NLP-EXP commited on
Commit
e256b74
·
verified ·
1 Parent(s): 81f899d

Upload 14 files

Browse files
README.md ADDED
@@ -0,0 +1,206 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: facebook/wav2vec2-large-xlsr-53
3
+ library_name: peft
4
+ tags:
5
+ - base_model:adapter:facebook/wav2vec2-large-xlsr-53
6
+ - lora
7
+ - transformers
8
+ ---
9
+
10
+ # Model Card for Model ID
11
+
12
+ <!-- Provide a quick summary of what the model is/does. -->
13
+
14
+
15
+
16
+ ## Model Details
17
+
18
+ ### Model Description
19
+
20
+ <!-- Provide a longer summary of what this model is. -->
21
+
22
+
23
+
24
+ - **Developed by:** [More Information Needed]
25
+ - **Funded by [optional]:** [More Information Needed]
26
+ - **Shared by [optional]:** [More Information Needed]
27
+ - **Model type:** [More Information Needed]
28
+ - **Language(s) (NLP):** [More Information Needed]
29
+ - **License:** [More Information Needed]
30
+ - **Finetuned from model [optional]:** [More Information Needed]
31
+
32
+ ### Model Sources [optional]
33
+
34
+ <!-- Provide the basic links for the model. -->
35
+
36
+ - **Repository:** [More Information Needed]
37
+ - **Paper [optional]:** [More Information Needed]
38
+ - **Demo [optional]:** [More Information Needed]
39
+
40
+ ## Uses
41
+
42
+ <!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
43
+
44
+ ### Direct Use
45
+
46
+ <!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
47
+
48
+ [More Information Needed]
49
+
50
+ ### Downstream Use [optional]
51
+
52
+ <!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
53
+
54
+ [More Information Needed]
55
+
56
+ ### Out-of-Scope Use
57
+
58
+ <!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
59
+
60
+ [More Information Needed]
61
+
62
+ ## Bias, Risks, and Limitations
63
+
64
+ <!-- This section is meant to convey both technical and sociotechnical limitations. -->
65
+
66
+ [More Information Needed]
67
+
68
+ ### Recommendations
69
+
70
+ <!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
71
+
72
+ Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
73
+
74
+ ## How to Get Started with the Model
75
+
76
+ Use the code below to get started with the model.
77
+
78
+ [More Information Needed]
79
+
80
+ ## Training Details
81
+
82
+ ### Training Data
83
+
84
+ <!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
85
+
86
+ [More Information Needed]
87
+
88
+ ### Training Procedure
89
+
90
+ <!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
91
+
92
+ #### Preprocessing [optional]
93
+
94
+ [More Information Needed]
95
+
96
+
97
+ #### Training Hyperparameters
98
+
99
+ - **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
100
+
101
+ #### Speeds, Sizes, Times [optional]
102
+
103
+ <!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
104
+
105
+ [More Information Needed]
106
+
107
+ ## Evaluation
108
+
109
+ <!-- This section describes the evaluation protocols and provides the results. -->
110
+
111
+ ### Testing Data, Factors & Metrics
112
+
113
+ #### Testing Data
114
+
115
+ <!-- This should link to a Dataset Card if possible. -->
116
+
117
+ [More Information Needed]
118
+
119
+ #### Factors
120
+
121
+ <!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
122
+
123
+ [More Information Needed]
124
+
125
+ #### Metrics
126
+
127
+ <!-- These are the evaluation metrics being used, ideally with a description of why. -->
128
+
129
+ [More Information Needed]
130
+
131
+ ### Results
132
+
133
+ [More Information Needed]
134
+
135
+ #### Summary
136
+
137
+
138
+
139
+ ## Model Examination [optional]
140
+
141
+ <!-- Relevant interpretability work for the model goes here -->
142
+
143
+ [More Information Needed]
144
+
145
+ ## Environmental Impact
146
+
147
+ <!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
148
+
149
+ Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
150
+
151
+ - **Hardware Type:** [More Information Needed]
152
+ - **Hours used:** [More Information Needed]
153
+ - **Cloud Provider:** [More Information Needed]
154
+ - **Compute Region:** [More Information Needed]
155
+ - **Carbon Emitted:** [More Information Needed]
156
+
157
+ ## Technical Specifications [optional]
158
+
159
+ ### Model Architecture and Objective
160
+
161
+ [More Information Needed]
162
+
163
+ ### Compute Infrastructure
164
+
165
+ [More Information Needed]
166
+
167
+ #### Hardware
168
+
169
+ [More Information Needed]
170
+
171
+ #### Software
172
+
173
+ [More Information Needed]
174
+
175
+ ## Citation [optional]
176
+
177
+ <!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
178
+
179
+ **BibTeX:**
180
+
181
+ [More Information Needed]
182
+
183
+ **APA:**
184
+
185
+ [More Information Needed]
186
+
187
+ ## Glossary [optional]
188
+
189
+ <!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
190
+
191
+ [More Information Needed]
192
+
193
+ ## More Information [optional]
194
+
195
+ [More Information Needed]
196
+
197
+ ## Model Card Authors [optional]
198
+
199
+ [More Information Needed]
200
+
201
+ ## Model Card Contact
202
+
203
+ [More Information Needed]
204
+ ### Framework versions
205
+
206
+ - PEFT 0.18.0
adapter_config.json ADDED
@@ -0,0 +1,48 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alora_invocation_tokens": null,
3
+ "alpha_pattern": {},
4
+ "arrow_config": null,
5
+ "auto_mapping": {
6
+ "base_model_class": "Wav2Vec2ForCTC",
7
+ "parent_library": "transformers.models.wav2vec2.modeling_wav2vec2"
8
+ },
9
+ "base_model_name_or_path": "facebook/wav2vec2-large-xlsr-53",
10
+ "bias": "none",
11
+ "corda_config": null,
12
+ "ensure_weight_tying": false,
13
+ "eva_config": null,
14
+ "exclude_modules": null,
15
+ "fan_in_fan_out": false,
16
+ "inference_mode": true,
17
+ "init_lora_weights": true,
18
+ "layer_replication": null,
19
+ "layers_pattern": null,
20
+ "layers_to_transform": null,
21
+ "loftq_config": {},
22
+ "lora_alpha": 32,
23
+ "lora_bias": false,
24
+ "lora_dropout": 0.05,
25
+ "megatron_config": null,
26
+ "megatron_core": "megatron.core",
27
+ "modules_to_save": null,
28
+ "peft_type": "LORA",
29
+ "peft_version": "0.18.0",
30
+ "qalora_group_size": 16,
31
+ "r": 16,
32
+ "rank_pattern": {},
33
+ "revision": null,
34
+ "target_modules": [
35
+ "output_dense",
36
+ "q_proj",
37
+ "k_proj",
38
+ "intermediate_dense",
39
+ "v_proj",
40
+ "out_proj"
41
+ ],
42
+ "target_parameters": null,
43
+ "task_type": null,
44
+ "trainable_token_indices": null,
45
+ "use_dora": false,
46
+ "use_qalora": false,
47
+ "use_rslora": false
48
+ }
adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:934cdb597fe0224ef64a50e56d4cb0ca0a9d3c67e2e6ff7fcd7f4fc348c9cf95
3
+ size 28354536
added_tokens.json ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ {
2
+ "</s>": 38,
3
+ "<s>": 37
4
+ }
optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:74eb7e1746cc10a56cbbe8796a4eaf5fa4bdd0cf98bb2093fa0fd47dc9e5c6bd
3
+ size 57190859
preprocessor_config.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_normalize": true,
3
+ "feature_extractor_type": "Wav2Vec2FeatureExtractor",
4
+ "feature_size": 1,
5
+ "padding_side": "right",
6
+ "padding_value": 0.0,
7
+ "processor_class": "Wav2Vec2Processor",
8
+ "return_attention_mask": true,
9
+ "sampling_rate": 16000
10
+ }
rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:11c6c90f1a0d57ffd763356f74e846cfaaa3a9025af757f392c0a71b3f187675
3
+ size 14709
scaler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:331020dd183bf5bd4b1430d9231cca331282992c44d9e5a2963d60aaddab40e2
3
+ size 1383
scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7c1cbf05d1a5d420ca1c71264cc117551123463425955c5350944bb1a56068cc
3
+ size 1465
special_tokens_map.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": "<s>",
3
+ "eos_token": "</s>",
4
+ "pad_token": "[PAD]",
5
+ "unk_token": "[UNK]"
6
+ }
tokenizer_config.json ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "added_tokens_decoder": {
3
+ "0": {
4
+ "content": "[PAD]",
5
+ "lstrip": true,
6
+ "normalized": false,
7
+ "rstrip": true,
8
+ "single_word": false,
9
+ "special": false
10
+ },
11
+ "1": {
12
+ "content": "[UNK]",
13
+ "lstrip": true,
14
+ "normalized": false,
15
+ "rstrip": true,
16
+ "single_word": false,
17
+ "special": false
18
+ },
19
+ "37": {
20
+ "content": "<s>",
21
+ "lstrip": false,
22
+ "normalized": false,
23
+ "rstrip": false,
24
+ "single_word": false,
25
+ "special": true
26
+ },
27
+ "38": {
28
+ "content": "</s>",
29
+ "lstrip": false,
30
+ "normalized": false,
31
+ "rstrip": false,
32
+ "single_word": false,
33
+ "special": true
34
+ }
35
+ },
36
+ "bos_token": "<s>",
37
+ "clean_up_tokenization_spaces": false,
38
+ "do_lower_case": false,
39
+ "eos_token": "</s>",
40
+ "extra_special_tokens": {},
41
+ "model_max_length": 1000000000000000019884624838656,
42
+ "pad_token": "[PAD]",
43
+ "processor_class": "Wav2Vec2Processor",
44
+ "replace_word_delimiter_char": " ",
45
+ "target_lang": null,
46
+ "tokenizer_class": "Wav2Vec2CTCTokenizer",
47
+ "unk_token": "[UNK]",
48
+ "word_delimiter_token": "|"
49
+ }
trainer_state.json ADDED
@@ -0,0 +1,3262 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": 18532,
3
+ "best_metric": 0.19350049787272564,
4
+ "best_model_checkpoint": "/content/drive/MyDrive/asr_loracheckpoint2_backup/checkpoint-18532",
5
+ "epoch": 44.0,
6
+ "eval_steps": 500,
7
+ "global_step": 19888,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "epoch": 0.11074197120708748,
14
+ "grad_norm": 25.114500045776367,
15
+ "learning_rate": 6.5044247787610615e-06,
16
+ "loss": 24.7234,
17
+ "step": 50
18
+ },
19
+ {
20
+ "epoch": 0.22148394241417496,
21
+ "grad_norm": 18.712068557739258,
22
+ "learning_rate": 1.3141592920353982e-05,
23
+ "loss": 24.4168,
24
+ "step": 100
25
+ },
26
+ {
27
+ "epoch": 0.33222591362126247,
28
+ "grad_norm": 27.625778198242188,
29
+ "learning_rate": 1.9778761061946902e-05,
30
+ "loss": 24.0671,
31
+ "step": 150
32
+ },
33
+ {
34
+ "epoch": 0.4429678848283499,
35
+ "grad_norm": 21.45062828063965,
36
+ "learning_rate": 2.641592920353982e-05,
37
+ "loss": 23.263,
38
+ "step": 200
39
+ },
40
+ {
41
+ "epoch": 0.5537098560354374,
42
+ "grad_norm": 38.891578674316406,
43
+ "learning_rate": 3.305309734513274e-05,
44
+ "loss": 21.274,
45
+ "step": 250
46
+ },
47
+ {
48
+ "epoch": 0.6644518272425249,
49
+ "grad_norm": 15.521417617797852,
50
+ "learning_rate": 3.9690265486725656e-05,
51
+ "loss": 9.7142,
52
+ "step": 300
53
+ },
54
+ {
55
+ "epoch": 0.7751937984496124,
56
+ "grad_norm": 2.311408758163452,
57
+ "learning_rate": 4.632743362831858e-05,
58
+ "loss": 3.6891,
59
+ "step": 350
60
+ },
61
+ {
62
+ "epoch": 0.8859357696566998,
63
+ "grad_norm": 0.999423623085022,
64
+ "learning_rate": 5.29646017699115e-05,
65
+ "loss": 3.2125,
66
+ "step": 400
67
+ },
68
+ {
69
+ "epoch": 0.9966777408637874,
70
+ "grad_norm": 0.38470658659935,
71
+ "learning_rate": 5.960176991150442e-05,
72
+ "loss": 3.1497,
73
+ "step": 450
74
+ },
75
+ {
76
+ "epoch": 1.0,
77
+ "eval_cer": 1.0,
78
+ "eval_loss": 3.15576434135437,
79
+ "eval_runtime": 483.1807,
80
+ "eval_samples_per_second": 21.071,
81
+ "eval_steps_per_second": 2.635,
82
+ "eval_wer": 1.0,
83
+ "step": 452
84
+ },
85
+ {
86
+ "epoch": 1.106312292358804,
87
+ "grad_norm": 0.8518805503845215,
88
+ "learning_rate": 6.623893805309734e-05,
89
+ "loss": 3.1317,
90
+ "step": 500
91
+ },
92
+ {
93
+ "epoch": 1.2170542635658914,
94
+ "grad_norm": 0.38256514072418213,
95
+ "learning_rate": 7.287610619469026e-05,
96
+ "loss": 3.1353,
97
+ "step": 550
98
+ },
99
+ {
100
+ "epoch": 1.327796234772979,
101
+ "grad_norm": 1.8106805086135864,
102
+ "learning_rate": 7.951327433628318e-05,
103
+ "loss": 3.1262,
104
+ "step": 600
105
+ },
106
+ {
107
+ "epoch": 1.4385382059800664,
108
+ "grad_norm": 0.5565470457077026,
109
+ "learning_rate": 8.61504424778761e-05,
110
+ "loss": 3.1362,
111
+ "step": 650
112
+ },
113
+ {
114
+ "epoch": 1.5492801771871538,
115
+ "grad_norm": 0.8854584097862244,
116
+ "learning_rate": 9.278761061946901e-05,
117
+ "loss": 3.1189,
118
+ "step": 700
119
+ },
120
+ {
121
