
2597 |
+
"eval_wer": 0.11286014366847333,
|
2598 |
+
"step": 55500
|
2599 |
+
},
|
2600 |
+
{
|
2601 |
+
"epoch": 23.48,
|
2602 |
+
"learning_rate": 9.010680983328624e-05,
|
2603 |
+
"loss": 0.0276,
|
2604 |
+
"step": 55800
|
2605 |
+
},
|
2606 |
+
{
|
2607 |
+
"epoch": 23.48,
|
2608 |
+
"eval_loss": 0.2656802833080292,
|
2609 |
+
"eval_runtime": 449.064,
|
2610 |
+
"eval_samples_per_second": 11.609,
|
2611 |
+
"eval_wer": 0.1109411062991105,
|
2612 |
+
"step": 55800
|
2613 |
+
},
|
2614 |
+
{
|
2615 |
+
"epoch": 23.61,
|
2616 |
+
"learning_rate": 8.836055382876518e-05,
|
2617 |
+
"loss": 0.026,
|
2618 |
+
"step": 56100
|
2619 |
+
},
|
2620 |
+
{
|
2621 |
+
"epoch": 23.61,
|
2622 |
+
"eval_loss": 0.27263152599334717,
|
2623 |
+
"eval_runtime": 447.0215,
|
2624 |
+
"eval_samples_per_second": 11.662,
|
2625 |
+
"eval_wer": 0.11044838048805788,
|
2626 |
+
"step": 56100
|
2627 |
+
},
|
2628 |
+
{
|
2629 |
+
"epoch": 23.74,
|
2630 |
+
"learning_rate": 8.661429782424413e-05,
|
2631 |
+
"loss": 0.0258,
|
2632 |
+
"step": 56400
|
2633 |
+
},
|
2634 |
+
{
|
2635 |
+
"epoch": 23.74,
|
2636 |
+
"eval_loss": 0.26607978343963623,
|
2637 |
+
"eval_runtime": 422.7932,
|
2638 |
+
"eval_samples_per_second": 12.33,
|
2639 |
+
"eval_wer": 0.10775135499598039,
|
2640 |
+
"step": 56400
|
2641 |
+
},
|
2642 |
+
{
|
2643 |
+
"epoch": 23.86,
|
2644 |
+
"learning_rate": 8.486804181972309e-05,
|
2645 |
+
"loss": 0.0252,
|
2646 |
+
"step": 56700
|
2647 |
+
},
|
2648 |
+
{
|
2649 |
+
"epoch": 23.86,
|
2650 |
+
"eval_loss": 0.26961830258369446,
|
2651 |
+
"eval_runtime": 442.9007,
|
2652 |
+
"eval_samples_per_second": 11.77,
|
2653 |
+
"eval_wer": 0.10891833718005238,
|
2654 |
+
"step": 56700
|
2655 |
+
},
|
2656 |
+
{
|
2657 |
+
"epoch": 23.99,
|
2658 |
+
"learning_rate": 8.312178581520203e-05,
|
2659 |
+
"loss": 0.0255,
|
2660 |
+
"step": 57000
|
2661 |
+
},
|
2662 |
+
{
|
2663 |
+
"epoch": 23.99,
|
2664 |
+
"eval_loss": 0.25439324975013733,
|
2665 |
+
"eval_runtime": 447.4952,
|
2666 |
+
"eval_samples_per_second": 11.649,
|
2667 |
+
"eval_wer": 0.10808848318249008,
|
2668 |
+
"step": 57000
|
2669 |
+
},
|
2670 |
+
{
|
2671 |
+
"epoch": 24.12,
|
2672 |
+
"learning_rate": 8.137552981068098e-05,
|
2673 |
+
"loss": 0.0235,
|
2674 |
+
"step": 57300
|
2675 |
+
},
|
2676 |
+
{
|
2677 |
+
"epoch": 24.12,
|
2678 |
+
"eval_loss": 0.26512473821640015,
|
2679 |
+
"eval_runtime": 447.2758,
|
2680 |
+
"eval_samples_per_second": 11.655,
|
2681 |
+
"eval_wer": 0.1096444594279194,
|
2682 |
+
"step": 57300
|
2683 |
+
},
|
2684 |
+
{
|
2685 |
+
"epoch": 24.24,
|
2686 |
+
"learning_rate": 7.962927380615993e-05,
|
2687 |
+
"loss": 0.0237,
|
2688 |
+
"step": 57600
|
2689 |
+
},
|
2690 |
+
{
|
2691 |
+
"epoch": 24.24,
|
2692 |
+
"eval_loss": 0.2661215364933014,
