Abid commited on
Commit
de1a0f3
1 Parent(s): 5a7ca9f

Sota Results

Browse files

Files changed (3) hide show
  1. metrics.csv +6 -13
  2. notebook.ipynb +0 -0
  3. params.yml +88 -89
metrics.csv CHANGED
@@ -1,14 +1,7 @@
1
  Name,Value,Timestamp,Step
2
- train_runtime,1310.619,1646910000000,1
3
- train_samples_per_second,0.618,1646910000000,1
4
- train_steps_per_second,0.019,1646910000000,1
5
- total_flos,1.03E+17,1646910000000,1
6
- train_loss,0.6416,1646910000000,1
7
- epoch,184.62,1646910000000,1
8
- eval_loss,0.9889,1646910000000,1
9
- eval_wer,0.5607,1646910000000,1
10
- eval_cer,0.237,1646910000000,1
11
- eval_runtime,119.2739,1646910000000,1
12
- eval_samples_per_second,2.859,1646910000000,1
13
- eval_steps_per_second,0.361,1646910000000,1
14
- epoch,184.62,1646910000000,1
1
  Name,Value,Timestamp,Step
2
+ "epoch",184.62,1647505277782,1
3
+ "eval_cer",0.237,1647505277782,1
4
+ "eval_wer",0.5607,1647505277782,1
5
+ "Steps",2400,1647505277782,1
6
+ "eval_loss",0.9889,1647505277782,1
7
+ "loss",0.6416,1647505277782,1
 
 
 
 
 
 
 