+ "epoch": 1.6600221483942414,
122
+ "grad_norm": 0.43876901268959045,
123
+ "learning_rate": 9.942477876106195e-05,
124
+ "loss": 3.1301,
125
+ "step": 750
126
+ },
127
+ {
128
+ "epoch": 1.770764119601329,
129
+ "grad_norm": 0.6520148515701294,
130
+ "learning_rate": 0.00010606194690265486,
131
+ "loss": 3.1249,
132
+ "step": 800
133
+ },
134
+ {
135
+ "epoch": 1.8815060908084162,
136
+ "grad_norm": 0.9588941931724548,
137
+ "learning_rate": 0.00011269911504424777,
138
+ "loss": 3.1179,
139
+ "step": 850
140
+ },
141
+ {
142
+ "epoch": 1.9922480620155039,
143
+ "grad_norm": 0.9581903219223022,
144
+ "learning_rate": 0.0001193362831858407,
145
+ "loss": 3.1061,
146
+ "step": 900
147
+ },
148
+ {
149
+ "epoch": 2.0,
150
+ "eval_cer": 1.0,
151
+ "eval_loss": 3.11116886138916,
152
+ "eval_runtime": 481.1502,
153
+ "eval_samples_per_second": 21.16,
154
+ "eval_steps_per_second": 2.646,
155
+ "eval_wer": 1.0,
156
+ "step": 904
157
+ },
158
+ {
159
+ "epoch": 2.1018826135105204,
160
+ "grad_norm": 1.0075995922088623,
161
+ "learning_rate": 0.0001259734513274336,
162
+ "loss": 3.0901,
163
+ "step": 950
164
+ },
165
+ {
166
+ "epoch": 2.212624584717608,
167
+ "grad_norm": 1.4122037887573242,
168
+ "learning_rate": 0.00013261061946902652,
169
+ "loss": 3.0824,
170
+ "step": 1000
171
+ },
172
+ {
173
+ "epoch": 2.3233665559246957,
174
+ "grad_norm": 0.37052252888679504,
175
+ "learning_rate": 0.00013924778761061947,
176
+ "loss": 3.0478,
177
+ "step": 1050
178
+ },
179
+ {
180
+ "epoch": 2.434108527131783,
181
+ "grad_norm": 0.9149488210678101,
182
+ "learning_rate": 0.00014588495575221239,
183
+ "loss": 3.0426,
184
+ "step": 1100
185
+ },
186
+ {
187
+ "epoch": 2.5448504983388704,
188
+ "grad_norm": 0.6695976853370667,
189
+ "learning_rate": 0.0001525221238938053,
190
+ "loss": 3.0352,
191
+ "step": 1150
192
+ },
193
+ {
194
+ "epoch": 2.655592469545958,
195
+ "grad_norm": 0.28660523891448975,
196
+ "learning_rate": 0.0001591592920353982,
197
+ "loss": 3.0372,
198
+ "step": 1200
199
+ },
200
+ {
201
+ "epoch": 2.7663344407530452,
202
+ "grad_norm": 2.475407838821411,
203
+ "learning_rate": 0.00016579646017699114,
204
+ "loss": 3.0092,
205
+ "step": 1250
206
+ },
207
+ {
208
+ "epoch": 2.877076411960133,
209
+ "grad_norm": 0.40837517380714417,
210
+ "learning_rate": 0.00017243362831858406,
211
+ "loss": 2.9405,
212
+ "step": 1300
213
+ },
214
+ {
215
+ "epoch": 2.9878183831672205,
216
+ "grad_norm": 0.7240709066390991,
217
+ "learning_rate": 0.000179070796460177,
218
+ "loss": 2.8017,
219
+ "step": 1350
220
+ },
221
+ {
222
+ "epoch": 3.0,
223
+ "eval_cer": 1.0,
224
+ "eval_loss": 2.5643131732940674,
225
+ "eval_runtime": 484.9725,
226
+ "eval_samples_per_second": 20.993,
227
+ "eval_steps_per_second": 2.625,
228
+ "eval_wer": 1.0,
229
+ "step": 1356
230
+ },
231
+ {
232
+ "epoch": 3.097452934662237,
233
+ "grad_norm": 1.4658317565917969,
234
+ "learning_rate": 0.0001857079646017699,
235
+ "loss": 2.5261,
236
+ "step": 1400
237
+ },
238
+ {
239
+ "epoch": 3.2081949058693247,
240
+ "grad_norm": 0.7111493945121765,
241
+ "learning_rate": 0.0001923451327433628,
242
+ "loss": 2.115,
243
+ "step": 1450
244
+ },
245
+ {
246
+ "epoch": 3.318936877076412,
247
+ "grad_norm": 1.02842378616333,
248
+ "learning_rate": 0.00019898230088495576,
249
+ "loss": 1.7595,
250
+ "step": 1500
251
+ },
252
+ {
253
+ "epoch": 3.4296788482834994,
254
+ "grad_norm": 1.2684359550476074,
255
+ "learning_rate": 0.00020561946902654865,
256
+ "loss": 1.4518,
257
+ "step": 1550
258
+ },
259
+ {
260
+ "epoch": 3.540420819490587,
261
+ "grad_norm": 1.0996291637420654,
262
+ "learning_rate": 0.00021225663716814157,
263
+ "loss": 1.3586,
264
+ "step": 1600
265
+ },
266
+ {
267
+ "epoch": 3.6511627906976747,
268
+ "grad_norm": 0.8183071613311768,
269
+ "learning_rate": 0.0002188938053097345,
270
+ "loss": 1.2308,
271
+ "step": 1650
272
+ },
273
+ {
274
+ "epoch": 3.761904761904762,
275
+ "grad_norm": 0.8036144971847534,
276
+ "learning_rate": 0.0002255309734513274,
277
+ "loss": 1.1334,
278
+ "step": 1700
279
+ },
280
+ {
281
+ "epoch": 3.8726467331118495,
282
+ "grad_norm": 1.1055711507797241,
283
+ "learning_rate": 0.00023216814159292032,
284
+ "loss": 1.0593,
285
+ "step": 1750
286
+ },
287
+ {
288
+ "epoch": 3.983388704318937,
289
+ "grad_norm": 0.8789339065551758,
290
+ "learning_rate": 0.00023880530973451327,
291
+ "loss": 1.0541,
292
+ "step": 1800
293
+ },
294
+ {
295
+ "epoch": 4.0,
296
+ "eval_cer": 0.1802003031379858,
297
+ "eval_loss": 0.645916223526001,
298
+ "eval_runtime": 487.2819,
299
+ "eval_samples_per_second": 20.893,
300
+ "eval_steps_per_second": 2.612,
301
+ "eval_wer": 0.6615370688874808,
302
+ "step": 1808
303
+ },
304
+ {
305
+ "epoch": 4.093023255813954,
306
+ "grad_norm": 1.180145025253296,
307
+ "learning_rate": 0.00024544247787610616,
308
+ "loss": 1.01,
309
+ "step": 1850
310
+ },
311
+ {
312
+ "epoch": 4.203765227021041,
313
+ "grad_norm": 1.0302808284759521,
314
+ "learning_rate": 0.0002520796460176991,
315
+ "loss": 0.9623,
316
+ "step": 1900
317
+ },
318
+ {
319
+ "epoch": 4.314507198228128,
320
+ "grad_norm": 0.7714611291885376,
321
+ "learning_rate": 0.000258716814159292,
322
+ "loss": 0.9025,
323
+ "step": 1950
324
+ },
325
+ {
326
+ "epoch": 4.425249169435216,
327
+ "grad_norm": 0.6419712901115417,
328
+ "learning_rate": 0.00026535398230088497,
329
+ "loss": 0.8814,
330
+ "step": 2000
331
+ },
332
+ {
333
+ "epoch": 4.535991140642303,
334
+ "grad_norm": 0.983061671257019,
335
+ "learning_rate": 0.00027199115044247783,
336
+ "loss": 0.8831,
337
+ "step": 2050
338
+ },
339
+ {
340
+ "epoch": 4.646733111849391,
341
+ "grad_norm": 0.7041178345680237,
342
+ "learning_rate": 0.0002786283185840708,
343
+ "loss": 0.901,
344
+ "step": 2100
345
+ },
346
+ {
347
+ "epoch": 4.7574750830564785,
348
+ "grad_norm": 0.8451335430145264,
349
+ "learning_rate": 0.0002852654867256637,
350
+ "loss": 0.8312,
351
+ "step": 2150
352
+ },
353
+ {
354
+ "epoch": 4.868217054263566,
355
+ "grad_norm": 0.9010829329490662,
356
+ "learning_rate": 0.0002919026548672566,
357
+ "loss": 0.8204,
358
+ "step": 2200
359
+ },
360
+ {
361
+ "epoch": 4.978959025470654,
362
+ "grad_norm": 0.7499646544456482,
363
+ "learning_rate": 0.00029853982300884956,
364
+ "loss": 0.8032,
365
+ "step": 2250
366
+ },
367
+ {
368
+ "epoch": 5.0,
369
+ "eval_cer": 0.13106729228158473,
370
+ "eval_loss": 0.4477018117904663,
371
+ "eval_runtime": 486.3853,
372
+ "eval_samples_per_second": 20.932,
373
+ "eval_steps_per_second": 2.617,
374
+ "eval_wer": 0.4958993391871096,
375
+ "step": 2260
376
+ },
377
+ {
378
+ "epoch": 5.08859357696567,
379
+ "grad_norm": 0.6447732448577881,
380
+ "learning_rate": 0.00029942477876106194,
381
+ "loss": 0.8203,
382
+ "step": 2300
383
+ },
384
+ {
385
+ "epoch": 5.1993355481727574,
386
+ "grad_norm": 0.6052475571632385,
387
+ "learning_rate": 0.00029868731563421826,
388
+ "loss": 0.7706,
389
+ "step": 2350
390
+ },
391
+ {
392
+ "epoch": 5.310077519379845,
393
+ "grad_norm": 0.812167227268219,
394
+ "learning_rate": 0.00029794985250737463,
395
+ "loss": 0.7426,
396
+ "step": 2400
397
+ },
398
+ {
399
+ "epoch": 5.420819490586933,
400
+ "grad_norm": 0.7104111909866333,
401
+ "learning_rate": 0.00029721238938053095,
402
+ "loss": 0.726,
403
+ "step": 2450
404
+ },
405
+ {
406
+ "epoch": 5.53156146179402,
407
+ "grad_norm": 0.8796670436859131,
408
+ "learning_rate": 0.0002964749262536873,
409
+ "loss": 0.7275,
410
+ "step": 2500
411
+ },
412
+ {
413
+ "epoch": 5.642303433001107,
414
+ "grad_norm": 0.7545107007026672,
415
+ "learning_rate": 0.00029573746312684364,
416
+ "loss": 0.737,
417
+ "step": 2550
418
+ },
419
+ {
420
+ "epoch": 5.753045404208195,
421
+ "grad_norm": 0.6987699866294861,
422
+ "learning_rate": 0.00029499999999999996,
423
+ "loss": 0.7179,
424
+ "step": 2600
425
+ },
426
+ {
427
+ "epoch": 5.863787375415282,
428
+ "grad_norm": 0.7108384966850281,
429
+ "learning_rate": 0.00029426253687315633,
430
+ "loss": 0.6831,
431
+ "step": 2650
432
+ },
433
+ {
434
+ "epoch": 5.97452934662237,
435
+ "grad_norm": 0.5486265420913696,
436
+ "learning_rate": 0.00029352507374631265,
437
+ "loss": 0.6855,
438
+ "step": 2700
439
+ },
440
+ {
441
+ "epoch": 6.0,
442
+ "eval_cer": 0.1090644059438252,
443
+ "eval_loss": 0.3938077688217163,
444
+ "eval_runtime": 486.997,
445
+ "eval_samples_per_second": 20.906,
446
+ "eval_steps_per_second": 2.614,
447
+ "eval_wer": 0.41743459762831536,
448
+ "step": 2712
449
+ },
450
+ {
451
+ "epoch": 6.084163898117387,
452
+ "grad_norm": 0.6461381316184998,
453
+ "learning_rate": 0.000292787610619469,
454
+ "loss": 0.6692,
455
+ "step": 2750
456
+ },
457
+ {
458
+ "epoch": 6.194905869324474,
459
+ "grad_norm": 0.6610192060470581,
460
+ "learning_rate": 0.00029205014749262534,
461
+ "loss": 0.6749,
462
+ "step": 2800
463
+ },
464
+ {
465
+ "epoch": 6.305647840531561,
466
+ "grad_norm": 0.6090611815452576,
467
+ "learning_rate": 0.0002913126843657817,
468
+ "loss": 0.6681,
469
+ "step": 2850
470
+ },
471
+ {
472
+ "epoch": 6.416389811738649,
473
+ "grad_norm": 0.7427356243133545,
474
+ "learning_rate": 0.00029057522123893803,
475
+ "loss": 0.6565,
476
+ "step": 2900
477
+ },
478
+ {
479
+ "epoch": 6.5271317829457365,
480
+ "grad_norm": 0.5656064748764038,
481
+ "learning_rate": 0.00028983775811209435,
482
+ "loss": 0.7006,
483
+ "step": 2950
484
+ },
485
+ {
486
+ "epoch": 6.637873754152824,
487
+ "grad_norm": 0.661216676235199,
488
+ "learning_rate": 0.0002891002949852507,
489
+ "loss": 0.6375,
490
+ "step": 3000
491
+ },
492
+ {
493
+ "epoch": 6.748615725359912,
494
+ "grad_norm": 0.6545315384864807,
495
+ "learning_rate": 0.00028836283185840704,
496
+ "loss": 0.6258,
497
+ "step": 3050
498
+ },
499
+ {
500
+ "epoch": 6.859357696566999,
501
+ "grad_norm": 0.7606847882270813,
502
+ "learning_rate": 0.0002876253687315634,
503
+ "loss": 0.6252,
504
+ "step": 3100
505
+ },
506
+ {
507
+ "epoch": 6.970099667774086,
508
+ "grad_norm": 0.7205169200897217,
509
+ "learning_rate": 0.00028688790560471973,
510
+ "loss": 0.6275,
511
+ "step": 3150
512
+ },
513
+ {
514
+ "epoch": 7.0,
515
+ "eval_cer": 0.09786352988955044,
516
+ "eval_loss": 0.34525957703590393,
517
+ "eval_runtime": 489.2546,
518
+ "eval_samples_per_second": 20.809,
519
+ "eval_steps_per_second": 2.602,
520
+ "eval_wer": 0.37516067710690687,
521
+ "step": 3164
522
+ },
523
+ {
524
+ "epoch": 7.079734219269103,
525
+ "grad_norm": 0.7388074398040771,
526
+ "learning_rate": 0.00028615044247787605,
527
+ "loss": 0.6025,
528
+ "step": 3200
529
+ },
530
+ {
531
+ "epoch": 7.190476190476191,
532
+ "grad_norm": 0.74979567527771,
533
+ "learning_rate": 0.0002854129793510324,
534
+ "loss": 0.6064,
535
+ "step": 3250
536
+ },
537
+ {
538
+ "epoch": 7.301218161683278,
539
+ "grad_norm": 0.7727375626564026,
540
+ "learning_rate": 0.00028467551622418874,
541
+ "loss": 0.6019,
542
+ "step": 3300
543
+ },
544
+ {
545
+ "epoch": 7.411960132890365,
546
+ "grad_norm": 0.6809210777282715,
547
+ "learning_rate": 0.0002839380530973451,
548
+ "loss": 0.6096,
549
+ "step": 3350
550
+ },
551
+ {
552
+ "epoch": 7.522702104097453,
553
+ "grad_norm": 0.9523106813430786,
554
+ "learning_rate": 0.00028320058997050143,
555
+ "loss": 0.6002,
556
+ "step": 3400
557
+ },
558
+ {
559
+ "epoch": 7.63344407530454,
560
+ "grad_norm": 0.7200210690498352,
561
+ "learning_rate": 0.0002824631268436578,
562
+ "loss": 0.577,
563
+ "step": 3450
564
+ },
565
+ {
566
+ "epoch": 7.7441860465116275,
567
+ "grad_norm": 0.5474680662155151,
568
+ "learning_rate": 0.0002817256637168141,
569
+ "loss": 0.5672,
570
+ "step": 3500
571
+ },
572
+ {
573
+ "epoch": 7.8549280177187155,
574
+ "grad_norm": 0.868034839630127,
575
+ "learning_rate": 0.0002809882005899705,
576
+ "loss": 0.571,
577
+ "step": 3550
578
+ },
579
+ {
580
+ "epoch": 7.965669988925803,
581
+ "grad_norm": 0.5585504770278931,
582
+ "learning_rate": 0.0002802507374631268,
583
+ "loss": 0.6264,
584
+ "step": 3600
585
+ },
586
+ {
587
+ "epoch": 8.0,
588
+ "eval_cer": 0.09037935760854587,
589
+ "eval_loss": 0.3181062638759613,
590
+ "eval_runtime": 489.6095,