|
2693 |
+
"eval_runtime": 422.3577,
|
2694 |
+
"eval_samples_per_second": 12.343,
|
2695 |
+
"eval_wer": 0.10912580067944296,
|
2696 |
+
"step": 57600
|
2697 |
+
},
|
2698 |
+
{
|
2699 |
+
"epoch": 24.37,
|
2700 |
+
"learning_rate": 7.788301780163888e-05,
|
2701 |
+
"loss": 0.0259,
|
2702 |
+
"step": 57900
|
2703 |
+
},
|
2704 |
+
{
|
2705 |
+
"epoch": 24.37,
|
2706 |
+
"eval_loss": 0.2603491246700287,
|
2707 |
+
"eval_runtime": 447.2071,
|
2708 |
+
"eval_samples_per_second": 11.657,
|
2709 |
+
"eval_wer": 0.1081144161199139,
|
2710 |
+
"step": 57900
|
2711 |
+
},
|
2712 |
+
{
|
2713 |
+
"epoch": 24.49,
|
2714 |
+
"learning_rate": 7.613676179711782e-05,
|
2715 |
+
"loss": 0.022,
|
2716 |
+
"step": 58200
|
2717 |
+
},
|
2718 |
+
{
|
2719 |
+
"epoch": 24.49,
|
2720 |
+
"eval_loss": 0.25694188475608826,
|
2721 |
+
"eval_runtime": 447.9008,
|
2722 |
+
"eval_samples_per_second": 11.639,
|
2723 |
+
"eval_wer": 0.10653250693706076,
|
2724 |
+
"step": 58200
|
2725 |
+
},
|
2726 |
+
{
|
2727 |
+
"epoch": 24.62,
|
2728 |
+
"learning_rate": 7.439050579259678e-05,
|
2729 |
+
"loss": 0.0239,
|
2730 |
+
"step": 58500
|
2731 |
+
},
|
2732 |
+
{
|
2733 |
+
"epoch": 24.62,
|
2734 |
+
"eval_loss": 0.2633558511734009,
|
2735 |
+
"eval_runtime": 422.3281,
|
2736 |
+
"eval_samples_per_second": 12.343,
|
2737 |
+
"eval_wer": 0.10868494074323799,
|
2738 |
+
"step": 58500
|
2739 |
+
},
|
2740 |
+
{
|
2741 |
+
"epoch": 24.75,
|
2742 |
+
"learning_rate": 7.264424978807573e-05,
|
2743 |
+
"loss": 0.0226,
|
2744 |
+
"step": 58800
|
2745 |
+
},
|
2746 |
+
{
|
2747 |
+
"epoch": 24.75,
|
2748 |
+
"eval_loss": 0.2715343236923218,
|
2749 |
+
"eval_runtime": 451.351,
|
2750 |
+
"eval_samples_per_second": 11.55,
|
2751 |
+
"eval_wer": 0.10580638468919375,
|
2752 |
+
"step": 58800
|
2753 |
+
},
|
2754 |
+
{
|
2755 |
+
"epoch": 24.87,
|
2756 |
+
"learning_rate": 7.089799378355467e-05,
|
2757 |
+
"loss": 0.0236,
|
2758 |
+
"step": 59100
|
2759 |
+
},
|
2760 |
+
{
|
2761 |
+
"epoch": 24.87,
|
2762 |
+
"eval_loss": 0.25477010011672974,
|
2763 |
+
"eval_runtime": 442.3516,
|
2764 |
+
"eval_samples_per_second": 11.785,
|
2765 |
+
"eval_wer": 0.10487279894193616,
|
2766 |
+
"step": 59100
|
2767 |
+
},
|
2768 |
+
{
|
2769 |
+
"epoch": 25.0,
|
2770 |
+
"learning_rate": 6.915173777903363e-05,
|
2771 |
+
"loss": 0.0221,
|
2772 |
+
"step": 59400
|
2773 |
+
},
|
2774 |
+
{
|
2775 |
+
"epoch": 25.0,
|
2776 |
+
"eval_loss": 0.26137542724609375,
|
2777 |
+
"eval_runtime": 423.9191,
|
2778 |
+
"eval_samples_per_second": 12.297,
|
2779 |
+
"eval_wer": 0.10531365887814113,
|
2780 |
+
"step": 59400
|
2781 |
+
},
|
2782 |
+
{
|
2783 |
+
"epoch": 25.13,
|
2784 |
+
"learning_rate": 6.740548177451257e-05,
|
2785 |
+
"loss": 0.0228,
|
2786 |
+
"step": 59700
|
2787 |
+
