Training in progress, step 300

config.json

@@ -17,7 +17,7 @@
   "decoder_layerdrop": 0.0,
   "decoder_layers": 24,
   "decoder_start_token_id": 50258,
-  "dropout": 0.0,
+  "dropout": 0.1,
   "encoder_attention_heads": 16,
   "encoder_ffn_dim": 4096,
   "encoder_layerdrop": 0.0,

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c552aaa81f911ae3fe51c5c92b5267348d4a3a4130f401374637dcf7da02e12f
+oid sha256:7636e3d95bfeff32e8b2bd10d0996660c605ffd61c13facfde26c8e2b5599938
 size 3055754841

run.sh

@@ -29,7 +29,7 @@ python run_speech_recognition_seq2seq_streaming.py \
 --load_best_model_at_end \
 --gradient_checkpointing \
 --fp16 \
---overwrite_output_dir="True" \
+--overwrite_output_dir="False" \
 --do_train \
 --do_eval \
 --predict_with_generate \

run_speech_recognition_seq2seq_streaming.py

@@ -556,6 +556,7 @@ def main():
             elif isinstance(train_dataloader.dataset, IterableDataset):
                 train_dataloader.dataset.set_epoch(train_dataloader.dataset._epoch + 1)
 
+    model.config.dropout=0.1
     # Initialize Trainer
     trainer = Seq2SeqTrainer(
         model=model,

runs/Dec14_09-09-02_129-213-22-31/1671009068.9843225/events.out.tfevents.1671009068.129-213-22-31.76517.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:78cc986e73b2a39d30783ce62e763818707dbb5a8526f805a7b55958b29b7b3b
+size 5869

runs/Dec14_09-09-02_129-213-22-31/events.out.tfevents.1671009068.129-213-22-31.76517.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7d74cebc2feab7a04ab7fde962da8d39c21532b57ef537c4a6242c1f298ea2d9
+size 5211

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c0f35dccc386ad3aa56ad7009153489022a8491fa55ee23bd71e2832a9f86ae9
+oid sha256:4c533f7abd990d9ef06c67fc5110076fffbb3d083bf1fbae49a66b54b16fd23b
 size 3579