591
+ "eval_samples_per_second": 20.794,
592
+ "eval_steps_per_second": 2.6,
593
+ "eval_wer": 0.35024893636281346,
594
+ "step": 3616
595
+ },
596
+ {
597
+ "epoch": 8.07530454042082,
598
+ "grad_norm": 0.6232703328132629,
599
+ "learning_rate": 0.0002795132743362832,
600
+ "loss": 0.5612,
601
+ "step": 3650
602
+ },
603
+ {
604
+ "epoch": 8.186046511627907,
605
+ "grad_norm": 0.5712502002716064,
606
+ "learning_rate": 0.0002787758112094395,
607
+ "loss": 0.5539,
608
+ "step": 3700
609
+ },
610
+ {
611
+ "epoch": 8.296788482834994,
612
+ "grad_norm": 0.6545235514640808,
613
+ "learning_rate": 0.0002780383480825958,
614
+ "loss": 0.6073,
615
+ "step": 3750
616
+ },
617
+ {
618
+ "epoch": 8.407530454042082,
619
+ "grad_norm": 0.7191429138183594,
620
+ "learning_rate": 0.0002773008849557522,
621
+ "loss": 0.5501,
622
+ "step": 3800
623
+ },
624
+ {
625
+ "epoch": 8.518272425249169,
626
+ "grad_norm": 0.6950603127479553,
627
+ "learning_rate": 0.0002765634218289085,
628
+ "loss": 0.5374,
629
+ "step": 3850
630
+ },
631
+ {
632
+ "epoch": 8.629014396456256,
633
+ "grad_norm": 0.6837615966796875,
634
+ "learning_rate": 0.0002758259587020649,
635
+ "loss": 0.5679,
636
+ "step": 3900
637
+ },
638
+ {
639
+ "epoch": 8.739756367663345,
640
+ "grad_norm": 0.6940374970436096,
641
+ "learning_rate": 0.0002750884955752212,
642
+ "loss": 0.558,
643
+ "step": 3950
644
+ },
645
+ {
646
+ "epoch": 8.850498338870432,
647
+ "grad_norm": 0.643427848815918,
648
+ "learning_rate": 0.0002743510324483776,
649
+ "loss": 0.5241,
650
+ "step": 4000
651
+ },
652
+ {
653
+ "epoch": 8.96124031007752,
654
+ "grad_norm": 0.6302089095115662,
655
+ "learning_rate": 0.0002736135693215339,
656
+ "loss": 0.5388,
657
+ "step": 4050
658
+ },
659
+ {
660
+ "epoch": 9.0,
661
+ "eval_cer": 0.08582140966415502,
662
+ "eval_loss": 0.30813154578208923,
663
+ "eval_runtime": 488.7207,
664
+ "eval_samples_per_second": 20.832,
665
+ "eval_steps_per_second": 2.605,
666
+ "eval_wer": 0.33156513080474337,
667
+ "step": 4068
668
+ },
669
+ {
670
+ "epoch": 9.070874861572536,
671
+ "grad_norm": 0.6800464987754822,
672
+ "learning_rate": 0.00027287610619469027,
673
+ "loss": 0.5446,
674
+ "step": 4100
675
+ },
676
+ {
677
+ "epoch": 9.181616832779623,
678
+ "grad_norm": 0.5239233374595642,
679
+ "learning_rate": 0.0002721386430678466,
680
+ "loss": 0.4957,
681
+ "step": 4150
682
+ },
683
+ {
684
+ "epoch": 9.29235880398671,
685
+ "grad_norm": 0.6076834201812744,
686
+ "learning_rate": 0.0002714011799410029,
687
+ "loss": 0.5316,
688
+ "step": 4200
689
+ },
690
+ {
691
+ "epoch": 9.4031007751938,
692
+ "grad_norm": 0.8120834827423096,
693
+ "learning_rate": 0.0002706637168141593,
694
+ "loss": 0.5766,
695
+ "step": 4250
696
+ },
697
+ {
698
+ "epoch": 9.513842746400886,
699
+ "grad_norm": 0.6490704417228699,
700
+ "learning_rate": 0.0002699262536873156,
701
+ "loss": 0.5044,
702
+ "step": 4300
703
+ },
704
+ {
705
+ "epoch": 9.624584717607974,
706
+ "grad_norm": 0.6406815648078918,
707
+ "learning_rate": 0.00026918879056047197,
708
+ "loss": 0.511,
709
+ "step": 4350
710
+ },
711
+ {
712
+ "epoch": 9.73532668881506,
713
+ "grad_norm": 0.5792708992958069,
714
+ "learning_rate": 0.0002684513274336283,
715
+ "loss": 0.5195,
716
+ "step": 4400
717
+ },
718
+ {
719
+ "epoch": 9.846068660022148,
720
+ "grad_norm": 0.7325103878974915,
721
+ "learning_rate": 0.0002677138643067846,
722
+ "loss": 0.5012,
723
+ "step": 4450
724
+ },
725
+ {
726
+ "epoch": 9.956810631229235,
727
+ "grad_norm": 0.7060527205467224,
728
+ "learning_rate": 0.000266976401179941,
729
+ "loss": 0.5194,
730
+ "step": 4500
731
+ },
732
+ {
733
+ "epoch": 10.0,
734
+ "eval_cer": 0.08035332255187068,
735
+ "eval_loss": 0.28676968812942505,
736
+ "eval_runtime": 487.2322,
737
+ "eval_samples_per_second": 20.896,
738
+ "eval_steps_per_second": 2.613,
739
+ "eval_wer": 0.3129718475604236,
740
+ "step": 4520
741
+ },
742
+ {
743
+ "epoch": 10.066445182724253,
744
+ "grad_norm": 0.6046482920646667,
745
+ "learning_rate": 0.0002662389380530973,
746
+ "loss": 0.489,
747
+ "step": 4550
748
+ },
749
+ {
750
+ "epoch": 10.17718715393134,
751
+ "grad_norm": 0.7882891893386841,
752
+ "learning_rate": 0.00026550147492625367,
753
+ "loss": 0.4864,
754
+ "step": 4600
755
+ },
756
+ {
757
+ "epoch": 10.287929125138428,
758
+ "grad_norm": 0.8176812529563904,
759
+ "learning_rate": 0.00026476401179941,
760
+ "loss": 0.4962,
761
+ "step": 4650
762
+ },
763
+ {
764
+ "epoch": 10.398671096345515,
765
+ "grad_norm": 0.5068700313568115,
766
+ "learning_rate": 0.00026402654867256636,
767
+ "loss": 0.4777,
768
+ "step": 4700
769
+ },
770
+ {
771
+ "epoch": 10.509413067552602,
772
+ "grad_norm": 0.5760357975959778,
773
+ "learning_rate": 0.0002632890855457227,
774
+ "loss": 0.4758,
775
+ "step": 4750
776
+ },
777
+ {
778
+ "epoch": 10.62015503875969,
779
+ "grad_norm": 0.704738974571228,
780
+ "learning_rate": 0.00026255162241887905,
781
+ "loss": 0.4889,
782
+ "step": 4800
783
+ },
784
+ {
785
+ "epoch": 10.730897009966778,
786
+ "grad_norm": 0.5809513926506042,
787
+ "learning_rate": 0.00026181415929203537,
788
+ "loss": 0.4811,
789
+ "step": 4850
790
+ },
791
+ {
792
+ "epoch": 10.841638981173865,
793
+ "grad_norm": 0.5770090222358704,
794
+ "learning_rate": 0.0002610766961651917,
795
+ "loss": 0.515,
796
+ "step": 4900
797
+ },
798
+ {
799
+ "epoch": 10.952380952380953,
800
+ "grad_norm": 0.5710371136665344,
801
+ "learning_rate": 0.00026033923303834806,
802
+ "loss": 0.4798,
803
+ "step": 4950
804
+ },
805
+ {
806
+ "epoch": 11.0,
807
+ "eval_cer": 0.07714426612324227,
808
+ "eval_loss": 0.27717578411102295,
809
+ "eval_runtime": 489.1751,
810
+ "eval_samples_per_second": 20.813,
811
+ "eval_steps_per_second": 2.602,
812
+ "eval_wer": 0.2999728433058749,
813
+ "step": 4972
814
+ },
815
+ {
816
+ "epoch": 11.062015503875969,
817
+ "grad_norm": 0.6379249691963196,
818
+ "learning_rate": 0.0002596017699115044,
819
+ "loss": 0.5031,
820
+ "step": 5000
821
+ },
822
+ {
823
+ "epoch": 11.172757475083056,
824
+ "grad_norm": 0.4557214677333832,
825
+ "learning_rate": 0.00025886430678466075,
826
+ "loss": 0.4546,
827
+ "step": 5050
828
+ },
829
+ {
830
+ "epoch": 11.283499446290143,
831
+ "grad_norm": 0.9845902919769287,
832
+ "learning_rate": 0.00025812684365781707,
833
+ "loss": 0.4494,
834
+ "step": 5100
835
+ },
836
+ {
837
+ "epoch": 11.39424141749723,
838
+ "grad_norm": 0.5986599922180176,
839
+ "learning_rate": 0.00025738938053097344,
840
+ "loss": 0.466,
841
+ "step": 5150
842
+ },
843
+ {
844
+ "epoch": 11.50498338870432,
845
+ "grad_norm": 0.5532345771789551,
846
+ "learning_rate": 0.00025665191740412976,
847
+ "loss": 0.4696,
848
+ "step": 5200
849
+ },
850
+ {
851
+ "epoch": 11.615725359911407,
852
+ "grad_norm": 0.5763828754425049,
853
+ "learning_rate": 0.00025591445427728613,
854
+ "loss": 0.4423,
855
+ "step": 5250
856
+ },
857
+ {
858
+ "epoch": 11.726467331118494,
859
+ "grad_norm": 0.5517535209655762,
860
+ "learning_rate": 0.00025517699115044245,
861
+ "loss": 0.4344,
862
+ "step": 5300
863
+ },
864
+ {
865
+ "epoch": 11.837209302325581,
866
+ "grad_norm": 0.6629925966262817,
867
+ "learning_rate": 0.0002544395280235988,
868
+ "loss": 0.452,
869
+ "step": 5350
870
+ },
871
+ {
872
+ "epoch": 11.947951273532668,
873
+ "grad_norm": 0.5328862071037292,
874
+ "learning_rate": 0.00025370206489675514,
875
+ "loss": 0.4523,
876
+ "step": 5400
877
+ },
878
+ {
879
+ "epoch": 12.0,
880
+ "eval_cer": 0.07210042714897999,
881
+ "eval_loss": 0.26620689034461975,
882
+ "eval_runtime": 489.7467,
883
+ "eval_samples_per_second": 20.788,
884
+ "eval_steps_per_second": 2.599,
885
+ "eval_wer": 0.2804562324613017,
886
+ "step": 5424
887
+ },
888
+ {
889
+ "epoch": 12.057585825027685,
890
+ "grad_norm": 0.5896037220954895,
891
+ "learning_rate": 0.0002529646017699115,
892
+ "loss": 0.4961,
893
+ "step": 5450
894
+ },
895
+ {
896
+ "epoch": 12.168327796234774,
897
+ "grad_norm": 0.6215909123420715,
898
+ "learning_rate": 0.00025222713864306783,
899
+ "loss": 0.4503,
900
+ "step": 5500
901
+ },
902
+ {
903
+ "epoch": 12.279069767441861,
904
+ "grad_norm": 0.7767740488052368,
905
+ "learning_rate": 0.00025148967551622415,
906
+ "loss": 0.4421,
907
+ "step": 5550
908
+ },
909
+ {
910
+ "epoch": 12.389811738648948,
911
+ "grad_norm": 0.7437320351600647,
912
+ "learning_rate": 0.0002507522123893805,
913
+ "loss": 0.4316,
914
+ "step": 5600
915
+ },
916
+ {
917
+ "epoch": 12.500553709856035,
918
+ "grad_norm": 0.8144622445106506,
919
+ "learning_rate": 0.00025001474926253684,
920
+ "loss": 0.4599,
921
+ "step": 5650
922
+ },
923
+ {
924
+ "epoch": 12.611295681063122,
925
+ "grad_norm": 0.5992530584335327,
926
+ "learning_rate": 0.00024927728613569316,
927
+ "loss": 0.4339,
928
+ "step": 5700
929
+ },
930
+ {
931
+ "epoch": 12.72203765227021,
932
+ "grad_norm": 0.7761603593826294,
933
+ "learning_rate": 0.00024853982300884953,
934
+ "loss": 0.4342,
935
+ "step": 5750
936
+ },
937
+ {
938
+ "epoch": 12.832779623477299,
939
+ "grad_norm": 0.5413389205932617,
940
+ "learning_rate": 0.00024780235988200585,
941
+ "loss": 0.4442,
942
+ "step": 5800
943
+ },
944
+ {
945
+ "epoch": 12.943521594684386,
946
+ "grad_norm": 0.5843682885169983,
947
+ "learning_rate": 0.0002470648967551622,
948
+ "loss": 0.4424,
949
+ "step": 5850
950
+ },
951
+ {
952
+ "epoch": 13.0,
953
+ "eval_cer": 0.07174870006019247,
954
+ "eval_loss": 0.26166489720344543,
955
+ "eval_runtime": 492.0407,
956
+ "eval_samples_per_second": 20.691,
957
+ "eval_steps_per_second": 2.587,
958
+ "eval_wer": 0.27929754684529734,
959
+ "step": 5876
960
+ },
961
+ {
962
+ "epoch": 13.053156146179402,
963
+ "grad_norm": 0.9268521070480347,
964
+ "learning_rate": 0.00024632743362831854,
965
+ "loss": 0.4322,
966
+ "step": 5900
967
+ },
968
+ {
969
+ "epoch": 13.16389811738649,
970
+ "grad_norm": 0.5855753421783447,
971
+ "learning_rate": 0.0002455899705014749,
972
+ "loss": 0.4314,
973
+ "step": 5950
974
+ },
975
+ {
976
+ "epoch": 13.274640088593577,
977
+ "grad_norm": 0.710689127445221,
978
+ "learning_rate": 0.00024485250737463123,
979
+ "loss": 0.4407,
980
+ "step": 6000
981
+ },
982
+ {
983
+ "epoch": 13.385382059800664,
984
+ "grad_norm": 0.6203360557556152,
985
+ "learning_rate": 0.0002441150442477876,
986
+ "loss": 0.4394,
987
+ "step": 6050
988
+ },
989
+ {
990
+ "epoch": 13.496124031007753,
991
+ "grad_norm": 0.6015070676803589,
992
+ "learning_rate": 0.00024337758112094392,
993
+ "loss": 0.4359,
994
+ "step": 6100
995
+ },
996
+ {
997
+ "epoch": 13.60686600221484,
998
+ "grad_norm": 0.7456609606742859,
999
+ "learning_rate": 0.00024264011799410027,
1000
+ "loss": 0.477,
1001
+ "step": 6150
1002
+ },
1003
+ {
1004
+ "epoch": 13.717607973421927,
1005
+ "grad_norm": 0.6540862917900085,
1006
+ "learning_rate": 0.00024190265486725662,
1007
+ "loss": 0.4415,
1008
+ "step": 6200
1009
+ },
1010
+ {
1011
+ "epoch": 13.828349944629014,
1012
+ "grad_norm": 0.4988921880722046,
1013
+ "learning_rate": 0.00024116519174041293,
1014
+ "loss": 0.4131,
1015
+ "step": 6250
1016
+ },
1017
+ {
1018
+ "epoch": 13.939091915836102,
1019
+ "grad_norm": 0.7218478322029114,
1020
+ "learning_rate": 0.0002404277286135693,
1021
+ "loss": 0.4179,
1022
+ "step": 6300
1023
+ },
1024
+ {
1025
+ "epoch": 14.0,
1026
+ "eval_cer": 0.06874995467434424,
1027
+ "eval_loss": 0.2501542270183563,
1028
+ "eval_runtime": 495.114,
1029
+ "eval_samples_per_second": 20.563,
1030
+ "eval_steps_per_second": 2.571,
1031
+ "eval_wer": 0.2687245405992577,
1032
+ "step": 6328
1033
+ },
1034
+ {
1035
+ "epoch": 14.048726467331118,
1036
+ "grad_norm": 0.5830133557319641,
1037
+ "learning_rate": 0.00023969026548672563,
1038
+ "loss": 0.4341,
1039
+ "step": 6350
1040
+ },
1041
+ {
1042
+ "epoch": 14.159468438538205,
1043
+ "grad_norm": 0.762266993522644,
1044
+ "learning_rate": 0.000238952802359882,
1045
+ "loss": 0.4129,
1046
+ "step": 6400
1047
+ },
1048
+ {
1049
+ "epoch": 14.270210409745294,
1050
+ "grad_norm": 0.49856090545654297,
1051
+ "learning_rate": 0.00023821533923303832,
1052
+ "loss": 0.4356,
1053