},
|
2788 |
+
{
|
2789 |
+
"epoch": 25.13,
|
2790 |
+
"eval_loss": 0.2630428969860077,
|
2791 |
+
"eval_runtime": 457.649,
|
2792 |
+
"eval_samples_per_second": 11.391,
|
2793 |
+
"eval_wer": 0.10508026244132673,
|
2794 |
+
"step": 59700
|
2795 |
+
},
|
2796 |
+
{
|
2797 |
+
"epoch": 25.25,
|
2798 |
+
"learning_rate": 6.565922576999152e-05,
|
2799 |
+
"loss": 0.0211,
|
2800 |
+
"step": 60000
|
2801 |
+
},
|
2802 |
+
{
|
2803 |
+
"epoch": 25.25,
|
2804 |
+
"eval_loss": 0.26376327872276306,
|
2805 |
+
"eval_runtime": 438.7038,
|
2806 |
+
"eval_samples_per_second": 11.883,
|
2807 |
+
"eval_wer": 0.10510619537875054,
|
2808 |
+
"step": 60000
|
2809 |
+
},
|
2810 |
+
{
|
2811 |
+
"epoch": 25.38,
|
2812 |
+
"learning_rate": 6.391296976547048e-05,
|
2813 |
+
"loss": 0.0216,
|
2814 |
+
"step": 60300
|
2815 |
+
},
|
2816 |
+
{
|
2817 |
+
"epoch": 25.38,
|
2818 |
+
"eval_loss": 0.27177637815475464,
|
2819 |
+
"eval_runtime": 453.0435,
|
2820 |
+
"eval_samples_per_second": 11.507,
|
2821 |
+
"eval_wer": 0.10443193900573118,
|
2822 |
+
"step": 60300
|
2823 |
+
},
|
2824 |
+
{
|
2825 |
+
"epoch": 25.51,
|
2826 |
+
"learning_rate": 6.216671376094942e-05,
|
2827 |
+
"loss": 0.0241,
|
2828 |
+
"step": 60600
|
2829 |
+
},
|
2830 |
+
{
|
2831 |
+
"epoch": 25.51,
|
2832 |
+
"eval_loss": 0.2619257867336273,
|
2833 |
+
"eval_runtime": 422.4781,
|
2834 |
+
"eval_samples_per_second": 12.339,
|
2835 |
+
"eval_wer": 0.10329088975908302,
|
2836 |
+
"step": 60600
|
2837 |
+
},
|
2838 |
+
{
|
2839 |
+
"epoch": 25.63,
|
2840 |
+
"learning_rate": 6.042045775642837e-05,
|
2841 |
+
"loss": 0.0211,
|
2842 |
+
"step": 60900
|
2843 |
+
},
|
2844 |
+
{
|
2845 |
+
"epoch": 25.63,
|
2846 |
+
"eval_loss": 0.26482462882995605,
|
2847 |
+
"eval_runtime": 439.7256,
|
2848 |
+
"eval_samples_per_second": 11.855,
|
2849 |
+
"eval_wer": 0.10217577344985866,
|
2850 |
+
"step": 60900
|
2851 |
+
},
|
2852 |
+
{
|
2853 |
+
"epoch": 25.76,
|
2854 |
+
"learning_rate": 5.867420175190732e-05,
|
2855 |
+
"loss": 0.0216,
|
2856 |
+
"step": 61200
|
2857 |
+
},
|
2858 |
+
{
|
2859 |
+
"epoch": 25.76,
|
2860 |
+
"eval_loss": 0.2719128131866455,
|
2861 |
+
"eval_runtime": 444.9118,
|
2862 |
+
"eval_samples_per_second": 11.717,
|
2863 |
+
"eval_wer": 0.10227950519955395,
|
2864 |
+
"step": 61200
|
2865 |
+
},
|
2866 |
+
{
|
2867 |
+
"epoch": 25.88,
|
2868 |
+
"learning_rate": 5.692794574738626e-05,
|
2869 |
+
"loss": 0.0209,
|
2870 |
+
"step": 61500
|
2871 |
+
},
|
2872 |
+
{
|
2873 |
+
"epoch": 25.88,
|
2874 |
+
"eval_loss": 0.2721671164035797,
|
2875 |
+
"eval_runtime": 422.0845,
|
2876 |
+
"eval_samples_per_second": 12.351,
|
2877 |
+
"eval_wer": 0.10134591945229636,
|
2878 |
+
"step": 61500
|
2879 |
+
},
|
2880 |
+
{
|
2881 |
+
"epoch": 26.01,
|
2882 |
+
