update model card README.md

README.md

@@ -34,11 +34,11 @@ More information needed
 
 The following hyperparameters were used during training:
 - learning_rate: 8.379967082059723e-06
-- train_batch_size: 32
-- eval_batch_size: 32
+- train_batch_size: 16
+- eval_batch_size: 16
 - seed: 42
 - gradient_accumulation_steps: 2
-- total_train_batch_size: 64
+- total_train_batch_size: 32
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
 - num_epochs: 0.1

wandb/run-20220206_130241-drcrhi7r/files/config.yaml

@@ -1,5 +1,11 @@
 wandb_version: 1
 
+_n_gpu:
+  desc: null
+  value: 1
+_name_or_path:
+  desc: null
+  value: ./
 _wandb:
   desc: null
   value:
@@ -9,6 +15,34 @@ _wandb:
     huggingface_version: 4.17.0.dev0
     is_jupyter_run: false
     is_kaggle_kernel: false
+    m:
+    - 1: train/global_step
+      6:
+      - 3
+    - 1: train/train_runtime
+      5: 1
+      6:
+      - 1
+    - 1: train/train_samples_per_second
+      5: 1
+      6:
+      - 1
+    - 1: train/train_steps_per_second
+      5: 1
+      6:
+      - 1
+    - 1: train/total_flos
+      5: 1
+      6:
+      - 1
+    - 1: train/train_loss
+      5: 1
+      6:
+      - 1
+    - 1: train/epoch
+      5: 1
+      6:
+      - 1
     python_version: 3.9.7
     start_time: 1644148961
     t:
@@ -25,3 +59,641 @@ _wandb:
       6: 4.17.0.dev0
       8:
       - 5
+activation_dropout:
+  desc: null
+  value: 0.055
+adafactor:
+  desc: null
+  value: false
+adam_beta1:
+  desc: null
+  value: 0.9
+adam_beta2:
+  desc: null
+  value: 0.999
+adam_epsilon:
+  desc: null
+  value: 1.0e-08
+adapter_kernel_size:
+  desc: null
+  value: 3
+adapter_stride:
+  desc: null
+  value: 2
+add_adapter:
+  desc: null
+  value: false
+add_cross_attention:
+  desc: null
+  value: false
+apply_spec_augment:
+  desc: null
+  value: true
+architectures:
+  desc: null
+  value:
+  - Wav2Vec2ForCTC
+attention_dropout:
+  desc: null
+  value: 0.094
+bad_words_ids:
+  desc: null
+  value: null
+bf16:
+  desc: null
+  value: false
+bf16_full_eval:
+  desc: null
+  value: false
+bos_token_id:
+  desc: null
+  value: 1
+chunk_size_feed_forward:
+  desc: null
+  value: 0
+classifier_proj_size:
+  desc: null
+  value: 256
+codevector_dim:
+  desc: null
+  value: 768
+contrastive_logits_temperature:
+  desc: null
+  value: 0.1
+conv_bias:
+  desc: null
+  value: true
+conv_dim:
+  desc: null
+  value:
+  - 512
+  - 512
+  - 512
+  - 512
+  - 512
+  - 512
+  - 512
+conv_kernel:
+  desc: null
+  value:
+  - 10
+  - 3
+  - 3
+  - 3
+  - 3
+  - 2
+  - 2
+conv_stride:
+  desc: null
+  value:
+  - 5
+  - 2
+  - 2
+  - 2
+  - 2
+  - 2
+  - 2
+cross_attention_hidden_size:
+  desc: null
+  value: null
+ctc_loss_reduction:
+  desc: null
+  value: mean
+ctc_zero_infinity:
+  desc: null
+  value: true
+dataloader_drop_last:
+  desc: null
+  value: false
+dataloader_num_workers:
+  desc: null
+  value: 0
+dataloader_pin_memory:
+  desc: null
+  value: true
+ddp_bucket_cap_mb:
+  desc: null
+  value: None
+ddp_find_unused_parameters:
+  desc: null
+  value: None
+debug:
+  desc: null
+  value: '[]'
+decoder_start_token_id:
+  desc: null
+  value: null
+deepspeed:
+  desc: null
+  value: None
+disable_tqdm:
+  desc: null
+  value: false
+diversity_loss_weight:
+  desc: null
+  value: 0.1
+diversity_penalty:
+  desc: null
+  value: 0.0
+do_eval:
+  desc: null
+  value: true
+do_predict:
+  desc: null
+  value: false
+do_sample:
+  desc: null
+  value: false
+do_stable_layer_norm:
+  desc: null
+  value: true
+do_train:
+  desc: null
+  value: true
+early_stopping:
+  desc: null
+  value: false
+encoder_no_repeat_ngram_size:
+  desc: null
+  value: 0
+eos_token_id:
+  desc: null
+  value: 2
+eval_accumulation_steps:
+  desc: null
+  value: None