+ "step": 6450
1054
+ },
1055
+ {
1056
+ "epoch": 14.380952380952381,
1057
+ "grad_norm": 0.6130903959274292,
1058
+ "learning_rate": 0.0002374778761061947,
1059
+ "loss": 0.4012,
1060
+ "step": 6500
1061
+ },
1062
+ {
1063
+ "epoch": 14.491694352159469,
1064
+ "grad_norm": 0.8215532898902893,
1065
+ "learning_rate": 0.000236740412979351,
1066
+ "loss": 0.4061,
1067
+ "step": 6550
1068
+ },
1069
+ {
1070
+ "epoch": 14.602436323366556,
1071
+ "grad_norm": 0.7206094861030579,
1072
+ "learning_rate": 0.00023600294985250735,
1073
+ "loss": 0.4021,
1074
+ "step": 6600
1075
+ },
1076
+ {
1077
+ "epoch": 14.713178294573643,
1078
+ "grad_norm": 0.6363425254821777,
1079
+ "learning_rate": 0.0002352654867256637,
1080
+ "loss": 0.412,
1081
+ "step": 6650
1082
+ },
1083
+ {
1084
+ "epoch": 14.82392026578073,
1085
+ "grad_norm": 0.5326490998268127,
1086
+ "learning_rate": 0.00023452802359882004,
1087
+ "loss": 0.3978,
1088
+ "step": 6700
1089
+ },
1090
+ {
1091
+ "epoch": 14.934662236987819,
1092
+ "grad_norm": 0.554918646812439,
1093
+ "learning_rate": 0.00023379056047197636,
1094
+ "loss": 0.415,
1095
+ "step": 6750
1096
+ },
1097
+ {
1098
+ "epoch": 15.0,
1099
+ "eval_cer": 0.06736842868642623,
1100
+ "eval_loss": 0.2500366270542145,
1101
+ "eval_runtime": 493.6452,
1102
+ "eval_samples_per_second": 20.624,
1103
+ "eval_steps_per_second": 2.579,
1104
+ "eval_wer": 0.2614465465737304,
1105
+ "step": 6780
1106
+ },
1107
+ {
1108
+ "epoch": 15.044296788482836,
1109
+ "grad_norm": 0.6449525356292725,
1110
+ "learning_rate": 0.00023305309734513273,
1111
+ "loss": 0.4243,
1112
+ "step": 6800
1113
+ },
1114
+ {
1115
+ "epoch": 15.155038759689923,
1116
+ "grad_norm": 1.446018099784851,
1117
+ "learning_rate": 0.00023231563421828905,
1118
+ "loss": 0.4122,
1119
+ "step": 6850
1120
+ },
1121
+ {
1122
+ "epoch": 15.26578073089701,
1123
+ "grad_norm": 0.5363207459449768,
1124
+ "learning_rate": 0.00023157817109144542,
1125
+ "loss": 0.4125,
1126
+ "step": 6900
1127
+ },
1128
+ {
1129
+ "epoch": 15.376522702104097,
1130
+ "grad_norm": 0.6152534484863281,
1131
+ "learning_rate": 0.00023084070796460174,
1132
+ "loss": 0.3998,
1133
+ "step": 6950
1134
+ },
1135
+ {
1136
+ "epoch": 15.487264673311184,
1137
+ "grad_norm": 0.592438817024231,
1138
+ "learning_rate": 0.00023010324483775812,
1139
+ "loss": 0.3853,
1140
+ "step": 7000
1141
+ },
1142
+ {
1143
+ "epoch": 15.598006644518273,
1144
+ "grad_norm": 0.6657503843307495,
1145
+ "learning_rate": 0.00022936578171091443,
1146
+ "loss": 0.3951,
1147
+ "step": 7050
1148
+ },
1149
+ {
1150
+ "epoch": 15.70874861572536,
1151
+ "grad_norm": 0.5211730599403381,
1152
+ "learning_rate": 0.00022862831858407078,
1153
+ "loss": 0.4023,
1154
+ "step": 7100
1155
+ },
1156
+ {
1157
+ "epoch": 15.819490586932448,
1158
+ "grad_norm": 0.5348416566848755,
1159
+ "learning_rate": 0.00022789085545722713,
1160
+ "loss": 0.4093,
1161
+ "step": 7150
1162
+ },
1163
+ {
1164
+ "epoch": 15.930232558139535,
1165
+ "grad_norm": 0.927044689655304,
1166
+ "learning_rate": 0.00022715339233038347,
1167
+ "loss": 0.4405,
1168
+ "step": 7200
1169
+ },
1170
+ {
1171
+ "epoch": 16.0,
1172
+ "eval_cer": 0.06604491953789587,
1173
+ "eval_loss": 0.2411191165447235,
1174
+ "eval_runtime": 494.0285,
1175
+ "eval_samples_per_second": 20.608,
1176
+ "eval_steps_per_second": 2.577,
1177
+ "eval_wer": 0.2560876255997103,
1178
+ "step": 7232
1179
+ },
1180
+ {
1181
+ "epoch": 16.039867109634553,
1182
+ "grad_norm": 0.5035960674285889,
1183
+ "learning_rate": 0.00022641592920353982,
1184
+ "loss": 0.4186,
1185
+ "step": 7250
1186
+ },
1187
+ {
1188
+ "epoch": 16.15060908084164,
1189
+ "grad_norm": 0.7938876748085022,
1190
+ "learning_rate": 0.00022567846607669613,
1191
+ "loss": 0.4016,
1192
+ "step": 7300
1193
+ },
1194
+ {
1195
+ "epoch": 16.261351052048727,
1196
+ "grad_norm": 0.6998234987258911,
1197
+ "learning_rate": 0.00022494100294985248,
1198
+ "loss": 0.3891,
1199
+ "step": 7350
1200
+ },
1201
+ {
1202
+ "epoch": 16.372093023255815,
1203
+ "grad_norm": 0.5634494423866272,
1204
+ "learning_rate": 0.00022420353982300883,
1205
+ "loss": 0.3768,
1206
+ "step": 7400
1207
+ },
1208
+ {
1209
+ "epoch": 16.482834994462902,
1210
+ "grad_norm": 0.6078976392745972,
1211
+ "learning_rate": 0.00022346607669616517,
1212
+ "loss": 0.3974,
1213
+ "step": 7450
1214
+ },
1215
+ {
1216
+ "epoch": 16.59357696566999,
1217
+ "grad_norm": 1.2107906341552734,
1218
+ "learning_rate": 0.0002227286135693215,
1219
+ "loss": 0.3791,
1220
+ "step": 7500
1221
+ },
1222
+ {
1223
+ "epoch": 16.704318936877076,
1224
+ "grad_norm": 0.7613789439201355,
1225
+ "learning_rate": 0.00022199115044247786,
1226
+ "loss": 0.3713,
1227
+ "step": 7550
1228
+ },
1229
+ {
1230
+ "epoch": 16.815060908084163,
1231
+ "grad_norm": 0.5816598534584045,
1232
+ "learning_rate": 0.00022125368731563418,
1233
+ "loss": 0.3817,
1234
+ "step": 7600
1235
+ },
1236
+ {
1237
+ "epoch": 16.92580287929125,
1238
+ "grad_norm": 0.8725369572639465,
1239
+ "learning_rate": 0.00022051622418879055,
1240
+ "loss": 0.3812,
1241
+ "step": 7650
1242
+ },
1243
+ {
1244
+ "epoch": 17.0,
1245
+ "eval_cer": 0.06427903198903481,
1246
+ "eval_loss": 0.23461875319480896,
1247
+ "eval_runtime": 492.7771,
1248
+ "eval_samples_per_second": 20.66,
1249
+ "eval_steps_per_second": 2.583,
1250
+ "eval_wer": 0.2501674662804381,
1251
+ "step": 7684
1252
+ },
1253
+ {
1254
+ "epoch": 17.035437430786267,
1255
+ "grad_norm": 0.5457949638366699,
1256
+ "learning_rate": 0.00021977876106194687,
1257
+ "loss": 0.389,
1258
+ "step": 7700
1259
+ },
1260
+ {
1261
+ "epoch": 17.146179401993354,
1262
+ "grad_norm": 0.6383467316627502,
1263
+ "learning_rate": 0.00021904129793510324,
1264
+ "loss": 0.3842,
1265
+ "step": 7750
1266
+ },
1267
+ {
1268
+ "epoch": 17.25692137320044,
1269
+ "grad_norm": 4.207304954528809,
1270
+ "learning_rate": 0.00021830383480825956,
1271
+ "loss": 0.3459,
1272
+ "step": 7800
1273
+ },
1274
+ {
1275
+ "epoch": 17.367663344407532,
1276
+ "grad_norm": 1.1675844192504883,
1277
+ "learning_rate": 0.0002175663716814159,
1278
+ "loss": 0.3825,
1279
+ "step": 7850
1280
+ },
1281
+ {
1282
+ "epoch": 17.47840531561462,
1283
+ "grad_norm": 0.44133588671684265,
1284
+ "learning_rate": 0.00021682890855457225,
1285
+ "loss": 0.3911,
1286
+ "step": 7900
1287
+ },
1288
+ {
1289
+ "epoch": 17.589147286821706,
1290
+ "grad_norm": 0.5262250304222107,
1291
+ "learning_rate": 0.0002160914454277286,
1292
+ "loss": 0.383,
1293
+ "step": 7950
1294
+ },
1295
+ {
1296
+ "epoch": 17.699889258028794,
1297
+ "grad_norm": 0.7032670378684998,
1298
+ "learning_rate": 0.00021535398230088494,
1299
+ "loss": 0.4,
1300
+ "step": 8000
1301
+ },
1302
+ {
1303
+ "epoch": 17.81063122923588,
1304
+ "grad_norm": 0.48254212737083435,
1305
+ "learning_rate": 0.0002146165191740413,
1306
+ "loss": 0.3697,
1307
+ "step": 8050
1308
+ },
1309
+ {
1310
+ "epoch": 17.921373200442968,
1311
+ "grad_norm": 0.629219651222229,
1312
+ "learning_rate": 0.0002138790560471976,
1313
+ "loss": 0.3752,
1314
+ "step": 8100
1315
+ },
1316
+ {
1317
+ "epoch": 18.0,
1318
+ "eval_cer": 0.06295189678804274,
1319
+ "eval_loss": 0.23475028574466705,
1320
+ "eval_runtime": 494.2681,
1321
+ "eval_samples_per_second": 20.598,
1322
+ "eval_steps_per_second": 2.576,
1323
+ "eval_wer": 0.2447180229926677,
1324
+ "step": 8136
1325
+ },
1326
+ {
1327
+ "epoch": 18.031007751937985,
1328
+ "grad_norm": 0.5522592663764954,
1329
+ "learning_rate": 0.00021314159292035398,
1330
+ "loss": 0.4118,
1331
+ "step": 8150
1332
+ },
1333
+ {
1334
+ "epoch": 18.14174972314507,
1335
+ "grad_norm": 0.6355589628219604,
1336
+ "learning_rate": 0.0002124041297935103,
1337
+ "loss": 0.3613,
1338
+ "step": 8200
1339
+ },
1340
+ {
1341
+ "epoch": 18.25249169435216,
1342
+ "grad_norm": 0.6213610172271729,
1343
+ "learning_rate": 0.00021166666666666667,
1344
+ "loss": 0.3561,
1345
+ "step": 8250
1346
+ },
1347
+ {
1348
+ "epoch": 18.363233665559246,
1349
+ "grad_norm": 0.6862598657608032,
1350
+ "learning_rate": 0.000210929203539823,
1351
+ "loss": 0.3822,
1352
+ "step": 8300
1353
+ },
1354
+ {
1355
+ "epoch": 18.473975636766333,
1356
+ "grad_norm": 0.4650695323944092,
1357
+ "learning_rate": 0.00021019174041297936,
1358
+ "loss": 0.3708,
1359
+ "step": 8350
1360
+ },
1361
+ {
1362
+ "epoch": 18.58471760797342,
1363
+ "grad_norm": 0.6250514388084412,
1364
+ "learning_rate": 0.00020945427728613568,
1365
+ "loss": 0.371,
1366
+ "step": 8400
1367
+ },
1368
+ {
1369
+ "epoch": 18.695459579180508,
1370
+ "grad_norm": 1.5113970041275024,
1371
+ "learning_rate": 0.000208716814159292,
1372
+ "loss": 0.3526,
1373
+ "step": 8450
1374
+ },
1375
+ {
1376
+ "epoch": 18.8062015503876,
1377
+ "grad_norm": 0.5471574068069458,
1378
+ "learning_rate": 0.00020797935103244837,
1379
+ "loss": 0.3789,
1380
+ "step": 8500
1381
+ },
1382
+ {
1383
+ "epoch": 18.916943521594686,
1384
+ "grad_norm": 0.5146489143371582,
1385
+ "learning_rate": 0.0002072418879056047,
1386
+ "loss": 0.3565,
1387
+ "step": 8550
1388
+ },
1389
+ {
1390
+ "epoch": 19.0,
1391
+ "eval_cer": 0.06287574968634646,
1392
+ "eval_loss": 0.23113583028316498,
1393
+ "eval_runtime": 494.568,
1394
+ "eval_samples_per_second": 20.586,
1395
+ "eval_steps_per_second": 2.574,
1396
+ "eval_wer": 0.24508011224766904,
1397
+ "step": 8588
1398
+ },
1399
+ {
1400
+ "epoch": 19.026578073089702,
1401
+ "grad_norm": 0.9078727960586548,
1402
+ "learning_rate": 0.00020650442477876103,
1403
+ "loss": 0.3678,
1404
+ "step": 8600
1405
+ },
1406
+ {
1407
+ "epoch": 19.13732004429679,
1408
+ "grad_norm": 0.6155833601951599,
1409
+ "learning_rate": 0.00020576696165191738,
1410
+ "loss": 0.3483,
1411
+ "step": 8650
1412
+ },
1413
+ {
1414
+ "epoch": 19.248062015503876,
1415
+ "grad_norm": 0.4708555042743683,
1416
+ "learning_rate": 0.00020502949852507373,
1417
+ "loss": 0.3504,
1418
+ "step": 8700
1419
+ },
1420
+ {
1421
+ "epoch": 19.358803986710964,
1422
+ "grad_norm": 0.5760027170181274,
1423
+ "learning_rate": 0.00020429203539823007,
1424
+ "loss": 0.3521,
1425
+ "step": 8750
1426
+ },
1427
+ {
1428
+ "epoch": 19.46954595791805,
1429
+ "grad_norm": 0.5396759510040283,
1430
+ "learning_rate": 0.00020355457227138642,
1431
+ "loss": 0.3583,
1432
+ "step": 8800
1433
+ },
1434
+ {
1435
+ "epoch": 19.580287929125138,
1436
+ "grad_norm": 0.4109143912792206,
1437
+ "learning_rate": 0.00020281710914454273,
1438
+ "loss": 0.3421,
1439
+ "step": 8850
1440
+ },
1441
+ {
1442
+ "epoch": 19.691029900332225,
1443
+ "grad_norm": 0.5602762699127197,
1444
+ "learning_rate": 0.0002020796460176991,
1445
+ "loss": 0.3616,
1446
+ "step": 8900
1447
+ },
1448
+ {
1449
+ "epoch": 19.801771871539312,
1450
+ "grad_norm": 0.4505002200603485,
1451
+ "learning_rate": 0.00020134218289085543,
1452
+ "loss": 0.3561,
1453
+ "step": 8950
1454
+ },
1455
+ {
1456
+ "epoch": 19.9125138427464,
1457
+ "grad_norm": 0.6884456872940063,
1458
+ "learning_rate": 0.0002006047197640118,
1459
+ "loss": 0.3675,
1460
+ "step": 9000
1461
+ },
1462
+ {
1463
+ "epoch": 20.0,
1464
+ "eval_cer": 0.06083065609793242,
1465
+ "eval_loss": 0.2296416014432907,
1466
+ "eval_runtime": 495.6675,
1467
+ "eval_samples_per_second": 20.54,
1468
+ "eval_steps_per_second": 2.568,
1469
+ "eval_wer": 0.23590114963338463,
1470
+ "step": 9040
1471
+ },
1472
+ {
1473
+ "epoch": 20.022148394241416,
1474
+ "grad_norm": 0.706529974937439,
1475
+ "learning_rate": 0.00019986725663716812,
1476
+ "loss": 0.3583,
1477
+ "step": 9050
1478
+ },
1479
+ {
1480
+ "epoch": 20.132890365448507,
1481
+ "grad_norm": 1.8911956548690796,
1482
+ "learning_rate": 0.0001991297935103245,
1483
+ "loss": 0.3548,
1484
+ "step": 9100
1485
+ },
1486
+ {
1487
+ "epoch": 20.243632336655594,
1488
+ "grad_norm": NaN,
1489
+ "learning_rate": 0.0001983923303834808,
1490
+ "loss": 0.3652,
1491
+ "step": 9150
1492
+ },
1493
+ {
1494
+ "epoch": 20.35437430786268,
1495
+ "grad_norm": 0.5242050290107727,
1496
+ "learning_rate": 0.00019765486725663715,
1497
+ "loss": 0.3481,
1498
+ "step": 9200
1499
+ },
1500
+ {
1501
+ "epoch": 20.46511627906977,
1502
+ "grad_norm": 0.4361858665943146,