"learning_rate": 5.518168974286522e-05,
|
2883 |
+
"loss": 0.0212,
|
2884 |
+
"step": 61800
|
2885 |
+
},
|
2886 |
+
{
|
2887 |
+
"epoch": 26.01,
|
2888 |
+
"eval_loss": 0.26508814096450806,
|
2889 |
+
"eval_runtime": 445.5664,
|
2890 |
+
"eval_samples_per_second": 11.7,
|
2891 |
+
"eval_wer": 0.10051606545473406,
|
2892 |
+
"step": 61800
|
2893 |
+
},
|
2894 |
+
{
|
2895 |
+
"epoch": 26.14,
|
2896 |
+
"learning_rate": 5.3435433738344156e-05,
|
2897 |
+
"loss": 0.0204,
|
2898 |
+
"step": 62100
|
2899 |
+
},
|
2900 |
+
{
|
2901 |
+
"epoch": 26.14,
|
2902 |
+
"eval_loss": 0.26492705941200256,
|
2903 |
+
"eval_runtime": 452.3889,
|
2904 |
+
"eval_samples_per_second": 11.523,
|
2905 |
+
"eval_wer": 0.10220170638728249,
|
2906 |
+
"step": 62100
|
2907 |
+
},
|
2908 |
+
{
|
2909 |
+
"epoch": 26.26,
|
2910 |
+
"learning_rate": 5.168917773382311e-05,
|
2911 |
+
"loss": 0.0202,
|
2912 |
+
"step": 62400
|
2913 |
+
},
|
2914 |
+
{
|
2915 |
+
"epoch": 26.26,
|
2916 |
+
"eval_loss": 0.278484046459198,
|
2917 |
+
"eval_runtime": 422.0533,
|
2918 |
+
"eval_samples_per_second": 12.352,
|
2919 |
+
"eval_wer": 0.10313529213454008,
|
2920 |
+
"step": 62400
|
2921 |
+
},
|
2922 |
+
{
|
2923 |
+
"epoch": 26.39,
|
2924 |
+
"learning_rate": 4.994292172930206e-05,
|
2925 |
+
"loss": 0.0213,
|
2926 |
+
"step": 62700
|
2927 |
+
},
|
2928 |
+
{
|
2929 |
+
"epoch": 26.39,
|
2930 |
+
"eval_loss": 0.272367924451828,
|
2931 |
+
"eval_runtime": 442.0178,
|
2932 |
+
"eval_samples_per_second": 11.794,
|
2933 |
+
"eval_wer": 0.10033453489276731,
|
2934 |
+
"step": 62700
|
2935 |
+
},
|
2936 |
+
{
|
2937 |
+
"epoch": 26.52,
|
2938 |
+
"learning_rate": 4.819666572478101e-05,
|
2939 |
+
"loss": 0.0215,
|
2940 |
+
"step": 63000
|
2941 |
+
},
|
2942 |
+
{
|
2943 |
+
"epoch": 26.52,
|
2944 |
+
"eval_loss": 0.2681904137134552,
|
2945 |
+
"eval_runtime": 443.7076,
|
2946 |
+
"eval_samples_per_second": 11.749,
|
2947 |
+
"eval_wer": 0.10157931588911076,
|
2948 |
+
"step": 63000
|
2949 |
+
},
|
2950 |
+
{
|
2951 |
+
"epoch": 26.64,
|
2952 |
+
"learning_rate": 4.645040972025996e-05,
|
2953 |
+
"loss": 0.0188,
|
2954 |
+
"step": 63300
|
2955 |
+
},
|
2956 |
+
{
|
2957 |
+
"epoch": 26.64,
|
2958 |
+
"eval_loss": 0.2568943500518799,
|
2959 |
+
"eval_runtime": 423.3689,
|
2960 |
+
"eval_samples_per_second": 12.313,
|
2961 |
+
"eval_wer": 0.10023080314307202,
|
2962 |
+
"step": 63300
|
2963 |
+
},
|
2964 |
+
{
|
2965 |
+
"epoch": 26.77,
|
2966 |
+
"learning_rate": 4.4704153715738906e-05,
|
2967 |
+
"loss": 0.0192,
|
2968 |
+
"step": 63600
|
2969 |
+
},
|
2970 |
+
{
|
2971 |
+
"epoch": 26.77,
|
2972 |
+
"eval_loss": 0.26485371589660645,
|
2973 |
+
"eval_runtime": 442.7329,
|
2974 |
+
"eval_samples_per_second": 11.775,
|
2975 |
+
"eval_wer": 0.10111252301548196,
|
2976 |
+