+eval_batch_size:
+  desc: null
+  value: 16
+eval_steps:
+  desc: null
+  value: 500
+evaluation_strategy:
+  desc: null
+  value: steps
+feat_extract_activation:
+  desc: null
+  value: gelu
+feat_extract_dropout:
+  desc: null
+  value: 0.0
+feat_extract_norm:
+  desc: null
+  value: layer
+feat_proj_dropout:
+  desc: null
+  value: 0.04
+feat_quantizer_dropout:
+  desc: null
+  value: 0.0
+final_dropout:
+  desc: null
+  value: 0.0
+finetuning_task:
+  desc: null
+  value: null
+forced_bos_token_id:
+  desc: null
+  value: null
+forced_eos_token_id:
+  desc: null
+  value: null
+fp16:
+  desc: null
+  value: true
+fp16_backend:
+  desc: null
+  value: auto
+fp16_full_eval:
+  desc: null
+  value: false
+fp16_opt_level:
+  desc: null
+  value: O1
+gradient_accumulation_steps:
+  desc: null
+  value: 2
+gradient_checkpointing:
+  desc: null
+  value: true
+greater_is_better:
+  desc: null
+  value: None
+group_by_length:
+  desc: null
+  value: true
+half_precision_backend:
+  desc: null
+  value: amp
+hidden_act:
+  desc: null
+  value: gelu
+hidden_dropout:
+  desc: null
+  value: 0.047
+hidden_size:
+  desc: null
+  value: 1024
+hub_model_id:
+  desc: null
+  value: NbAiLab/wav2vec2-large-voxrex-npsc-bokmaal
+hub_strategy:
+  desc: null
+  value: every_save
+hub_token:
+  desc: null
+  value: <HUB_TOKEN>
+id2label:
+  desc: null
+  value:
+    '0': LABEL_0
+    '1': LABEL_1
+ignore_data_skip:
+  desc: null
+  value: false
+initializer_range:
+  desc: null
+  value: 0.02
+intermediate_size:
+  desc: null
+  value: 4096
+is_decoder:
+  desc: null
+  value: false
+is_encoder_decoder:
+  desc: null
+  value: false
+label2id:
+  desc: null
+  value:
+    LABEL_0: 0
+    LABEL_1: 1
+label_names:
+  desc: null
+  value: None
+label_smoothing_factor:
+  desc: null
+  value: 0.0
+layer_norm_eps:
+  desc: null
+  value: 1.0e-05
+layerdrop:
+  desc: null
+  value: 0.041
+learning_rate:
+  desc: null
+  value: 8.379967082059723e-06
+length_column_name:
+  desc: null
+  value: input_length
+length_penalty:
+  desc: null
+  value: 1.0
+load_best_model_at_end:
+  desc: null
+  value: false
+local_rank:
+  desc: null
+  value: -1
+log_level:
+  desc: null
+  value: -1
+log_level_replica:
+  desc: null
+  value: -1
+log_on_each_node:
+  desc: null
+  value: true
+logging_dir:
+  desc: null
+  value: ./runs/Feb06_13-01-43_dante
+logging_first_step:
+  desc: null
+  value: false
+logging_nan_inf_filter:
+  desc: null
+  value: true
+logging_steps:
+  desc: null
+  value: 100
+logging_strategy:
+  desc: null
+  value: steps
+lr_scheduler_type:
+  desc: null
+  value: linear
+mask_channel_length:
+  desc: null
+  value: 10
+mask_channel_min_space:
+  desc: null
+  value: 1
+mask_channel_other:
+  desc: null
+  value: 0.0
+mask_channel_prob:
+  desc: null
+  value: 0.0
+mask_channel_selection:
+  desc: null
+  value: static
+mask_feature_length:
+  desc: null
+  value: 64
+mask_feature_min_masks:
+  desc: null
+  value: 0
+mask_feature_prob:
+  desc: null
+  value: 0.25
+mask_time_length:
+  desc: null
+  value: 10
+mask_time_min_masks:
+  desc: null
+  value: 2
+mask_time_min_space:
+  desc: null
+  value: 1
+mask_time_other:
+  desc: null
+  value: 0.0
+mask_time_prob:
+  desc: null
+  value: 0.082
+mask_time_selection:
+  desc: null
+  value: static
+max_grad_norm:
+  desc: null
+  value: 1.0
+max_length:
+  desc: null
+  value: 20
+max_steps:
+  desc: null
+  value: -1
+metric_for_best_model:
+  desc: null
+  value: None
+min_length:
+  desc: null
+  value: 0
+model_type:
+  desc: null
+  value: wav2vec2
+mp_parameters:
+  desc: null
+  value: ''
+no_cuda:
+  desc: null
+  value: false
+no_repeat_ngram_size:
+  desc: null
+  value: 0
+num_adapter_layers:
+  desc: null
