Adding recover. Fix README

README.md

@@ -26,7 +26,7 @@ model-index:
     metrics:
     - name: Wer
       type: wer
-      value: 47.08
+      value: 45.73
 ---
 
 # Whisper Tiny Norwegian Bokmål
@@ -34,8 +34,8 @@ model-index:
 This model is a fine-tuned version of [openai/whisper-medium](https://huggingface.co/openai/whisper-medium) trained on several datasets.
 
 It is currently in the middle of a large training. Currently it achieves the following results on the evaluation set:
-- Loss: 1.464
-- Wer: 47.08
+- Loss: 1.4616
+- Wer: 45.73
 
 ## Model description
 
@@ -55,7 +55,7 @@ The following hyperparameters were used during training:
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
 - lr_scheduler_warmup_steps: 1000
-- training_steps: 100.000 (currently @4.000)
+- training_steps: 100.000 (currently @5.000)
 - mixed_precision_training: fp16
 
 ### Live Training results

run.recover.sh

@@ -0,0 +1,40 @@
+python run_speech_recognition_seq2seq_streaming.py \
+	--model_name_or_path="openai/whisper-tiny" \
+        --resume_from_checkpoint="checkpoint-5000" \
+	--dataset_name="NbAiLab/NCC_S" \
+	--language="Norwegian" \
+	--train_split_name="train" \
+	--eval_split_name="validation" \
+	--model_index_name="Whisper Tiny Norwegian Bokmål" \
+	--max_steps="100000" \
+	--output_dir="./" \
+	--per_device_train_batch_size="128" \
+	--per_device_eval_batch_size="32" \
+	--gradient_accumulation_step="1" \
+	--logging_steps="50" \
+	--learning_rate="3e-6" \
+	--lr_scheduler_type="constant_with_warmup" \
+	--warmup_steps="1000" \
+	--evaluation_strategy="steps" \
+	--eval_steps="1000" \
+	--save_strategy="steps" \
+	--save_steps="1000" \
+	--generation_max_length="225" \
+	--length_column_name="duration" \
+	--max_duration_in_seconds="30" \
+	--text_column_name="text" \
+	--freeze_feature_encoder="False" \
+	--report_to="tensorboard" \
+	--metric_for_best_model="wer" \
+	--greater_is_better="False" \
+	--load_best_model_at_end \
+	--gradient_checkpointing \
+	--fp16 \
+	--overwrite_output_dir="false" \
+	--do_train \
+	--do_eval \
+	--predict_with_generate \
+	--do_normalize_eval \
+	--use_auth_token \
+	--push_to_hub \
+	--hub_model_id="NbAiLab/whisper-tiny-nob"