"step": 63600
|
2977 |
+
},
|
2978 |
+
{
|
2979 |
+
"epoch": 26.89,
|
2980 |
+
"learning_rate": 4.295789771121786e-05,
|
2981 |
+
"loss": 0.0209,
|
2982 |
+
"step": 63900
|
2983 |
+
},
|
2984 |
+
{
|
2985 |
+
"epoch": 26.89,
|
2986 |
+
"eval_loss": 0.25886571407318115,
|
2987 |
+
"eval_runtime": 451.6321,
|
2988 |
+
"eval_samples_per_second": 11.543,
|
2989 |
+
"eval_wer": 0.099504680895205,
|
2990 |
+
"step": 63900
|
2991 |
+
},
|
2992 |
+
{
|
2993 |
+
"epoch": 27.02,
|
2994 |
+
"learning_rate": 4.121164170669681e-05,
|
2995 |
+
"loss": 0.0194,
|
2996 |
+
"step": 64200
|
2997 |
+
},
|
2998 |
+
{
|
2999 |
+
"epoch": 27.02,
|
3000 |
+
"eval_loss": 0.2557186484336853,
|
3001 |
+
"eval_runtime": 442.5519,
|
3002 |
+
"eval_samples_per_second": 11.779,
|
3003 |
+
"eval_wer": 0.09893415627188092,
|
3004 |
+
"step": 64200
|
3005 |
+
},
|
3006 |
+
{
|
3007 |
+
"epoch": 27.15,
|
3008 |
+
"learning_rate": 3.946538570217575e-05,
|
3009 |
+
"loss": 0.0196,
|
3010 |
+
"step": 64500
|
3011 |
+
},
|
3012 |
+
{
|
3013 |
+
"epoch": 27.15,
|
3014 |
+
"eval_loss": 0.25841090083122253,
|
3015 |
+
"eval_runtime": 424.1061,
|
3016 |
+
"eval_samples_per_second": 12.292,
|
3017 |
+
"eval_wer": 0.09989367495656233,
|
3018 |
+
"step": 64500
|
3019 |
+
},
|
3020 |
+
{
|
3021 |
+
"epoch": 27.27,
|
3022 |
+
"learning_rate": 3.77191296976547e-05,
|
3023 |
+
"loss": 0.019,
|
3024 |
+
"step": 64800
|
3025 |
+
},
|
3026 |
+
{
|
3027 |
+
"epoch": 27.27,
|
3028 |
+
"eval_loss": 0.2668149173259735,
|
3029 |
+
"eval_runtime": 448.6143,
|
3030 |
+
"eval_samples_per_second": 11.62,
|
3031 |
+
"eval_wer": 0.09877855864733798,
|
3032 |
+
"step": 64800
|
3033 |
+
},
|
3034 |
+
{
|
3035 |
+
"epoch": 27.4,
|
3036 |
+
"learning_rate": 3.5972873693133656e-05,
|
3037 |
+
"loss": 0.0181,
|
3038 |
+
"step": 65100
|
3039 |
+
},
|
3040 |
+
{
|
3041 |
+
"epoch": 27.4,
|
3042 |
+
"eval_loss": 0.2628437876701355,
|
3043 |
+
"eval_runtime": 450.825,
|
3044 |
+
"eval_samples_per_second": 11.563,
|
3045 |
+
"eval_wer": 0.09768937527553746,
|
3046 |
+
"step": 65100
|
3047 |
+
},
|
3048 |
+
{
|
3049 |
+
"epoch": 27.53,
|
3050 |
+
"learning_rate": 3.42266176886126e-05,
|
3051 |
+
"loss": 0.0175,
|
3052 |
+
"step": 65400
|
3053 |
+
},
|
3054 |
+
{
|
3055 |
+
"epoch": 27.53,
|
3056 |
+
"eval_loss": 0.26442092657089233,
|
3057 |
+
"eval_runtime": 420.8378,
|
3058 |
+
"eval_samples_per_second": 12.387,
|
3059 |
+
"eval_wer": 0.0998677420191385,
|
3060 |
+
"step": 65400
|
3061 |
+
},
|
3062 |
+
{
|
3063 |
+
"epoch": 27.65,
|
3064 |
+
"learning_rate": 3.248036168409155e-05,
|
3065 |
+
"loss": 0.0201,
|
3066 |
+
"step": 65700
|
3067 |
+
},
|
3068 |
+
{
|
3069 |
+
"epoch": 27.65,
|
3070 |
+
"eval_loss": 0.2726499140262604,
|
3071 |
+
"eval_runtime": 445.7776,
|
3072 |
+