+  value: 3
+num_attention_heads:
+  desc: null
+  value: 16
+num_beam_groups:
+  desc: null
+  value: 1
+num_beams:
+  desc: null
+  value: 1
+num_codevector_groups:
+  desc: null
+  value: 2
+num_codevectors_per_group:
+  desc: null
+  value: 320
+num_conv_pos_embedding_groups:
+  desc: null
+  value: 16
+num_conv_pos_embeddings:
+  desc: null
+  value: 128
+num_feat_extract_layers:
+  desc: null
+  value: 7
+num_hidden_layers:
+  desc: null
+  value: 24
+num_negatives:
+  desc: null
+  value: 100
+num_return_sequences:
+  desc: null
+  value: 1
+num_train_epochs:
+  desc: null
+  value: 0.1
+optim:
+  desc: null
+  value: adamw_hf
+output_attentions:
+  desc: null
+  value: false
+output_dir:
+  desc: null
+  value: ./
+output_hidden_size:
+  desc: null
+  value: 1024
+output_hidden_states:
+  desc: null
+  value: false
+output_scores:
+  desc: null
+  value: false
+overwrite_output_dir:
+  desc: null
+  value: true
+pad_token_id:
+  desc: null
+  value: 31
+past_index:
+  desc: null
+  value: -1
+per_device_eval_batch_size:
+  desc: null
+  value: 16
+per_device_train_batch_size:
+  desc: null
+  value: 16
+per_gpu_eval_batch_size:
+  desc: null
+  value: None
+per_gpu_train_batch_size:
+  desc: null
+  value: None
+prediction_loss_only:
+  desc: null
+  value: false
+prefix:
+  desc: null
+  value: null
+problem_type:
+  desc: null
+  value: null
+proj_codevector_dim:
+  desc: null
+  value: 768
+pruned_heads:
+  desc: null
+  value: {}
+push_to_hub:
+  desc: null
+  value: true
+push_to_hub_model_id:
+  desc: null
+  value: None
+push_to_hub_organization:
+  desc: null
+  value: None
+push_to_hub_token:
+  desc: null
+  value: <PUSH_TO_HUB_TOKEN>
+remove_invalid_values:
+  desc: null
+  value: false
+remove_unused_columns:
+  desc: null
+  value: true
+repetition_penalty:
+  desc: null
+  value: 1.0
+report_to:
+  desc: null
+  value: '[''tensorboard'', ''wandb'']'
+resume_from_checkpoint:
+  desc: null
+  value: None
+return_dict:
+  desc: null
+  value: true
+return_dict_in_generate:
+  desc: null
+  value: false
+run_name:
+  desc: null
+  value: ./
+save_on_each_node:
+  desc: null
+  value: false
+save_steps:
+  desc: null
+  value: 500
+save_strategy:
+  desc: null
+  value: steps
+save_total_limit:
+  desc: null
+  value: 3
+seed:
+  desc: null
+  value: 42
+sep_token_id:
+  desc: null
+  value: null
+sharded_ddp:
+  desc: null
+  value: '[]'
+skip_memory_metrics:
+  desc: null
+  value: true
+task_specific_params:
+  desc: null
+  value: null
+tdnn_dilation:
+  desc: null
+  value:
+  - 1
+  - 2
+  - 3
+  - 1
+  - 1
+tdnn_dim:
+  desc: null
+  value:
+  - 512
+  - 512
+  - 512
+  - 512
+  - 1500
+tdnn_kernel:
+  desc: null
+  value:
+  - 5
+  - 3
+  - 3
+  - 1
+  - 1
+temperature:
+  desc: null
+  value: 1.0
+tf32:
+  desc: null
+  value: None
+tie_encoder_decoder:
+  desc: null
+  value: false
+tie_word_embeddings:
+  desc: null
+  value: true
+tokenizer_class:
+  desc: null
+  value: null
+top_k:
+  desc: null
+  value: 50
+top_p:
+  desc: null
+  value: 1.0
+torch_dtype:
+  desc: null
+  value: float32
+torchscript:
+  desc: null
+  value: false
+tpu_metrics_debug:
+  desc: null
+  value: false
+tpu_num_cores:
+  desc: null
+  value: None
+train_batch_size:
+  desc: null
+  value: 16
+transformers_version:
+  desc: null
+  value: 4.17.0.dev0
+use_bfloat16:
+  desc: null
+  value: false
+use_legacy_prediction_loop:
+  desc: null
+  value: false
+use_weighted_layer_sum:
+  desc: null
+  value: false
+vocab_size:
+  desc: null
+  value: 34
+warmup_ratio:
+  desc: null
+  value: 0.0
+warmup_steps:
+  desc: null
+  value: 0
+weight_decay:
+  desc: null
+  value: 0.0
+xpu_backend:
+  desc: null
+  value: None
+xvector_output_dim:
+  desc: null
+  value: 512