1503
+ "learning_rate": 0.0001969174041297935,
1504
+ "loss": 0.346,
1505
+ "step": 9250
1506
+ },
1507
+ {
1508
+ "epoch": 20.575858250276855,
1509
+ "grad_norm": 0.4760480523109436,
1510
+ "learning_rate": 0.00019617994100294984,
1511
+ "loss": 0.3394,
1512
+ "step": 9300
1513
+ },
1514
+ {
1515
+ "epoch": 20.686600221483943,
1516
+ "grad_norm": 0.5500379800796509,
1517
+ "learning_rate": 0.00019544247787610616,
1518
+ "loss": 0.3631,
1519
+ "step": 9350
1520
+ },
1521
+ {
1522
+ "epoch": 20.79734219269103,
1523
+ "grad_norm": 0.5537223815917969,
1524
+ "learning_rate": 0.00019470501474926253,
1525
+ "loss": 0.3517,
1526
+ "step": 9400
1527
+ },
1528
+ {
1529
+ "epoch": 20.908084163898117,
1530
+ "grad_norm": 1.2052114009857178,
1531
+ "learning_rate": 0.00019396755162241885,
1532
+ "loss": 0.3318,
1533
+ "step": 9450
1534
+ },
1535
+ {
1536
+ "epoch": 21.0,
1537
+ "eval_cer": 0.06032663480575237,
1538
+ "eval_loss": 0.2251521646976471,
1539
+ "eval_runtime": 497.982,
1540
+ "eval_samples_per_second": 20.445,
1541
+ "eval_steps_per_second": 2.556,
1542
+ "eval_wer": 0.23347515162487553,
1543
+ "step": 9492
1544
+ },
1545
+ {
1546
+ "epoch": 21.017718715393134,
1547
+ "grad_norm": 0.4967649281024933,
1548
+ "learning_rate": 0.00019323008849557523,
1549
+ "loss": 0.3693,
1550
+ "step": 9500
1551
+ },
1552
+ {
1553
+ "epoch": 21.12846068660022,
1554
+ "grad_norm": 0.40202295780181885,
1555
+ "learning_rate": 0.00019249262536873154,
1556
+ "loss": 0.3454,
1557
+ "step": 9550
1558
+ },
1559
+ {
1560
+ "epoch": 21.239202657807308,
1561
+ "grad_norm": 0.4794892370700836,
1562
+ "learning_rate": 0.00019175516224188792,
1563
+ "loss": 0.3286,
1564
+ "step": 9600
1565
+ },
1566
+ {
1567
+ "epoch": 21.349944629014395,
1568
+ "grad_norm": 0.5660750865936279,
1569
+ "learning_rate": 0.00019101769911504423,
1570
+ "loss": 0.3384,
1571
+ "step": 9650
1572
+ },
1573
+ {
1574
+ "epoch": 21.460686600221482,
1575
+ "grad_norm": 0.5895997881889343,
1576
+ "learning_rate": 0.00019028023598820055,
1577
+ "loss": 0.3634,
1578
+ "step": 9700
1579
+ },
1580
+ {
1581
+ "epoch": 21.571428571428573,
1582
+ "grad_norm": 0.33758455514907837,
1583
+ "learning_rate": 0.00018954277286135693,
1584
+ "loss": 0.3376,
1585
+ "step": 9750
1586
+ },
1587
+ {
1588
+ "epoch": 21.68217054263566,
1589
+ "grad_norm": 0.44639402627944946,
1590
+ "learning_rate": 0.00018880530973451324,
1591
+ "loss": 0.3532,
1592
+ "step": 9800
1593
+ },
1594
+ {
1595
+ "epoch": 21.792912513842747,
1596
+ "grad_norm": 0.4666491746902466,
1597
+ "learning_rate": 0.00018806784660766962,
1598
+ "loss": 0.3467,
1599
+ "step": 9850
1600
+ },
1601
+ {
1602
+ "epoch": 21.903654485049834,
1603
+ "grad_norm": 0.36474844813346863,
1604
+ "learning_rate": 0.00018733038348082593,
1605
+ "loss": 0.3479,
1606
+ "step": 9900
1607
+ },
1608
+ {
1609
+ "epoch": 22.0,
1610
+ "eval_cer": 0.059242445119695994,
1611
+ "eval_loss": 0.2179824411869049,
1612
+ "eval_runtime": 496.9044,
1613
+ "eval_samples_per_second": 20.489,
1614
+ "eval_steps_per_second": 2.562,
1615
+ "eval_wer": 0.23155607857336832,
1616
+ "step": 9944
1617
+ },
1618
+ {
1619
+ "epoch": 22.01328903654485,
1620
+ "grad_norm": 0.42274388670921326,
1621
+ "learning_rate": 0.00018659292035398228,
1622
+ "loss": 0.3619,
1623
+ "step": 9950
1624
+ },
1625
+ {
1626
+ "epoch": 22.124031007751938,
1627
+ "grad_norm": 0.6178750395774841,
1628
+ "learning_rate": 0.00018585545722713863,
1629
+ "loss": 0.3358,
1630
+ "step": 10000
1631
+ },
1632
+ {
1633
+ "epoch": 22.234772978959025,
1634
+ "grad_norm": 0.39765465259552,
1635
+ "learning_rate": 0.00018511799410029497,
1636
+ "loss": 0.3354,
1637
+ "step": 10050
1638
+ },
1639
+ {
1640
+ "epoch": 22.345514950166113,
1641
+ "grad_norm": 1.0347713232040405,
1642
+ "learning_rate": 0.0001843805309734513,
1643
+ "loss": 0.3266,
1644
+ "step": 10100
1645
+ },
1646
+ {
1647
+ "epoch": 22.4562569213732,
1648
+ "grad_norm": 0.4053027927875519,
1649
+ "learning_rate": 0.00018364306784660766,
1650
+ "loss": 0.3445,
1651
+ "step": 10150
1652
+ },
1653
+ {
1654
+ "epoch": 22.566998892580287,
1655
+ "grad_norm": 0.5446871519088745,
1656
+ "learning_rate": 0.00018290560471976398,
1657
+ "loss": 0.3585,
1658
+ "step": 10200
1659
+ },
1660
+ {
1661
+ "epoch": 22.677740863787374,
1662
+ "grad_norm": 0.6107918620109558,
1663
+ "learning_rate": 0.00018216814159292035,
1664
+ "loss": 0.3418,
1665
+ "step": 10250
1666
+ },
1667
+ {
1668
+ "epoch": 22.78848283499446,
1669
+ "grad_norm": 0.5900046825408936,
1670
+ "learning_rate": 0.00018143067846607667,
1671
+ "loss": 0.3483,
1672
+ "step": 10300
1673
+ },
1674
+ {
1675
+ "epoch": 22.899224806201552,
1676
+ "grad_norm": 0.42756062746047974,
1677
+ "learning_rate": 0.00018069321533923304,
1678
+ "loss": 0.33,
1679
+ "step": 10350
1680
+ },
1681
+ {
1682
+ "epoch": 23.0,
1683
+ "eval_cer": 0.05697253627865488,
1684
+ "eval_loss": 0.21119394898414612,
1685
+ "eval_runtime": 496.1257,
1686
+ "eval_samples_per_second": 20.521,
1687
+ "eval_steps_per_second": 2.566,
1688
+ "eval_wer": 0.22308319000633656,
1689
+ "step": 10396
1690
+ },
1691
+ {
1692
+ "epoch": 23.00885935769657,
1693
+ "grad_norm": 0.6858476996421814,
1694
+ "learning_rate": 0.00017995575221238936,
1695
+ "loss": 0.3312,
1696
+ "step": 10400
1697
+ },
1698
+ {
1699
+ "epoch": 23.119601328903656,
1700
+ "grad_norm": 0.5675090551376343,
1701
+ "learning_rate": 0.0001792182890855457,
1702
+ "loss": 0.3219,
1703
+ "step": 10450
1704
+ },
1705
+ {
1706
+ "epoch": 23.230343300110743,
1707
+ "grad_norm": 0.36263322830200195,
1708
+ "learning_rate": 0.00017848082595870205,
1709
+ "loss": 0.3325,
1710
+ "step": 10500
1711
+ },
1712
+ {
1713
+ "epoch": 23.34108527131783,
1714
+ "grad_norm": 0.8479202389717102,
1715
+ "learning_rate": 0.0001777433628318584,
1716
+ "loss": 0.3265,
1717
+ "step": 10550
1718
+ },
1719
+ {
1720
+ "epoch": 23.451827242524917,
1721
+ "grad_norm": 0.6954662799835205,
1722
+ "learning_rate": 0.00017700589970501474,
1723
+ "loss": 0.3578,
1724
+ "step": 10600
1725
+ },
1726
+ {
1727
+ "epoch": 23.562569213732004,
1728
+ "grad_norm": 0.5398178100585938,
1729
+ "learning_rate": 0.0001762684365781711,
1730
+ "loss": 0.3234,
1731
+ "step": 10650
1732
+ },
1733
+ {
1734
+ "epoch": 23.67331118493909,
1735
+ "grad_norm": 0.6818826794624329,
1736
+ "learning_rate": 0.0001755309734513274,
1737
+ "loss": 0.3287,
1738
+ "step": 10700
1739
+ },
1740
+ {
1741
+ "epoch": 23.78405315614618,
1742
+ "grad_norm": 0.6283251047134399,
1743
+ "learning_rate": 0.00017479351032448378,
1744
+ "loss": 0.339,
1745
+ "step": 10750
1746
+ },
1747
+ {
1748
+ "epoch": 23.894795127353266,
1749
+ "grad_norm": 0.4861604869365692,
1750
+ "learning_rate": 0.0001740560471976401,
1751
+ "loss": 0.3288,
1752
+ "step": 10800
1753
+ },
1754
+ {
1755
+ "epoch": 24.0,
1756
+ "eval_cer": 0.05748018362329666,
1757
+ "eval_loss": 0.2127908617258072,
1758
+ "eval_runtime": 499.441,
1759
+ "eval_samples_per_second": 20.385,
1760
+ "eval_steps_per_second": 2.549,
1761
+ "eval_wer": 0.22348148818683805,
1762
+ "step": 10848
1763
+ },
1764
+ {
1765
+ "epoch": 24.004429678848282,
1766
+ "grad_norm": 0.2947724461555481,
1767
+ "learning_rate": 0.00017331858407079642,
1768
+ "loss": 0.3282,
1769
+ "step": 10850
1770
+ },
1771
+ {
1772
+ "epoch": 24.11517165005537,
1773
+ "grad_norm": 0.4310826063156128,
1774
+ "learning_rate": 0.0001725811209439528,
1775
+ "loss": 0.316,
1776
+ "step": 10900
1777
+ },
1778
+ {
1779
+ "epoch": 24.225913621262457,
1780
+ "grad_norm": 0.8667187094688416,
1781
+ "learning_rate": 0.0001718436578171091,
1782
+ "loss": 0.3381,
1783
+ "step": 10950
1784
+ },
1785
+ {
1786
+ "epoch": 24.336655592469548,
1787
+ "grad_norm": 0.5773921012878418,
1788
+ "learning_rate": 0.00017110619469026548,
1789
+ "loss": 0.3154,
1790
+ "step": 11000
1791
+ },
1792
+ {
1793
+ "epoch": 24.447397563676635,
1794
+ "grad_norm": 0.3089664578437805,
1795
+ "learning_rate": 0.0001703687315634218,
1796
+ "loss": 0.3151,
1797
+ "step": 11050
1798
+ },
1799
+ {
1800
+ "epoch": 24.558139534883722,
1801
+ "grad_norm": 0.5007433891296387,
1802
+ "learning_rate": 0.00016963126843657817,
1803
+ "loss": 0.3229,
1804
+ "step": 11100
1805
+ },
1806
+ {
1807
+ "epoch": 24.66888150609081,
1808
+ "grad_norm": 2.2603423595428467,
1809
+ "learning_rate": 0.0001688938053097345,
1810
+ "loss": 0.3313,
1811
+ "step": 11150
1812
+ },
1813
+ {
1814
+ "epoch": 24.779623477297896,
1815
+ "grad_norm": 0.34253740310668945,
1816
+ "learning_rate": 0.00016815634218289084,
1817
+ "loss": 0.3175,
1818
+ "step": 11200
1819
+ },
1820
+ {
1821
+ "epoch": 24.890365448504983,
1822
+ "grad_norm": 0.4452781081199646,
1823
+ "learning_rate": 0.00016741887905604718,
1824
+ "loss": 0.3382,
1825
+ "step": 11250
1826
+ },
1827
+ {
1828
+ "epoch": 25.0,
1829
+ "grad_norm": 2.1675078868865967,
1830
+ "learning_rate": 0.00016668141592920353,
1831
+ "loss": 0.3192,
1832
+ "step": 11300
1833
+ },
1834
+ {
1835
+ "epoch": 25.0,
1836
+ "eval_cer": 0.056867380757264795,
1837
+ "eval_loss": 0.21508556604385376,
1838
+ "eval_runtime": 501.0298,
1839
+ "eval_samples_per_second": 20.32,
1840
+ "eval_steps_per_second": 2.541,
1841
+ "eval_wer": 0.22150810174708066,
1842
+ "step": 11300
1843
+ },
1844
+ {
1845
+ "epoch": 25.110741971207087,
1846
+ "grad_norm": 1.4607517719268799,
1847
+ "learning_rate": 0.00016594395280235987,
1848
+ "loss": 0.3148,
1849
+ "step": 11350
1850
+ },
1851
+ {
1852
+ "epoch": 25.221483942414174,
1853
+ "grad_norm": 1.1707969903945923,
1854
+ "learning_rate": 0.00016520648967551622,
1855
+ "loss": 0.3314,
1856
+ "step": 11400
1857
+ },
1858
+ {
1859
+ "epoch": 25.33222591362126,
1860
+ "grad_norm": 0.954971969127655,
1861
+ "learning_rate": 0.00016446902654867254,
1862
+ "loss": 0.3206,
1863
+ "step": 11450
1864
+ },
1865
+ {
1866
+ "epoch": 25.44296788482835,
1867
+ "grad_norm": 0.9992377758026123,
1868
+ "learning_rate": 0.0001637315634218289,
1869
+ "loss": 0.3116,
1870
+ "step": 11500
1871
+ },
1872
+ {
1873
+ "epoch": 25.553709856035436,
1874
+ "grad_norm": 0.6818801760673523,
1875
+ "learning_rate": 0.00016299410029498523,
1876
+ "loss": 0.3104,
1877
+ "step": 11550
1878
+ },
1879
+ {
1880
+ "epoch": 25.664451827242527,
1881
+ "grad_norm": 0.8313333988189697,
1882
+ "learning_rate": 0.0001622566371681416,
1883
+ "loss": 0.3047,
1884
+ "step": 11600
1885
+ },
1886
+ {
1887
+ "epoch": 25.775193798449614,
1888
+ "grad_norm": 2.883293390274048,
1889
+ "learning_rate": 0.00016151917404129792,
1890
+ "loss": 0.3036,
1891
+ "step": 11650
1892
+ },
1893
+ {
1894
+ "epoch": 25.8859357696567,
1895
+ "grad_norm": 1.2632169723510742,
1896
+ "learning_rate": 0.00016078171091445426,
1897
+ "loss": 0.3331,
1898
+ "step": 11700
1899
+ },
1900
+ {
1901
+ "epoch": 25.996677740863788,
1902
+ "grad_norm": 0.8442880511283875,
1903
+ "learning_rate": 0.0001600442477876106,
1904
+ "loss": 0.3158,
1905
+ "step": 11750
1906
+ },
1907
+ {
1908
+ "epoch": 26.0,
1909
+ "eval_cer": 0.05608415342553176,
1910
+ "eval_loss": 0.21561041474342346,
1911
+ "eval_runtime": 500.4961,
1912
+ "eval_samples_per_second": 20.342,
1913
+ "eval_steps_per_second": 2.543,
1914
+ "eval_wer": 0.21783289580881687,
1915
+ "step": 11752
1916
+ },
1917
+ {
1918
+ "epoch": 26.106312292358805,
1919
+ "grad_norm": 1.1655757427215576,
1920
+ "learning_rate": 0.00015930678466076695,
1921
+ "loss": 0.3046,
1922
+ "step": 11800
1923
+ },
1924
+ {
1925
+ "epoch": 26.217054263565892,
1926
+ "grad_norm": 1.1917389631271362,
1927
+ "learning_rate": 0.0001585693215339233,
1928
+ "loss": 0.3056,
1929
+ "step": 11850
1930
+ },
1931
+ {
1932
+ "epoch": 26.32779623477298,
1933
+ "grad_norm": 0.7422528862953186,
1934
+ "learning_rate": 0.00015783185840707964,
1935
+ "loss": 0.3326,
1936
+ "step": 11900
1937
+ },
1938
+ {
1939
+ "epoch": 26.438538205980066,
1940
+ "grad_norm": 0.8917913436889648,
1941
+ "learning_rate": 0.00015709439528023596,
1942
+ "loss": 0.3091,
1943
+ "step": 11950
1944
+ },
1945
+ {
1946
+ "epoch": 26.549280177187153,
1947
+ "grad_norm": 1.8631386756896973,
1948