"eval_samples_per_second": 11.694,
|
3073 |
+
"eval_wer": 0.09924535152096678,
|
3074 |
+
"step": 65700
|
3075 |
+
},
|
3076 |
+
{
|
3077 |
+
"epoch": 27.78,
|
3078 |
+
"learning_rate": 3.07341056795705e-05,
|
3079 |
+
"loss": 0.0182,
|
3080 |
+
"step": 66000
|
3081 |
+
},
|
3082 |
+
{
|
3083 |
+
"epoch": 27.78,
|
3084 |
+
"eval_loss": 0.265592485666275,
|
3085 |
+
"eval_runtime": 453.8658,
|
3086 |
+
"eval_samples_per_second": 11.486,
|
3087 |
+
"eval_wer": 0.09704105183994191,
|
3088 |
+
"step": 66000
|
3089 |
+
},
|
3090 |
+
{
|
3091 |
+
"epoch": 27.9,
|
3092 |
+
"learning_rate": 2.8987849675049448e-05,
|
3093 |
+
"loss": 0.0205,
|
3094 |
+
"step": 66300
|
3095 |
+
},
|
3096 |
+
{
|
3097 |
+
"epoch": 27.9,
|
3098 |
+
"eval_loss": 0.2697185277938843,
|
3099 |
+
"eval_runtime": 421.7967,
|
3100 |
+
"eval_samples_per_second": 12.359,
|
3101 |
+
"eval_wer": 0.09831176577370919,
|
3102 |
+
"step": 66300
|
3103 |
+
},
|
3104 |
+
{
|
3105 |
+
"epoch": 28.03,
|
3106 |
+
"learning_rate": 2.7241593670528396e-05,
|
3107 |
+
"loss": 0.0197,
|
3108 |
+
"step": 66600
|
3109 |
+
},
|
3110 |
+
{
|
3111 |
+
"epoch": 28.03,
|
3112 |
+
"eval_loss": 0.26293227076530457,
|
3113 |
+
"eval_runtime": 454.9345,
|
3114 |
+
"eval_samples_per_second": 11.459,
|
3115 |
+
"eval_wer": 0.09675578952827987,
|
3116 |
+
"step": 66600
|
3117 |
+
},
|
3118 |
+
{
|
3119 |
+
"epoch": 28.16,
|
3120 |
+
"learning_rate": 2.5495337666007347e-05,
|
3121 |
+
"loss": 0.018,
|
3122 |
+
"step": 66900
|
3123 |
+
},
|
3124 |
+
{
|
3125 |
+
"epoch": 28.16,
|
3126 |
+
"eval_loss": 0.2722020149230957,
|
3127 |
+
"eval_runtime": 451.9089,
|
3128 |
+
"eval_samples_per_second": 11.536,
|
3129 |
+
"eval_wer": 0.09683358834055134,
|
3130 |
+
"step": 66900
|
3131 |
+
},
|
3132 |
+
{
|
3133 |
+
"epoch": 28.28,
|
3134 |
+
"learning_rate": 2.3749081661486295e-05,
|
3135 |
+
"loss": 0.0172,
|
3136 |
+
"step": 67200
|
3137 |
+
},
|
3138 |
+
{
|
3139 |
+
"epoch": 28.28,
|
3140 |
+
"eval_loss": 0.2774170935153961,
|
3141 |
+
"eval_runtime": 451.1229,
|
3142 |
+
"eval_samples_per_second": 11.556,
|
3143 |
+
"eval_wer": 0.09685952127797516,
|
3144 |
+
"step": 67200
|
3145 |
+
},
|
3146 |
+
{
|
3147 |
+
"epoch": 28.41,
|
3148 |
+
"learning_rate": 2.2002825656965243e-05,
|
3149 |
+
"loss": 0.0158,
|
3150 |
+
"step": 67500
|
3151 |
+
},
|
3152 |
+
{
|
3153 |
+
"epoch": 28.41,
|
3154 |
+
"eval_loss": 0.27579203248023987,
|
3155 |
+
"eval_runtime": 425.5661,
|
3156 |
+
"eval_samples_per_second": 12.25,
|
3157 |
+
"eval_wer": 0.09735224708902777,
|
3158 |
+
"step": 67500
|
3159 |
+
},
|
3160 |
+
{
|
3161 |
+
"epoch": 28.54,
|
3162 |
+
"learning_rate": 2.025656965244419e-05,
|
3163 |
+
"loss": 0.0189,
|
3164 |
+
"step": 67800
|
3165 |
+
},
|
3166 |
+
{
|
3167 |
+
"epoch": 28.54,
|
3168 |
+