notebook.ipynb CHANGED
The diff for this file is too large to render. See raw diff
params.yml CHANGED
@@ -1,89 +1,88 @@
1
- model: facebook/wav2vec2-xls-r-300m
2
- trainer:
3
- _n_gpu: 1
4
- adafactor: false
5
- adam_beta1: 0.9
6
- adam_beta2: 0.999
7
- adam_epsilon: 1.0e-08
8
- bf16: false
9
- bf16_full_eval: false
10
- dataloader_drop_last: false
11
- dataloader_num_workers: 0
12
- dataloader_pin_memory: true
13
- ddp_bucket_cap_mb: null
14
- ddp_find_unused_parameters: null
15
- debug: '[]'
16
- deepspeed: null
17
- disable_tqdm: false
18
- do_eval: true
19
- do_predict: false
20
- do_train: false
21
- eval_accumulation_steps: null
22
- eval_steps: 500
23
- evaluation_strategy: steps
24
- fp16: true
25
- fp16_backend: auto
26
- fp16_full_eval: false
27
- fp16_opt_level: O1
28
- gradient_accumulation_steps: 2
29
- gradient_checkpointing: true
30
- greater_is_better: null
31
- group_by_length: true
32
- half_precision_backend: amp
33
- hub_model_id: null
34
- hub_strategy: every_save
35
- hub_token: <HUB_TOKEN>
36
- ignore_data_skip: false
37
- label_names: null
38
- label_smoothing_factor: 0.0
39
- learning_rate: 0.0001
40
- length_column_name: length
41
- load_best_model_at_end: false
42
- local_rank: -1
43
- log_level: -1
44
- log_level_replica: -1
45
- log_on_each_node: true
46
- logging_dir: wav2vec2-large-xls-r-300m-Urdu-CV8/runs/Mar10_09-33-33_c7ab98cbc2d3
47
- logging_first_step: false
48
- logging_nan_inf_filter: true
49
- logging_steps: 500
50
- logging_strategy: steps
51
- lr_scheduler_type: linear
52
- max_grad_norm: 1.0
53
- max_steps: -1
54
- metric_for_best_model: null
55
- mp_parameters: ''
56
- no_cuda: false
57
- num_train_epochs: 200
58
- optim: adamw_hf
59
- output_dir: wav2vec2-large-xls-r-300m-Urdu-CV8
60
- overwrite_output_dir: false
61
- past_index: -1
62
- per_device_eval_batch_size: 8
63
- per_device_train_batch_size: 32
64
- per_gpu_eval_batch_size: null
65
- per_gpu_train_batch_size: null
66
- prediction_loss_only: false
67
- push_to_hub: true
68
- push_to_hub_model_id: null
69
- push_to_hub_organization: null
70
- push_to_hub_token: <PUSH_TO_HUB_TOKEN>
71
- remove_unused_columns: true
72
- report_to: '[''mlflow'', ''tensorboard'']'
73
- resume_from_checkpoint: null
74
- run_name: wav2vec2-large-xls-r-300m-Urdu-CV8
75
- save_on_each_node: false
76
- save_steps: 500
77
- save_strategy: steps
78
- save_total_limit: 1
79
- seed: 42
80
- sharded_ddp: '[]'
81
- skip_memory_metrics: true
82
- tf32: null
83
- tpu_metrics_debug: false
84
- tpu_num_cores: null
85
- use_legacy_prediction_loop: false
86
- warmup_ratio: 0.0
87
- warmup_steps: 10
88
- weight_decay: 0.0
89
- xpu_backend: null
1
+ _n_gpu: 1
2
+ adafactor: false
3
+ adam_beta1: 0.9
4
+ adam_beta2: 0.999
5
+ adam_epsilon: 1.0e-08
6
+ bf16: false
7
+ bf16_full_eval: false
8
+ dataloader_drop_last: false
9
+ dataloader_num_workers: 0
10
+ dataloader_pin_memory: true
11
+ ddp_bucket_cap_mb: null
12
+ ddp_find_unused_parameters: null
13
+ debug: '[]'
14
+ deepspeed: null
15
+ disable_tqdm: false
16
+ do_eval: true
17
+ do_predict: false
18
+ do_train: false
19
+ eval_accumulation_steps: null
20
+ eval_steps: 400
21
+ evaluation_strategy: steps
22
+ fp16: true
23
+ fp16_backend: auto
24
+ fp16_full_eval: false
25
+ fp16_opt_level: O1
26
+ gradient_accumulation_steps: 2
27
+ gradient_checkpointing: true
28
+ greater_is_better: null
29
+ group_by_length: true
30
+ half_precision_backend: amp
31
+ hub_model_id: null
32
+ hub_strategy: every_save
33
+ hub_token: <HUB_TOKEN>
34
+ ignore_data_skip: false
35
+ label_names: null
36
+ label_smoothing_factor: 0.0
37
+ learning_rate: 0.0001
38
+ length_column_name: length
39
+ load_best_model_at_end: false
40
+ local_rank: -1
41
+ log_level: -1
42
+ log_level_replica: -1
43
+ log_on_each_node: true
44
+ logging_dir: Model/runs/Mar17_07-52-06_7da6d8561539
45
+ logging_first_step: false
46
+ logging_nan_inf_filter: true
47
+ logging_steps: 400
48
+ logging_strategy: steps
49
+ lr_scheduler_type: linear
50
+ max_grad_norm: 1.0
51
+ max_steps: -1
52
+ metric_for_best_model: null
53
+ model_class: wav2vec2-xls-r-300m
54
+ mp_parameters: ''
55
+ no_cuda: false
56
+ num_train_epochs: 200
57
+ optim: adamw_hf
58
+ output_dir: Model
59
+ overwrite_output_dir: false
60
+ past_index: -1
61
+ per_device_eval_batch_size: 8
62
+ per_device_train_batch_size: 32
63
+ per_gpu_eval_batch_size: null
64
+ per_gpu_train_batch_size: null
65
+ prediction_loss_only: false
66
+ push_to_hub: false
67
+ push_to_hub_model_id: null
68
+ push_to_hub_organization: null
69
+ push_to_hub_token: <PUSH_TO_HUB_TOKEN>
70
+ remove_unused_columns: true
71
+ report_to: '[''mlflow'', ''tensorboard'']'
72
+ resume_from_checkpoint: null
73
+ run_name: Model
74
+ save_on_each_node: false
75
+ save_steps: 400
76
+ save_strategy: steps
77
+ save_total_limit: 2
78
+ seed: 42
79
+ sharded_ddp: '[]'
80
+ skip_memory_metrics: true
81
+ tf32: null
82
+ tpu_metrics_debug: false
83
+ tpu_num_cores: null
84
+ use_legacy_prediction_loop: false
85
+ warmup_ratio: 0.0
86
+ warmup_steps: 1000
87
+ weight_decay: 0.0
88
+ xpu_backend: null