Training in progress, step 1000

config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "openai/whisper-small",
+  "_name_or_path": "gerryc/whisper-small-ar",
   "activation_dropout": 0.0,
   "activation_function": "gelu",
   "architectures": [

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e6bd105af5822c4ea555248d7265a2e4584a35330a88bd95dac230a2ab282d6c
+oid sha256:adf21435a36297bdf8ca67e060fb08761854d387efe59c9d87afb8fb3c055cd9
 size 967102601

runs/Dec09_22-29-52_150-136-66-91/1670625011.7124658/events.out.tfevents.1670625011.150-136-66-91.153804.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bb4ec7f1ffd1583ec745882940471dce0ead727edf65321819a0f1dde03ed067
+size 5869

runs/Dec09_22-29-52_150-136-66-91/events.out.tfevents.1670625011.150-136-66-91.153804.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:69bbd4ccbfdc1e0cccada7050c910e17c4d56061a6e40966cbe41af90f465299
+size 10853

tokenizer_config.json

@@ -19,7 +19,7 @@
   },
   "errors": "replace",
   "model_max_length": 1024,
-  "name_or_path": "openai/whisper-small",
+  "name_or_path": "gerryc/whisper-small-ar",
   "pad_token": null,
   "processor_class": "WhisperProcessor",
   "return_attention_mask": false,

train2.sh

@@ -0,0 +1,41 @@
+python run_speech_recognition_seq2seq_streaming.py \
+--model_name_or_path="gerryc/whisper-small-ar" \
+--dataset_name="mozilla-foundation/common_voice_11_0" \
+--dataset_config_name="ar" \
+--language="arabic" \
+--train_split_name="train" \
+--eval_split_name="validation" \
+--max_eval_samples=256 \
+--do_lower_case \
+--do_remove_punctuation \
+--model_index_name="Whisper Small Arabic" \
+--max_steps="50000" \
+--output_dir="./" \
+--per_device_train_batch_size="64" \
+--per_device_eval_batch_size="32" \
+--logging_steps="25" \
+--learning_rate="1e-5" \
+--warmup_steps="500" \
+--evaluation_strategy="steps" \
+--eval_steps="1000" \
+--save_strategy="steps" \
+--save_steps="1000" \
+--generation_max_length="225" \
+--length_column_name="input_length" \
+--max_duration_in_seconds="30" \
+--text_column_name="sentence" \
+--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 \
+--do_train \
+--do_eval \
+--predict_with_generate \
+--do_normalize_eval \
+--streaming \
+--use_auth_token \
+--push_to_hub

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cdd16fa657d48d514adf5732acdd109ad7c1338a57d1f8b857c022b07a7e05df
+oid sha256:587e06ffef62adbf1f508d5d25bcb28386f1133172f2384339b580f2f0e5cc43
 size 3579