"eval_loss": 0.2757831811904907,
|
3169 |
+
"eval_runtime": 453.6044,
|
3170 |
+
"eval_samples_per_second": 11.492,
|
3171 |
+
"eval_wer": 0.09717071652706102,
|
3172 |
+
"step": 67800
|
3173 |
+
},
|
3174 |
+
{
|
3175 |
+
"epoch": 28.66,
|
3176 |
+
"learning_rate": 1.8510313647923142e-05,
|
3177 |
+
"loss": 0.0188,
|
3178 |
+
"step": 68100
|
3179 |
+
},
|
3180 |
+
{
|
3181 |
+
"epoch": 28.66,
|
3182 |
+
"eval_loss": 0.281482994556427,
|
3183 |
+
"eval_runtime": 443.0118,
|
3184 |
+
"eval_samples_per_second": 11.767,
|
3185 |
+
"eval_wer": 0.09660019190373693,
|
3186 |
+
"step": 68100
|
3187 |
+
},
|
3188 |
+
{
|
3189 |
+
"epoch": 28.79,
|
3190 |
+
"learning_rate": 1.676405764340209e-05,
|
3191 |
+
"loss": 0.0167,
|
3192 |
+
"step": 68400
|
3193 |
+
},
|
3194 |
+
{
|
3195 |
+
"epoch": 28.79,
|
3196 |
+
"eval_loss": 0.2745157480239868,
|
3197 |
+
"eval_runtime": 422.5171,
|
3198 |
+
"eval_samples_per_second": 12.338,
|
3199 |
+
"eval_wer": 0.09587406965586992,
|
3200 |
+
"step": 68400
|
3201 |
+
},
|
3202 |
+
{
|
3203 |
+
"epoch": 28.91,
|
3204 |
+
"learning_rate": 1.501780163888104e-05,
|
3205 |
+
"loss": 0.0149,
|
3206 |
+
"step": 68700
|
3207 |
+
},
|
3208 |
+
{
|
3209 |
+
"epoch": 28.91,
|
3210 |
+
"eval_loss": 0.27913355827331543,
|
3211 |
+
"eval_runtime": 451.1316,
|
3212 |
+
"eval_samples_per_second": 11.555,
|
3213 |
+
"eval_wer": 0.09660019190373693,
|
3214 |
+
"step": 68700
|
3215 |
+
},
|
3216 |
+
{
|
3217 |
+
"epoch": 29.04,
|
3218 |
+
"learning_rate": 1.3271545634359989e-05,
|
3219 |
+
"loss": 0.0168,
|
3220 |
+
"step": 69000
|
3221 |
+
},
|
3222 |
+
{
|
3223 |
+
"epoch": 29.04,
|
3224 |
+
"eval_loss": 0.27442270517349243,
|
3225 |
+
"eval_runtime": 453.3058,
|
3226 |
+
"eval_samples_per_second": 11.5,
|
3227 |
+
"eval_wer": 0.0956147402816317,
|
3228 |
+
"step": 69000
|
3229 |
+
},
|
3230 |
+
{
|
3231 |
+
"epoch": 29.17,
|
3232 |
+
"learning_rate": 1.1525289629838937e-05,
|
3233 |
+
"loss": 0.0163,
|
3234 |
+
"step": 69300
|
3235 |
+
},
|
3236 |
+
{
|
3237 |
+
"epoch": 29.17,
|
3238 |
+
"eval_loss": 0.27673137187957764,
|
3239 |
+
"eval_runtime": 423.4053,
|
3240 |
+
"eval_samples_per_second": 12.312,
|
3241 |
+
"eval_wer": 0.09512201447057908,
|
3242 |
+
"step": 69300
|
3243 |
+
},
|
3244 |
+
{
|
3245 |
+
"epoch": 29.29,
|
3246 |
+
"learning_rate": 9.779033625317885e-06,
|
3247 |
+
"loss": 0.0177,
|
3248 |
+
"step": 69600
|
3249 |
+
},
|
3250 |
+
{
|
3251 |
+
"epoch": 29.29,
|
3252 |
+
"eval_loss": 0.2782154083251953,
|
3253 |
+
"eval_runtime": 450.2277,
|
3254 |
+
"eval_samples_per_second": 11.579,
|
3255 |
+
"eval_wer": 0.09512201447057908,
|
3256 |
+
"step": 69600
|
3257 |
+
},
|
3258 |
+
{
|
3259 |
+
"epoch": 29.42,
|
3260 |
+
"learning_rate": 8.032777620796835e-06,
|
3261 |
+
"loss": 0.0166,
|