+ "learning_rate": 0.0001563569321533923,
1949
+ "loss": 0.2933,
1950
+ "step": 12000
1951
+ },
1952
+ {
1953
+ "epoch": 26.66002214839424,
1954
+ "grad_norm": 1.4756150245666504,
1955
+ "learning_rate": 0.00015561946902654865,
1956
+ "loss": 0.2987,
1957
+ "step": 12050
1958
+ },
1959
+ {
1960
+ "epoch": 26.770764119601328,
1961
+ "grad_norm": 0.9095970988273621,
1962
+ "learning_rate": 0.000154882005899705,
1963
+ "loss": 0.308,
1964
+ "step": 12100
1965
+ },
1966
+ {
1967
+ "epoch": 26.881506090808415,
1968
+ "grad_norm": 0.768785834312439,
1969
+ "learning_rate": 0.00015414454277286134,
1970
+ "loss": 0.3219,
1971
+ "step": 12150
1972
+ },
1973
+ {
1974
+ "epoch": 26.992248062015506,
1975
+ "grad_norm": 1.0584475994110107,
1976
+ "learning_rate": 0.00015340707964601766,
1977
+ "loss": 0.323,
1978
+ "step": 12200
1979
+ },
1980
+ {
1981
+ "epoch": 27.0,
1982
+ "eval_cer": 0.05621831736661566,
1983
+ "eval_loss": 0.21734623610973358,
1984
+ "eval_runtime": 502.1183,
1985
+ "eval_samples_per_second": 20.276,
1986
+ "eval_steps_per_second": 2.535,
1987
+ "eval_wer": 0.21875622340907033,
1988
+ "step": 12204
1989
+ },
1990
+ {
1991
+ "epoch": 27.101882613510522,
1992
+ "grad_norm": 0.8777967691421509,
1993
+ "learning_rate": 0.00015266961651917404,
1994
+ "loss": 0.3043,
1995
+ "step": 12250
1996
+ },
1997
+ {
1998
+ "epoch": 27.21262458471761,
1999
+ "grad_norm": 0.6613890528678894,
2000
+ "learning_rate": 0.00015193215339233035,
2001
+ "loss": 0.3055,
2002
+ "step": 12300
2003
+ },
2004
+ {
2005
+ "epoch": 27.323366555924697,
2006
+ "grad_norm": 1.175444483757019,
2007
+ "learning_rate": 0.00015119469026548673,
2008
+ "loss": 0.3027,
2009
+ "step": 12350
2010
+ },
2011
+ {
2012
+ "epoch": 27.434108527131784,
2013
+ "grad_norm": 0.9431079626083374,
2014
+ "learning_rate": 0.00015045722713864304,
2015
+ "loss": 0.2968,
2016
+ "step": 12400
2017
+ },
2018
+ {
2019
+ "epoch": 27.54485049833887,
2020
+ "grad_norm": 0.8549213409423828,
2021
+ "learning_rate": 0.0001497197640117994,
2022
+ "loss": 0.2959,
2023
+ "step": 12450
2024
+ },
2025
+ {
2026
+ "epoch": 27.655592469545958,
2027
+ "grad_norm": 2.223446846008301,
2028
+ "learning_rate": 0.00014898230088495574,
2029
+ "loss": 0.3125,
2030
+ "step": 12500
2031
+ },
2032
+ {
2033
+ "epoch": 27.766334440753045,
2034
+ "grad_norm": 2.218383312225342,
2035
+ "learning_rate": 0.00014824483775811208,
2036
+ "loss": 0.307,
2037
+ "step": 12550
2038
+ },
2039
+ {
2040
+ "epoch": 27.877076411960132,
2041
+ "grad_norm": 12.813185691833496,
2042
+ "learning_rate": 0.00014750737463126843,
2043
+ "loss": 0.3076,
2044
+ "step": 12600
2045
+ },
2046
+ {
2047
+ "epoch": 27.98781838316722,
2048
+ "grad_norm": 0.6163749694824219,
2049
+ "learning_rate": 0.00014676991150442477,
2050
+ "loss": 0.3328,
2051
+ "step": 12650
2052
+ },
2053
+ {
2054
+ "epoch": 28.0,
2055
+ "eval_cer": 0.054448803765292875,
2056
+ "eval_loss": 0.2115621566772461,
2057
+ "eval_runtime": 500.9948,
2058
+ "eval_samples_per_second": 20.322,
2059
+ "eval_steps_per_second": 2.541,
2060
+ "eval_wer": 0.21100751335204126,
2061
+ "step": 12656
2062
+ },
2063
+ {
2064
+ "epoch": 28.097452934662236,
2065
+ "grad_norm": 1.2707273960113525,
2066
+ "learning_rate": 0.0001460324483775811,
2067
+ "loss": 0.3012,
2068
+ "step": 12700
2069
+ },
2070
+ {
2071
+ "epoch": 28.208194905869323,
2072
+ "grad_norm": 1.3631172180175781,
2073
+ "learning_rate": 0.00014529498525073744,
2074
+ "loss": 0.2988,
2075
+ "step": 12750
2076
+ },
2077
+ {
2078
+ "epoch": 28.31893687707641,
2079
+ "grad_norm": 1.3375684022903442,
2080
+ "learning_rate": 0.00014455752212389378,
2081
+ "loss": 0.307,
2082
+ "step": 12800
2083
+ },
2084
+ {
2085
+ "epoch": 28.4296788482835,
2086
+ "grad_norm": 1.4623414278030396,
2087
+ "learning_rate": 0.00014382005899705013,
2088
+ "loss": 0.3189,
2089
+ "step": 12850
2090
+ },
2091
+ {
2092
+ "epoch": 28.54042081949059,
2093
+ "grad_norm": 0.9495463967323303,
2094
+ "learning_rate": 0.00014308259587020647,
2095
+ "loss": 0.2902,
2096
+ "step": 12900
2097
+ },
2098
+ {
2099
+ "epoch": 28.651162790697676,
2100
+ "grad_norm": 0.6960974931716919,
2101
+ "learning_rate": 0.00014234513274336282,
2102
+ "loss": 0.3025,
2103
+ "step": 12950
2104
+ },
2105
+ {
2106
+ "epoch": 28.761904761904763,
2107
+ "grad_norm": 1.6524468660354614,
2108
+ "learning_rate": 0.00014160766961651916,
2109
+ "loss": 0.3054,
2110
+ "step": 13000
2111
+ },
2112
+ {
2113
+ "epoch": 28.87264673311185,
2114
+ "grad_norm": 0.7738968729972839,
2115
+ "learning_rate": 0.0001408702064896755,
2116
+ "loss": 0.3023,
2117
+ "step": 13050
2118
+ },
2119
+ {
2120
+ "epoch": 28.983388704318937,
2121
+ "grad_norm": 1.094317078590393,
2122
+ "learning_rate": 0.00014013274336283185,
2123
+ "loss": 0.3171,
2124
+ "step": 13100
2125
+ },
2126
+ {
2127
+ "epoch": 29.0,
2128
+ "eval_cer": 0.054949199005011204,
2129
+ "eval_loss": 0.21789278090000153,
2130
+ "eval_runtime": 503.5992,
2131
+ "eval_samples_per_second": 20.216,
2132
+ "eval_steps_per_second": 2.528,
2133
+ "eval_wer": 0.21321625780754955,
2134
+ "step": 13108
2135
+ },
2136
+ {
2137
+ "epoch": 29.093023255813954,
2138
+ "grad_norm": 0.723978579044342,
2139
+ "learning_rate": 0.0001393952802359882,
2140
+ "loss": 0.2999,
2141
+ "step": 13150
2142
+ },
2143
+ {
2144
+ "epoch": 29.20376522702104,
2145
+ "grad_norm": 0.9239691495895386,
2146
+ "learning_rate": 0.00013865781710914454,
2147
+ "loss": 0.2868,
2148
+ "step": 13200
2149
+ },
2150
+ {
2151
+ "epoch": 29.314507198228128,
2152
+ "grad_norm": 0.812727689743042,
2153
+ "learning_rate": 0.00013792035398230086,
2154
+ "loss": 0.2994,
2155
+ "step": 13250
2156
+ },
2157
+ {
2158
+ "epoch": 29.425249169435215,
2159
+ "grad_norm": 0.6128324866294861,
2160
+ "learning_rate": 0.0001371828908554572,
2161
+ "loss": 0.2986,
2162
+ "step": 13300
2163
+ },
2164
+ {
2165
+ "epoch": 29.535991140642302,
2166
+ "grad_norm": 0.9569301009178162,
2167
+ "learning_rate": 0.00013644542772861355,
2168
+ "loss": 0.309,
2169
+ "step": 13350
2170
+ },
2171
+ {
2172
+ "epoch": 29.64673311184939,
2173
+ "grad_norm": 1.0407278537750244,
2174
+ "learning_rate": 0.0001357079646017699,
2175
+ "loss": 0.2961,
2176
+ "step": 13400
2177
+ },
2178
+ {
2179
+ "epoch": 29.757475083056477,
2180
+ "grad_norm": 1.2411153316497803,
2181
+ "learning_rate": 0.00013497050147492624,
2182
+ "loss": 0.2863,
2183
+ "step": 13450
2184
+ },
2185
+ {
2186
+ "epoch": 29.868217054263567,
2187
+ "grad_norm": 1.976034164428711,
2188
+ "learning_rate": 0.0001342330383480826,
2189
+ "loss": 0.3,
2190
+ "step": 13500
2191
+ },
2192
+ {
2193
+ "epoch": 29.978959025470655,
2194
+ "grad_norm": 0.6932374238967896,
2195
+ "learning_rate": 0.00013349557522123894,
2196
+ "loss": 0.3038,
2197
+ "step": 13550
2198
+ },
2199
+ {
2200
+ "epoch": 30.0,
2201
+ "eval_cer": 0.0540898245715819,
2202
+ "eval_loss": 0.21122422814369202,
2203
+ "eval_runtime": 502.738,
2204
+ "eval_samples_per_second": 20.251,
2205
+ "eval_steps_per_second": 2.532,
2206
+ "eval_wer": 0.21066352855978998,
2207
+ "step": 13560
2208
+ },
2209
+ {
2210
+ "epoch": 30.08859357696567,
2211
+ "grad_norm": 0.7096422910690308,
2212
+ "learning_rate": 0.00013275811209439528,
2213
+ "loss": 0.2918,
2214
+ "step": 13600
2215
+ },
2216
+ {
2217
+ "epoch": 30.19933554817276,
2218
+ "grad_norm": 1.9430056810379028,
2219
+ "learning_rate": 0.00013202064896755163,
2220
+ "loss": 0.2878,
2221
+ "step": 13650
2222
+ },
2223
+ {
2224
+ "epoch": 30.310077519379846,
2225
+ "grad_norm": 2.081293821334839,
2226
+ "learning_rate": 0.00013128318584070797,
2227
+ "loss": 0.2791,
2228
+ "step": 13700
2229
+ },
2230
+ {
2231
+ "epoch": 30.420819490586933,
2232
+ "grad_norm": 1.2402158975601196,
2233
+ "learning_rate": 0.0001305457227138643,
2234
+ "loss": 0.2833,
2235
+ "step": 13750
2236
+ },
2237
+ {
2238
+ "epoch": 30.53156146179402,
2239
+ "grad_norm": 0.9645398855209351,
2240
+ "learning_rate": 0.00012980825958702064,
2241
+ "loss": 0.2828,
2242
+ "step": 13800
2243
+ },
2244
+ {
2245
+ "epoch": 30.642303433001107,
2246
+ "grad_norm": 0.6930665969848633,
2247
+ "learning_rate": 0.00012907079646017698,
2248
+ "loss": 0.3018,
2249
+ "step": 13850
2250
+ },
2251
+ {
2252
+ "epoch": 30.753045404208194,
2253
+ "grad_norm": 0.9082468748092651,
2254
+ "learning_rate": 0.00012833333333333333,
2255
+ "loss": 0.3037,
2256
+ "step": 13900
2257
+ },
2258
+ {
2259
+ "epoch": 30.86378737541528,
2260
+ "grad_norm": 0.7977900505065918,
2261
+ "learning_rate": 0.00012759587020648967,
2262
+ "loss": 0.3122,
2263
+ "step": 13950
2264
+ },
2265
+ {
2266
+ "epoch": 30.97452934662237,
2267
+ "grad_norm": 0.7910751104354858,
2268
+ "learning_rate": 0.000126858407079646,
2269
+ "loss": 0.2927,
2270
+ "step": 14000
2271
+ },
2272
+ {
2273
+ "epoch": 31.0,
2274
+ "eval_cer": 0.053854131161569646,
2275
+ "eval_loss": 0.21427859365940094,
2276
+ "eval_runtime": 504.444,
2277
+ "eval_samples_per_second": 20.183,
2278
+ "eval_steps_per_second": 2.524,
2279
+ "eval_wer": 0.2105549017832896,
2280
+ "step": 14012
2281
+ },
2282
+ {
2283
+ "epoch": 31.084163898117385,
2284
+ "grad_norm": 1.2776386737823486,
2285
+ "learning_rate": 0.00012612094395280234,
2286
+ "loss": 0.2853,
2287
+ "step": 14050
2288
+ },
2289
+ {
2290
+ "epoch": 31.194905869324472,
2291
+ "grad_norm": 1.0792760848999023,
2292
+ "learning_rate": 0.00012538348082595868,
2293
+ "loss": 0.2994,
2294
+ "step": 14100
2295
+ },
2296
+ {
2297
+ "epoch": 31.305647840531563,
2298
+ "grad_norm": 1.4567831754684448,
2299
+ "learning_rate": 0.00012464601769911503,
2300
+ "loss": 0.3088,
2301
+ "step": 14150
2302
+ },
2303
+ {
2304
+ "epoch": 31.41638981173865,
2305
+ "grad_norm": 1.4427411556243896,
2306
+ "learning_rate": 0.00012390855457227137,
2307
+ "loss": 0.3066,
2308
+ "step": 14200
2309
+ },
2310
+ {
2311
+ "epoch": 31.527131782945737,
2312
+ "grad_norm": 0.8337031006813049,
2313
+ "learning_rate": 0.00012317109144542772,
2314
+ "loss": 0.2906,
2315
+ "step": 14250
2316
+ },
2317
+ {
2318
+ "epoch": 31.637873754152825,
2319
+ "grad_norm": 0.5533416271209717,
2320
+ "learning_rate": 0.00012243362831858406,
2321
+ "loss": 0.2885,
2322
+ "step": 14300
2323
+ },
2324
+ {
2325
+ "epoch": 31.74861572535991,
2326
+ "grad_norm": 1.1606074571609497,
2327
+ "learning_rate": 0.00012169616519174041,
2328
+ "loss": 0.2802,
2329
+ "step": 14350
2330
+ },
2331
+ {
2332
+ "epoch": 31.859357696567,
2333
+ "grad_norm": 0.7705062627792358,
2334
+ "learning_rate": 0.00012095870206489674,
2335
+ "loss": 0.2766,
2336
+ "step": 14400
2337
+ },
2338
+ {
2339
+ "epoch": 31.970099667774086,
2340
+ "grad_norm": 1.1826943159103394,
2341
+ "learning_rate": 0.00012022123893805309,
2342
+ "loss": 0.2866,
2343
+ "step": 14450
2344
+ },
2345
+ {
2346
+ "epoch": 32.0,
2347
+ "eval_cer": 0.05413333720112263,
2348
+ "eval_loss": 0.2068941295146942,
2349
+ "eval_runtime": 502.2625,
2350
+ "eval_samples_per_second": 20.27,
2351
+ "eval_steps_per_second": 2.535,
2352
+ "eval_wer": 0.21153254277179326,
2353
+ "step": 14464
2354
+ },
2355
+ {
2356
+ "epoch": 32.079734219269106,
2357
+ "grad_norm": 0.8447093963623047,
2358
+ "learning_rate": 0.00011948377581120943,
2359
+ "loss": 0.2951,
2360
+ "step": 14500
2361
+ },
2362
+ {
2363
+ "epoch": 32.19047619047619,
2364
+ "grad_norm": 1.153529167175293,
2365
+ "learning_rate": 0.00011874631268436578,
2366
+ "loss": 0.2881,
2367
+ "step": 14550
2368
+ },
2369
+ {
2370
+ "epoch": 32.30121816168328,
2371
+ "grad_norm": 0.9775529503822327,
2372
+ "learning_rate": 0.00011800884955752212,
2373
+ "loss": 0.277,
2374
+ "step": 14600
2375
+ },
2376
+ {
2377
+ "epoch": 32.411960132890364,
2378
+ "grad_norm": 1.1522144079208374,
2379
+ "learning_rate": 0.00011727138643067845,
2380
+ "loss": 0.2783,
2381
+ "step": 14650
2382
+ },
2383
+ {
2384
+ "epoch": 32.522702104097455,
2385
+ "grad_norm": 0.7744702696800232,
2386
+ "learning_rate": 0.0001165339233038348,
2387
+ "loss": 0.2844,
2388
+ "step": 14700
2389
+ },
2390
+ {
2391
+ "epoch": 32.63344407530454,
2392
+ "grad_norm": 1.2563271522521973,
2393
+ "learning_rate": 0.00011579646017699114,