wandb/run-20220206_130241-drcrhi7r/files/output.log

@@ -11,3 +11,15 @@ Configuration saved in ./preprocessor_config.json
 Saving model checkpoint to ./
 Configuration saved in ./config.json
 Model weights saved in ./pytorch_model.bin
+Configuration saved in ./preprocessor_config.json
+Upload file wandb/run-20220206_130008-2e07dm9k/run-2e07dm9k.wandb: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 25.4k/25.4k [00:00<?, ?B/s]
+Upload file wandb/run-20220206_130241-drcrhi7r/run-drcrhi7r.wandb: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 5.92k/5.92k [00:00<?, ?B/s]
+Upload file runs/Feb06_13-01-43_dante/events.out.tfevents.1644148960.dante.2793694.0: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 5.16k/5.16k [00:00<?, ?B/s]
+Upload file runs/Feb06_13-01-43_dante/1644148960.1271484/events.out.tfevents.1644148960.dante.2793694.1: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 4.65k/4.65k [00:00<?, ?B/s]
+Upload file training_args.bin: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 2.98k/2.98k [00:00<?, ?B/s]
+02/06/2022 13:03:18 - WARNING - huggingface_hub.repository - To https://huggingface.co/NbAiLab/wav2vec2-large-voxrex-npsc-bokmaal
+Upload file wandb/run-20220206_130241-drcrhi7r/run-drcrhi7r.wandb: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 5.92k/5.92k [00:02<?, ?B/s]
+Upload file runs/Feb06_13-01-43_dante/events.out.tfevents.1644148960.dante.2793694.0: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 5.16k/5.16k [00:02<?, ?B/s]
+Upload file runs/Feb06_13-01-43_dante/1644148960.1271484/events.out.tfevents.1644148960.dante.2793694.1: 100%|███████████████████████████████████████████████████████████████████��███████████████████████████████████████████████████████████████████████████| 4.65k/4.65k [00:02<?, ?B/s]
+Upload file training_args.bin: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 2.98k/2.98k [00:02<?, ?B/s]
+Upload file runs/Feb06_13-01-43_dante/1644148960.1271484/events.out.tfevents.1644148960.dante.2793694.1: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 4.65k/4.65k [00:02<?, ?B/s]

wandb/run-20220206_130241-drcrhi7r/logs/debug-internal.log

@@ -57,3 +57,9 @@
 2022-02-06 13:02:52,262 INFO    Thread-8  :2795095 [dir_watcher.py:_on_file_modified():230] file/dir modified: /mnt/lv_ai_1_dante/javierr/wav2vec2-large-voxrex-npsc-bokmaal/wandb/run-20220206_130241-drcrhi7r/files/output.log
 2022-02-06 13:02:59,345 DEBUG   HandlerThread:2795095 [handler.py:handle_request():130] handle_request: stop_status
 2022-02-06 13:02:59,346 DEBUG   SenderThread:2795095 [sender.py:send_request():248] send_request: stop_status
+2022-02-06 13:03:12,529 DEBUG   SenderThread:2795095 [sender.py:send():234] send: stats
+2022-02-06 13:03:13,275 INFO    Thread-8  :2795095 [dir_watcher.py:_on_file_modified():230] file/dir modified: /mnt/lv_ai_1_dante/javierr/wav2vec2-large-voxrex-npsc-bokmaal/wandb/run-20220206_130241-drcrhi7r/files/config.yaml
+2022-02-06 13:03:14,574 DEBUG   HandlerThread:2795095 [handler.py:handle_request():130] handle_request: stop_status
+2022-02-06 13:03:14,575 DEBUG   SenderThread:2795095 [sender.py:send_request():248] send_request: stop_status
+2022-02-06 13:03:18,277 INFO    Thread-8  :2795095 [dir_watcher.py:_on_file_modified():230] file/dir modified: /mnt/lv_ai_1_dante/javierr/wav2vec2-large-voxrex-npsc-bokmaal/wandb/run-20220206_130241-drcrhi7r/files/output.log
+2022-02-06 13:03:20,279 INFO    Thread-8  :2795095 [dir_watcher.py:_on_file_modified():230] file/dir modified: /mnt/lv_ai_1_dante/javierr/wav2vec2-large-voxrex-npsc-bokmaal/wandb/run-20220206_130241-drcrhi7r/files/output.log

wandb/run-20220206_130241-drcrhi7r/run-drcrhi7r.wandb

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a2b98a0ef81ed25512f97c98016f79e9004302cd0cabb2087e53fc384304dc90
-size 6066
+oid sha256:80cdbd04efa3996f26012b1e222873358ec602095e93070c0550435a2bc50348
+size 12121