3262 |
+
"step": 69900
|
3263 |
+
},
|
3264 |
+
{
|
3265 |
+
"epoch": 29.42,
|
3266 |
+
"eval_loss": 0.2790413200855255,
|
3267 |
+
"eval_runtime": 446.9628,
|
3268 |
+
"eval_samples_per_second": 11.663,
|
3269 |
+
"eval_wer": 0.09553694146936023,
|
3270 |
+
"step": 69900
|
3271 |
+
},
|
3272 |
+
{
|
3273 |
+
"epoch": 29.55,
|
3274 |
+
"learning_rate": 6.286521616275784e-06,
|
3275 |
+
"loss": 0.0152,
|
3276 |
+
"step": 70200
|
3277 |
+
},
|
3278 |
+
{
|
3279 |
+
"epoch": 29.55,
|
3280 |
+
"eval_loss": 0.2786215543746948,
|
3281 |
+
"eval_runtime": 447.9811,
|
3282 |
+
"eval_samples_per_second": 11.637,
|
3283 |
+
"eval_wer": 0.0953813438448173,
|
3284 |
+
"step": 70200
|
3285 |
+
},
|
3286 |
+
{
|
3287 |
+
"epoch": 29.67,
|
3288 |
+
"learning_rate": 4.540265611754733e-06,
|
3289 |
+
"loss": 0.0157,
|
3290 |
+
"step": 70500
|
3291 |
+
},
|
3292 |
+
{
|
3293 |
+
"epoch": 29.67,
|
3294 |
+
"eval_loss": 0.27765703201293945,
|
3295 |
+
"eval_runtime": 423.4541,
|
3296 |
+
"eval_samples_per_second": 12.311,
|
3297 |
+
"eval_wer": 0.09517388034542673,
|
3298 |
+
"step": 70500
|
3299 |
+
},
|
3300 |
+
{
|
3301 |
+
"epoch": 29.8,
|
3302 |
+
"learning_rate": 2.7940096072336816e-06,
|
3303 |
+
"loss": 0.0175,
|
3304 |
+
"step": 70800
|
3305 |
+
},
|
3306 |
+
{
|
3307 |
+
"epoch": 29.8,
|
3308 |
+
"eval_loss": 0.2788052558898926,
|
3309 |
+
"eval_runtime": 450.7945,
|
3310 |
+
"eval_samples_per_second": 11.564,
|
3311 |
+
"eval_wer": 0.09486268509634087,
|
3312 |
+
"step": 70800
|
3313 |
+
},
|
3314 |
+
{
|
3315 |
+
"epoch": 29.92,
|
3316 |
+
"learning_rate": 1.0477536027126306e-06,
|
3317 |
+
"loss": 0.0166,
|
3318 |
+
"step": 71100
|
3319 |
+
},
|
3320 |
+
{
|
3321 |
+
"epoch": 29.92,
|
3322 |
+
"eval_loss": 0.2789298892021179,
|
3323 |
+
"eval_runtime": 455.4523,
|
3324 |
+
"eval_samples_per_second": 11.446,
|
3325 |
+
"eval_wer": 0.09540727678224112,
|
3326 |
+
"step": 71100
|
3327 |
+
},
|
3328 |
+
{
|
3329 |
+
"epoch": 30.0,
|
3330 |
+
"step": 71280,
|
3331 |
+
"total_flos": 4.535964846237055e+20,
|
3332 |
+
"train_runtime": 470143.8106,
|
3333 |
+
"train_samples_per_second": 0.152
|
3334 |
+
}
|
3335 |
+
],
|
3336 |
+
"max_steps": 71280,
|
3337 |
+
"num_train_epochs": 30,
|
3338 |
+
"total_flos": 4.535964846237055e+20,
|
3339 |
+
"trial_name": null,
|
3340 |
+
"trial_params": null
|
3341 |
+
}
|
vocab.json
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
{"<pad>": 0, "<s>": 1, "</s>": 2, "<unk>": 3, "|": 4, "آ": 5, "ئ": 6, "ا": 7, "ب": 8, "ت": 9, "ث": 10, "ج": 11, "ح": 12, "خ": 13, "د": 14, "ذ": 15, "ر": 16, "ز": 17, "س": 18, "ش": 19, "ص": 20, "ض": 21, "ط": 22, "ظ": 23, "ع": 24, "غ": 25, "ف": 26, "ق": 27, "ل": 28, "م": 29, "ن": 30, "ه": 31, "و": 32, "پ": 33, "چ": 34, "ژ": 35, "ک": 36, "گ": 37, "ی": 38, "": 39}
|