2394
+ "loss": 0.2768,
2395
+ "step": 14750
2396
+ },
2397
+ {
2398
+ "epoch": 32.74418604651163,
2399
+ "grad_norm": 0.5680158734321594,
2400
+ "learning_rate": 0.00011505899705014749,
2401
+ "loss": 0.2723,
2402
+ "step": 14800
2403
+ },
2404
+ {
2405
+ "epoch": 32.85492801771871,
2406
+ "grad_norm": 1.74644136428833,
2407
+ "learning_rate": 0.00011432153392330384,
2408
+ "loss": 0.2922,
2409
+ "step": 14850
2410
+ },
2411
+ {
2412
+ "epoch": 32.965669988925804,
2413
+ "grad_norm": 0.6420773863792419,
2414
+ "learning_rate": 0.00011358407079646018,
2415
+ "loss": 0.2988,
2416
+ "step": 14900
2417
+ },
2418
+ {
2419
+ "epoch": 33.0,
2420
+ "eval_cer": 0.05228767649810358,
2421
+ "eval_loss": 0.20571845769882202,
2422
+ "eval_runtime": 505.3192,
2423
+ "eval_samples_per_second": 20.148,
2424
+ "eval_steps_per_second": 2.519,
2425
+ "eval_wer": 0.2027699828007604,
2426
+ "step": 14916
2427
+ },
2428
+ {
2429
+ "epoch": 33.07530454042082,
2430
+ "grad_norm": 1.2276347875595093,
2431
+ "learning_rate": 0.0001128466076696165,
2432
+ "loss": 0.2973,
2433
+ "step": 14950
2434
+ },
2435
+ {
2436
+ "epoch": 33.18604651162791,
2437
+ "grad_norm": 1.1782604455947876,
2438
+ "learning_rate": 0.00011210914454277285,
2439
+ "loss": 0.2867,
2440
+ "step": 15000
2441
+ },
2442
+ {
2443
+ "epoch": 33.296788482835,
2444
+ "grad_norm": 1.052660346031189,
2445
+ "learning_rate": 0.00011137168141592919,
2446
+ "loss": 0.2911,
2447
+ "step": 15050
2448
+ },
2449
+ {
2450
+ "epoch": 33.40753045404208,
2451
+ "grad_norm": 0.8341677188873291,
2452
+ "learning_rate": 0.00011063421828908554,
2453
+ "loss": 0.2822,
2454
+ "step": 15100
2455
+ },
2456
+ {
2457
+ "epoch": 33.51827242524917,
2458
+ "grad_norm": 0.6574891209602356,
2459
+ "learning_rate": 0.00010989675516224187,
2460
+ "loss": 0.2734,
2461
+ "step": 15150
2462
+ },
2463
+ {
2464
+ "epoch": 33.629014396456256,
2465
+ "grad_norm": 0.5809639096260071,
2466
+ "learning_rate": 0.00010915929203539821,
2467
+ "loss": 0.2897,
2468
+ "step": 15200
2469
+ },
2470
+ {
2471
+ "epoch": 33.73975636766335,
2472
+ "grad_norm": 0.6616008281707764,
2473
+ "learning_rate": 0.00010842182890855456,
2474
+ "loss": 0.2716,
2475
+ "step": 15250
2476
+ },
2477
+ {
2478
+ "epoch": 33.85049833887043,
2479
+ "grad_norm": 1.0783253908157349,
2480
+ "learning_rate": 0.0001076843657817109,
2481
+ "loss": 0.3086,
2482
+ "step": 15300
2483
+ },
2484
+ {
2485
+ "epoch": 33.96124031007752,
2486
+ "grad_norm": 1.0111535787582397,
2487
+ "learning_rate": 0.00010694690265486725,
2488
+ "loss": 0.276,
2489
+ "step": 15350
2490
+ },
2491
+ {
2492
+ "epoch": 34.0,
2493
+ "eval_cer": 0.05319781566599706,
2494
+ "eval_loss": 0.20418909192085266,
2495
+ "eval_runtime": 502.8988,
2496
+ "eval_samples_per_second": 20.245,
2497
+ "eval_steps_per_second": 2.531,
2498
+ "eval_wer": 0.20802027699828007,
2499
+ "step": 15368
2500
+ },
2501
+ {
2502
+ "epoch": 34.070874861572534,
2503
+ "grad_norm": 0.7386752963066101,
2504
+ "learning_rate": 0.00010620943952802358,
2505
+ "loss": 0.2782,
2506
+ "step": 15400
2507
+ },
2508
+ {
2509
+ "epoch": 34.181616832779625,
2510
+ "grad_norm": 0.9310299754142761,
2511
+ "learning_rate": 0.00010547197640117993,
2512
+ "loss": 0.2745,
2513
+ "step": 15450
2514
+ },
2515
+ {
2516
+ "epoch": 34.29235880398671,
2517
+ "grad_norm": 1.1085954904556274,
2518
+ "learning_rate": 0.00010473451327433627,
2519
+ "loss": 0.265,
2520
+ "step": 15500
2521
+ },
2522
+ {
2523
+ "epoch": 34.4031007751938,
2524
+ "grad_norm": 0.7299510836601257,
2525
+ "learning_rate": 0.00010399705014749262,
2526
+ "loss": 0.2858,
2527
+ "step": 15550
2528
+ },
2529
+ {
2530
+ "epoch": 34.51384274640088,
2531
+ "grad_norm": 1.8487365245819092,
2532
+ "learning_rate": 0.00010325958702064896,
2533
+ "loss": 0.2941,
2534
+ "step": 15600
2535
+ },
2536
+ {
2537
+ "epoch": 34.62458471760797,
2538
+ "grad_norm": 1.3170086145401,
2539
+ "learning_rate": 0.00010252212389380531,
2540
+ "loss": 0.2873,
2541
+ "step": 15650
2542
+ },
2543
+ {
2544
+ "epoch": 34.735326688815064,
2545
+ "grad_norm": 0.7510431408882141,
2546
+ "learning_rate": 0.00010178466076696164,
2547
+ "loss": 0.2651,
2548
+ "step": 15700
2549
+ },
2550
+ {
2551
+ "epoch": 34.84606866002215,
2552
+ "grad_norm": 0.8078789710998535,
2553
+ "learning_rate": 0.00010104719764011799,
2554
+ "loss": 0.26,
2555
+ "step": 15750
2556
+ },
2557
+ {
2558
+ "epoch": 34.95681063122924,
2559
+ "grad_norm": 0.6926655769348145,
2560
+ "learning_rate": 0.00010030973451327433,
2561
+ "loss": 0.2757,
2562
+ "step": 15800
2563
+ },
2564
+ {
2565
+ "epoch": 35.0,
2566
+ "eval_cer": 0.051932323356854326,
2567
+ "eval_loss": 0.20506548881530762,
2568
+ "eval_runtime": 504.012,
2569
+ "eval_samples_per_second": 20.2,
2570
+ "eval_steps_per_second": 2.526,
2571
+ "eval_wer": 0.20217253553000814,
2572
+ "step": 15820
2573
+ },
2574
+ {
2575
+ "epoch": 35.06644518272425,
2576
+ "grad_norm": 0.6662414073944092,
2577
+ "learning_rate": 9.957227138643068e-05,
2578
+ "loss": 0.2745,
2579
+ "step": 15850
2580
+ },
2581
+ {
2582
+ "epoch": 35.17718715393134,
2583
+ "grad_norm": 0.7030307054519653,
2584
+ "learning_rate": 9.883480825958702e-05,
2585
+ "loss": 0.2773,
2586
+ "step": 15900
2587
+ },
2588
+ {
2589
+ "epoch": 35.287929125138426,
2590
+ "grad_norm": 1.443966031074524,
2591
+ "learning_rate": 9.809734513274335e-05,
2592
+ "loss": 0.2666,
2593
+ "step": 15950
2594
+ },
2595
+ {
2596
+ "epoch": 35.39867109634552,
2597
+ "grad_norm": 0.5570212602615356,
2598
+ "learning_rate": 9.73598820058997e-05,
2599
+ "loss": 0.2723,
2600
+ "step": 16000
2601
+ },
2602
+ {
2603
+ "epoch": 35.5094130675526,
2604
+ "grad_norm": 0.9807683229446411,
2605
+ "learning_rate": 9.662241887905605e-05,
2606
+ "loss": 0.2675,
2607
+ "step": 16050
2608
+ },
2609
+ {
2610
+ "epoch": 35.62015503875969,
2611
+ "grad_norm": 0.8136458992958069,
2612
+ "learning_rate": 9.588495575221239e-05,
2613
+ "loss": 0.2754,
2614
+ "step": 16100
2615
+ },
2616
+ {
2617
+ "epoch": 35.730897009966775,
2618
+ "grad_norm": 1.0001354217529297,
2619
+ "learning_rate": 9.514749262536871e-05,
2620
+ "loss": 0.2678,
2621
+ "step": 16150
2622
+ },
2623
+ {
2624
+ "epoch": 35.841638981173865,
2625
+ "grad_norm": 1.3988227844238281,
2626
+ "learning_rate": 9.441002949852505e-05,
2627
+ "loss": 0.2751,
2628
+ "step": 16200
2629
+ },
2630
+ {
2631
+ "epoch": 35.95238095238095,
2632
+ "grad_norm": 0.6191902756690979,
2633
+ "learning_rate": 9.36725663716814e-05,
2634
+ "loss": 0.2853,
2635
+ "step": 16250
2636
+ },
2637
+ {
2638
+ "epoch": 36.0,
2639
+ "eval_cer": 0.052490735435960285,
2640
+ "eval_loss": 0.20767323672771454,
2641
+ "eval_runtime": 504.3578,
2642
+ "eval_samples_per_second": 20.186,
2643
+ "eval_steps_per_second": 2.524,
2644
+ "eval_wer": 0.2030415497420114,
2645
+ "step": 16272
2646
+ },
2647
+ {
2648
+ "epoch": 36.06201550387597,
2649
+ "grad_norm": 0.6683135628700256,
2650
+ "learning_rate": 9.293510324483775e-05,
2651
+ "loss": 0.2632,
2652
+ "step": 16300
2653
+ },
2654
+ {
2655
+ "epoch": 36.17275747508306,
2656
+ "grad_norm": 0.541814923286438,
2657
+ "learning_rate": 9.219764011799409e-05,
2658
+ "loss": 0.2948,
2659
+ "step": 16350
2660
+ },
2661
+ {
2662
+ "epoch": 36.28349944629014,
2663
+ "grad_norm": 0.746822714805603,
2664
+ "learning_rate": 9.146017699115044e-05,
2665
+ "loss": 0.2708,
2666
+ "step": 16400
2667
+ },
2668
+ {
2669
+ "epoch": 36.394241417497234,
2670
+ "grad_norm": 0.9451239109039307,
2671
+ "learning_rate": 9.072271386430677e-05,
2672
+ "loss": 0.2697,
2673
+ "step": 16450
2674
+ },
2675
+ {
2676
+ "epoch": 36.50498338870432,
2677
+ "grad_norm": 2.489182233810425,
2678
+ "learning_rate": 8.998525073746311e-05,
2679
+ "loss": 0.2739,
2680
+ "step": 16500
2681
+ },
2682
+ {
2683
+ "epoch": 36.61572535991141,
2684
+ "grad_norm": 0.8976282477378845,
2685
+ "learning_rate": 8.924778761061946e-05,
2686
+ "loss": 0.2649,
2687
+ "step": 16550
2688
+ },
2689
+ {
2690
+ "epoch": 36.72646733111849,
2691
+ "grad_norm": 1.0089658498764038,
2692
+ "learning_rate": 8.85103244837758e-05,
2693
+ "loss": 0.2661,
2694
+ "step": 16600
2695
+ },
2696
+ {
2697
+ "epoch": 36.83720930232558,
2698
+ "grad_norm": 0.7189285755157471,
2699
+ "learning_rate": 8.777286135693215e-05,
2700
+ "loss": 0.2659,
2701
+ "step": 16650
2702
+ },
2703
+ {
2704
+ "epoch": 36.94795127353267,
2705
+ "grad_norm": 1.8182567358016968,
2706
+ "learning_rate": 8.703539823008848e-05,
2707
+ "loss": 0.2785,
2708
+ "step": 16700
2709
+ },
2710
+ {
2711
+ "epoch": 37.0,
2712
+ "eval_cer": 0.051094705238195384,
2713
+ "eval_loss": 0.20127856731414795,
2714
+ "eval_runtime": 506.5771,
2715
+ "eval_samples_per_second": 20.098,
2716
+ "eval_steps_per_second": 2.513,
2717
+ "eval_wer": 0.19960170181949852,
2718
+ "step": 16724
2719
+ },
2720
+ {
2721
+ "epoch": 37.05758582502769,
2722
+ "grad_norm": 0.747435450553894,
2723
+ "learning_rate": 8.629793510324483e-05,
2724
+ "loss": 0.2703,
2725
+ "step": 16750
2726
+ },
2727
+ {
2728
+ "epoch": 37.16832779623477,
2729
+ "grad_norm": 0.985905110836029,
2730
+ "learning_rate": 8.556047197640117e-05,
2731
+ "loss": 0.2638,
2732
+ "step": 16800
2733
+ },
2734
+ {
2735
+ "epoch": 37.27906976744186,
2736
+ "grad_norm": 0.4967564642429352,
2737
+ "learning_rate": 8.482300884955752e-05,
2738
+ "loss": 0.2574,
2739
+ "step": 16850
2740
+ },
2741
+ {
2742
+ "epoch": 37.389811738648945,
2743
+ "grad_norm": 3.6833720207214355,
2744
+ "learning_rate": 8.408554572271386e-05,
2745
+ "loss": 0.2663,
2746
+ "step": 16900
2747
+ },
2748
+ {
2749
+ "epoch": 37.500553709856035,
2750
+ "grad_norm": 0.45124557614326477,
2751
+ "learning_rate": 8.334808259587021e-05,
2752
+ "loss": 0.2625,
2753
+ "step": 16950
2754
+ },
2755
+ {
2756
+ "epoch": 37.611295681063126,
2757
+ "grad_norm": 0.6523784399032593,
2758
+ "learning_rate": 8.261061946902654e-05,
2759
+ "loss": 0.2652,
2760
+ "step": 17000
2761
+ },
2762
+ {
2763
+ "epoch": 37.72203765227021,
2764
+ "grad_norm": 0.9279195070266724,
2765
+ "learning_rate": 8.187315634218289e-05,
2766
+ "loss": 0.2632,
2767
+ "step": 17050
2768
+ },
2769
+ {
2770
+ "epoch": 37.8327796234773,
2771
+ "grad_norm": 2.810720205307007,
2772
+ "learning_rate": 8.113569321533923e-05,
2773
+ "loss": 0.2673,
2774
+ "step": 17100
2775
+ },
2776
+ {
2777
+ "epoch": 37.943521594684384,
2778
+ "grad_norm": 0.7658355832099915,
2779
+ "learning_rate": 8.039823008849558e-05,
2780
+ "loss": 0.2744,
2781
+ "step": 17150
2782
+ },
2783
+ {
2784
+ "epoch": 38.0,
2785
+ "eval_cer": 0.050583431841091875,
2786
+ "eval_loss": 0.19885742664337158,
2787
+ "eval_runtime": 502.4471,
2788
+ "eval_samples_per_second": 20.263,
2789
+ "eval_steps_per_second": 2.534,
2790
+ "eval_wer": 0.19665067439123743,
2791
+ "step": 17176
2792
+ },
2793
+ {
2794
+ "epoch": 38.053156146179404,
2795
+ "grad_norm": 0.9786189198493958,
2796
+ "learning_rate": 7.966076696165192e-05,
2797
+ "loss": 0.2677,
2798
+ "step": 17200
2799
+ },
2800
+ {
2801
+ "epoch": 38.16389811738649,
2802
+ "grad_norm": 0.7271732091903687,
2803
+ "learning_rate": 7.892330383480825e-05,
2804
+ "loss": 0.2758,
2805
+ "step": 17250
2806
+ },
2807
+ {
2808
+ "epoch": 38.27464008859358,
2809
+ "grad_norm": 1.0245264768600464,
2810
+ "learning_rate": 7.81858407079646e-05,
2811
+ "loss": 0.2769,
2812
+ "step": 17300
2813
+ },
2814
+ {
2815
+ "epoch": 38.38538205980066,
2816
+ "grad_norm": 0.5469414591789246,
2817
+ "learning_rate": 7.744837758112093e-05,
2818
+ "loss": 0.2946,
2819
+ "step": 17350
2820
+ },
2821
+ {
2822
+ "epoch": 38.49612403100775,
2823
+ "grad_norm": 1.3482476472854614,
2824
+ "learning_rate": 7.671091445427728e-05,
2825
+ "loss": 0.2756,
2826
+ "step": 17400
2827
+ },
2828
+ {
2829
+ "epoch": 38.606866002214836,
2830
+ "grad_norm": 1.6001390218734741,
2831
+ "learning_rate": 7.597345132743361e-05,
2832
+ "loss": 0.2797,
2833
+ "step": 17450
2834
+ },
2835
+ {
2836
+ "epoch": 38.71760797342193,
2837
+ "grad_norm": 0.7820600271224976,
2838
+ "learning_rate": 7.523598820058995e-05,
2839
+ "loss": 0.2655,
2840
+ "step": 17500
2841
+ },
2842
+ {
2843
+ "epoch": 38.82834994462901,
2844
+ "grad_norm": 1.1540759801864624,
2845
+ "learning_rate": 7.449852507374631e-05,
2846
+ "loss": 0.2672,
2847
+ "step": 17550
2848
+ },
2849
+ {
2850
+ "epoch": 38.9390919158361,
2851
+ "grad_norm": 0.48907431960105896,
2852
+ "learning_rate": 7.376106194690265e-05,
2853
+ "loss": 0.258,
2854
+ "step": 17600
2855
+ },
2856
+ {
2857
+ "epoch": 39.0,
2858
+ "eval_cer": 0.0508408815658745,
2859
+ "eval_loss": 0.2034897357225418,
2860
+ "eval_runtime": 505.9396,
2861
+ "eval_samples_per_second": 20.123,
2862
+ "eval_steps_per_second": 2.516,
2863
+ "eval_wer": 0.19721191273648955,
2864
+ "step": 17628
2865
+ },
2866
+ {
2867
+ "epoch": 39.04872646733112,
2868
+ "grad_norm": 0.5294991135597229,
2869
+ "learning_rate": 7.302359882005899e-05,
2870
+ "loss": 0.2743,
2871
+ "step": 17650
2872
+ },
2873
+ {
2874
+ "epoch": 39.159468438538205,
2875
+ "grad_norm": 0.49712470173835754,
2876
+ "learning_rate": 7.228613569321534e-05,
2877
+ "loss": 0.2649,
2878
+ "step": 17700
2879
+ },
2880
+ {
2881
+ "epoch": 39.270210409745296,
2882
+ "grad_norm": 1.055932641029358,
2883
+ "learning_rate": 7.154867256637167e-05,
2884
+ "loss": 0.2804,
2885
+ "step": 17750
2886
+ },
2887
+ {
2888
+ "epoch": 39.38095238095238,
2889
+ "grad_norm": 6.277849197387695,
2890
+ "learning_rate": 7.081120943952801e-05,
2891
+ "loss": 0.2576,
2892
+ "step": 17800
2893
+ },
2894
+ {
2895
+ "epoch": 39.49169435215947,
2896
+ "grad_norm": 0.6586483716964722,
2897
+ "learning_rate": 7.007374631268436e-05,
2898
+ "loss": 0.2789,
2899
+ "step": 17850
2900
+ },
2901
+ {
2902
+ "epoch": 39.602436323366554,
2903
+ "grad_norm": 0.6979815363883972,
2904
+ "learning_rate": 6.93362831858407e-05,
2905
+ "loss": 0.2625,
2906
+ "step": 17900
2907
+ },
2908
+ {
2909
+ "epoch": 39.713178294573645,
2910
+ "grad_norm": 0.6091492176055908,
2911
+ "learning_rate": 6.859882005899705e-05,
2912
+ "loss": 0.2645,
2913
+ "step": 17950
2914
+ },
2915
+ {
2916
+ "epoch": 39.82392026578073,
2917
+ "grad_norm": 0.5214503407478333,
2918
+ "learning_rate": 6.786135693215338e-05,
2919
+ "loss": 0.2578,
2920
+ "step": 18000
2921
+ },
2922
+ {
2923
+ "epoch": 39.93466223698782,
2924
+ "grad_norm": 2.7882440090179443,
2925
+ "learning_rate": 6.712389380530973e-05,
2926
+ "loss": 0.2709,
2927
+ "step": 18050
2928
+ },
2929
+ {
2930
+ "epoch": 40.0,
2931
+ "eval_cer": 0.051054818661116386,
2932
+ "eval_loss": 0.20327427983283997,
2933
+ "eval_runtime": 506.6892,
2934
+ "eval_samples_per_second": 20.093,
2935
+ "eval_steps_per_second": 2.512,
2936
+ "eval_wer": 0.19846112066624422,
2937
+ "step": 18080
2938
+ },
2939
+ {
2940
+ "epoch": 40.04429678848283,
2941
+ "grad_norm": 0.5243245363235474,
2942
+ "learning_rate": 6.638643067846607e-05,
2943
+ "loss": 0.2726,
2944
+ "step": 18100
2945
+ },
2946
+ {
2947
+ "epoch": 40.15503875968992,
2948
+ "grad_norm": 2.323392152786255,
2949
+ "learning_rate": 6.564896755162242e-05,
2950
+ "loss": 0.2977,
2951
+ "step": 18150
2952
+ },
2953
+ {
2954
+ "epoch": 40.26578073089701,
2955
+ "grad_norm": 0.7332302927970886,
2956
+ "learning_rate": 6.491150442477875e-05,
2957
+ "loss": 0.2459,
2958
+ "step": 18200
2959
+ },
2960
+ {
2961
+ "epoch": 40.3765227021041,
2962
+ "grad_norm": 0.7128420472145081,
2963
+ "learning_rate": 6.41740412979351e-05,
2964
+ "loss": 0.254,
2965
+ "step": 18250
2966
+ },
2967
+ {
2968
+ "epoch": 40.48726467331119,
2969
+ "grad_norm": 0.6172935366630554,
2970
+ "learning_rate": 6.343657817109144e-05,
2971
+ "loss": 0.271,
2972
+ "step": 18300
2973
+ },
2974
+ {
2975
+ "epoch": 40.59800664451827,
2976
+ "grad_norm": 0.8429868817329407,
2977
+ "learning_rate": 6.269911504424779e-05,
2978
+ "loss": 0.2801,
2979
+ "step": 18350
2980
+ },
2981
+ {
2982
+ "epoch": 40.70874861572536,
2983
+ "grad_norm": 1.258792519569397,
2984
+ "learning_rate": 6.196165191740412e-05,
2985
+ "loss": 0.2458,
2986
+ "step": 18400
2987
+ },
2988
+ {
2989
+ "epoch": 40.819490586932446,
2990
+ "grad_norm": 0.4864571690559387,
2991
+ "learning_rate": 6.122418879056046e-05,
2992
+ "loss": 0.2615,
2993
+ "step": 18450
2994
+ },
2995
+ {
2996
+ "epoch": 40.93023255813954,
2997
+ "grad_norm": 0.49407410621643066,
2998
+ "learning_rate": 6.048672566371681e-05,
2999
+ "loss": 0.2419,
3000
+ "step": 18500
3001
+ },
3002
+ {
3003
+ "epoch": 41.0,
3004
+ "eval_cer": 0.04970230109289221,
3005
+ "eval_loss": 0.20170935988426208,
3006
+ "eval_runtime": 504.9031,
3007
+ "eval_samples_per_second": 20.164,
3008
+ "eval_steps_per_second": 2.521,
3009
+ "eval_wer": 0.19350049787272564,
3010
+ "step": 18532
3011
+ },
3012
+ {
3013
+ "epoch": 41.03986710963455,
3014
+ "grad_norm": 28.640010833740234,
3015
+ "learning_rate": 5.9749262536873155e-05,
3016
+ "loss": 18.8906,
3017
+ "step": 18550
3018
+ },
3019
+ {
3020
+ "epoch": 41.15060908084164,
3021
+ "grad_norm": 0.40494856238365173,
3022
+ "learning_rate": 5.9011799410029494e-05,
3023
+ "loss": 6.7845,
3024
+ "step": 18600
3025
+ },
3026
+ {
3027
+ "epoch": 41.261351052048724,
3028
+ "grad_norm": 1.3764681816101074,
3029
+ "learning_rate": 5.827433628318584e-05,
3030
+ "loss": 3.1387,
3031
+ "step": 18650
3032
+ },
3033
+ {
3034
+ "epoch": 41.372093023255815,
3035
+ "grad_norm": 0.8231045007705688,
3036
+ "learning_rate": 5.753687315634218e-05,
3037
+ "loss": 3.1009,
3038
+ "step": 18700
3039
+ },
3040
+ {
3041
+ "epoch": 41.4828349944629,
3042
+ "grad_norm": 0.6420056223869324,
3043
+ "learning_rate": 5.679941002949852e-05,
3044
+ "loss": 3.0734,
3045
+ "step": 18750
3046
+ },
3047
+ {
3048
+ "epoch": 41.59357696566999,
3049
+ "grad_norm": 1.9376227855682373,
3050
+ "learning_rate": 5.606194690265486e-05,
3051
+ "loss": 2.8817,
3052
+ "step": 18800
3053
+ },
3054
+ {
3055
+ "epoch": 41.70431893687708,
3056
+ "grad_norm": 1.5035388469696045,
3057
+ "learning_rate": 5.53244837758112e-05,
3058
+ "loss": 1.8786,
3059
+ "step": 18850
3060
+ },
3061
+ {
3062
+ "epoch": 41.81506090808416,
3063
+ "grad_norm": 0.897224485874176,
3064
+ "learning_rate": 5.4587020648967546e-05,
3065
+ "loss": 1.1317,
3066
+ "step": 18900
3067
+ },
3068
+ {
3069
+ "epoch": 41.925802879291254,
3070
+ "grad_norm": 0.7507091760635376,
3071
+ "learning_rate": 5.3849557522123885e-05,
3072
+ "loss": 0.8011,
3073
+ "step": 18950
3074
+ },
3075
+ {
3076
+ "epoch": 42.0,
3077
+ "eval_cer": 0.062023627357840616,
3078
+ "eval_loss": 0.4412783980369568,
3079
+ "eval_runtime": 457.2135,
3080
+ "eval_samples_per_second": 22.267,
3081
+ "eval_steps_per_second": 2.784,
3082
+ "eval_wer": 0.24151353308590567,
3083
+ "step": 18984
3084
+ },
3085
+ {
3086
+ "epoch": 42.03543743078627,
3087
+ "grad_norm": 0.92021244764328,
3088
+ "learning_rate": 5.311209439528023e-05,
3089
+ "loss": 0.6342,
3090
+ "step": 19000
3091
+ },
3092
+ {
3093
+ "epoch": 42.14617940199336,
3094
+ "grad_norm": 1.2486605644226074,
3095
+ "learning_rate": 5.2374631268436575e-05,
3096
+ "loss": 0.5457,
3097
+ "step": 19050
3098
+ },
3099
+ {
3100
+ "epoch": 42.25692137320044,
3101
+ "grad_norm": 1.1830624341964722,
3102
+ "learning_rate": 5.1637168141592914e-05,
3103
+ "loss": 0.4583,
3104
+ "step": 19100
3105
+ },
3106
+ {
3107
+ "epoch": 42.36766334440753,
3108
+ "grad_norm": 0.9644584059715271,
3109
+ "learning_rate": 5.089970501474926e-05,
3110
+ "loss": 0.4325,
3111
+ "step": 19150
3112
+ },
3113
+ {
3114
+ "epoch": 42.478405315614616,
3115
+ "grad_norm": 0.5905857086181641,
3116
+ "learning_rate": 5.0162241887905605e-05,
3117
+ "loss": 0.4296,
3118
+ "step": 19200
3119
+ },
3120
+ {
3121
+ "epoch": 42.58914728682171,
3122
+ "grad_norm": 0.7383226752281189,
3123
+ "learning_rate": 4.9424778761061944e-05,
3124
+ "loss": 0.3936,
3125
+ "step": 19250
3126
+ },
3127
+ {
3128
+ "epoch": 42.69988925802879,
3129
+ "grad_norm": 1.9408543109893799,
3130
+ "learning_rate": 4.868731563421829e-05,
3131
+ "loss": 0.3912,
3132
+ "step": 19300
3133
+ },
3134
+ {
3135
+ "epoch": 42.81063122923588,
3136
+ "grad_norm": 0.9167234897613525,
3137
+ "learning_rate": 4.794985250737463e-05,
3138
+ "loss": 0.3729,
3139
+ "step": 19350
3140
+ },
3141
+ {
3142
+ "epoch": 42.921373200442964,
3143
+ "grad_norm": 0.4777889847755432,
3144
+ "learning_rate": 4.7212389380530966e-05,
3145
+ "loss": 0.3462,
3146
+ "step": 19400
3147
+ },
3148
+ {
3149
+ "epoch": 43.0,
3150
+ "eval_cer": 0.05363656801386602,
3151
+ "eval_loss": 0.23979651927947998,
3152
+ "eval_runtime": 447.7874,
3153
+ "eval_samples_per_second": 22.736,
3154
+ "eval_steps_per_second": 2.843,
3155
+ "eval_wer": 0.20780302344527926,
3156
+ "step": 19436
3157
+ },
3158
+ {
3159
+ "epoch": 43.031007751937985,
3160
+ "grad_norm": 0.6891963481903076,
3161
+ "learning_rate": 4.6474926253687305e-05,
3162
+ "loss": 0.3587,
3163
+ "step": 19450
3164
+ },
3165
+ {
3166
+ "epoch": 43.141749723145075,
3167
+ "grad_norm": 0.5241971611976624,
3168
+ "learning_rate": 4.573746312684365e-05,
3169
+ "loss": 0.3362,
3170
+ "step": 19500
3171
+ },
3172
+ {
3173
+ "epoch": 43.25249169435216,
3174
+ "grad_norm": 0.9411165714263916,
3175
+ "learning_rate": 4.4999999999999996e-05,
3176
+ "loss": 0.3442,
3177
+ "step": 19550
3178
+ },
3179
+ {
3180
+ "epoch": 43.36323366555925,
3181
+ "grad_norm": 0.5330546498298645,
3182
+ "learning_rate": 4.4262536873156335e-05,
3183
+ "loss": 0.3323,
3184
+ "step": 19600
3185
+ },
3186
+ {
3187
+ "epoch": 43.47397563676633,
3188
+ "grad_norm": 5.120829105377197,
3189
+ "learning_rate": 4.352507374631268e-05,
3190
+ "loss": 0.3192,
3191
+ "step": 19650
3192
+ },
3193
+ {
3194
+ "epoch": 43.584717607973424,
3195
+ "grad_norm": 0.6921789050102234,
3196
+ "learning_rate": 4.2787610619469026e-05,
3197
+ "loss": 0.3134,
3198
+ "step": 19700
3199
+ },
3200
+ {
3201
+ "epoch": 43.69545957918051,
3202
+ "grad_norm": 1.0152885913848877,
3203
+ "learning_rate": 4.2050147492625364e-05,
3204
+ "loss": 0.3127,
3205
+ "step": 19750
3206
+ },
3207
+ {
3208
+ "epoch": 43.8062015503876,
3209
+ "grad_norm": 0.5916356444358826,
3210
+ "learning_rate": 4.131268436578171e-05,
3211
+ "loss": 0.3069,
3212
+ "step": 19800
3213
+ },
3214
+ {
3215
+ "epoch": 43.91694352159468,
3216
+ "grad_norm": 1.337470531463623,
3217
+ "learning_rate": 4.0575221238938055e-05,
3218
+ "loss": 0.3384,
3219
+ "step": 19850
3220
+ },
3221
+ {
3222
+ "epoch": 44.0,
3223
+ "eval_cer": 0.050960541297111486,
3224
+ "eval_loss": 0.2203516960144043,
3225
+ "eval_runtime": 444.0494,
3226
+ "eval_samples_per_second": 22.928,
3227
+ "eval_steps_per_second": 2.867,
3228
+ "eval_wer": 0.19795419570924233,
3229
+ "step": 19888
3230
+ }
3231
+ ],
3232
+ "logging_steps": 50,
3233
+ "max_steps": 22600,
3234
+ "num_input_tokens_seen": 0,
3235
+ "num_train_epochs": 50,
3236
+ "save_steps": 500,
3237
+ "stateful_callbacks": {
3238
+ "EarlyStoppingCallback": {
3239
+ "args": {
3240
+ "early_stopping_patience": 3,
3241
+ "early_stopping_threshold": 0.0
3242
+ },
3243
+ "attributes": {
3244
+ "early_stopping_patience_counter": 3
3245
+ }
3246
+ },
3247
+ "TrainerControl": {
3248
+ "args": {
3249
+ "should_epoch_stop": false,
3250
+ "should_evaluate": false,
3251
+ "should_log": false,
3252
+ "should_save": true,
3253
+ "should_training_stop": true
3254
+ },
3255
+ "attributes": {}
3256
+ }
3257
+ },
3258
+ "total_flos": 1.638225383074747e+20,
3259
+ "train_batch_size": 32,
3260
+ "trial_name": null,
3261
+ "trial_params": null
3262
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:14f31b49e1ae26d733ffa216aa585d121e3452f906aae2067058ad7c3e4cf5d8
3
+ size 5905
vocab.json ADDED
@@ -0,0 +1,39 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "[PAD]": 0,
3
+ "[UNK]": 1,
4
+ "|": 2,
5
+ "ء": 3,
6
+ "ؤ": 4,
7
+ "ئ": 5,
8
+ "ا": 6,
9
+ "ب": 7,
10
+ "ة": 8,
11
+ "ت": 9,
12
+ "ث": 10,
13
+ "ج": 11,
14
+ "ح": 12,
15
+ "خ": 13,
16
+ "د": 14,
17
+ "ذ": 15,
18
+ "ر": 16,
19
+ "ز": 17,
20
+ "س": 18,
21
+ "ش": 19,
22
+ "ص": 20,
23
+ "ض": 21,
24
+ "ط": 22,
25
+ "ظ": 23,
26
+ "ع": 24,
27
+ "غ": 25,
28
+ "ف": 26,
29
+ "ق": 27,
30
+ "ك": 28,
31
+ "ل": 29,
32
+ "م": 30,
33
+ "ن": 31,
34
+ "ه": 32,
35
+ "و": 33,
36
+ "ى": 34,
37
+ "ي": 35,
38
+ "ی": 36
39
+ }