Upload trainer.py with huggingface_hub

trainer.py

@@ -61,10 +61,12 @@ import torch
 from dataclasses import dataclass
 from typing import Any, Dict, List, Union
 
+@dataclass
 @dataclass
 class DataCollatorSpeechSeq2SeqWithPadding:
     processor: Any
     decoder_start_token_id: int
+    max_target_length: int = 448  # Add a parameter to control max length
 
     def __call__(self, features: List[Dict[str, Union[List[int], torch.Tensor]]]) -> Dict[str, torch.Tensor]:
         # split inputs and labels since they have to be of different lengths and need different padding methods
@@ -77,8 +79,13 @@ class DataCollatorSpeechSeq2SeqWithPadding:
         # pad the labels to max length
         labels_batch = self.processor.tokenizer.pad(label_features, return_tensors="pt")
 
+        # Truncate labels to max_target_length
+        labels = labels_batch["input_ids"]
+        if labels.shape[1] > self.max_target_length:
+            labels = labels[:, :self.max_target_length]
+
         # replace padding with -100 to ignore loss correctly
-        labels = labels_batch["input_ids"].masked_fill(labels_batch.attention_mask.ne(1), -100)
+        labels = labels.masked_fill(labels_batch.attention_mask[:, :labels.shape[1]].ne(1), -100)
 
         # if bos token is appended in previous tokenization step,
         # cut bos token here as it's append later anyways
@@ -89,9 +96,11 @@ class DataCollatorSpeechSeq2SeqWithPadding:
 
         return batch
 
+# When initializing the data collator, you can now specify the max length
 data_collator = DataCollatorSpeechSeq2SeqWithPadding(
     processor=processor,
     decoder_start_token_id=model.config.decoder_start_token_id,
+    max_target_length=448  # Explicitly set the max length
 )
 
 import evaluate
@@ -124,12 +133,14 @@ training_args = Seq2SeqTrainingArguments(
     max_steps=4000,
     gradient_checkpointing=True,
     fp16=True,
+    fp16_full_eval=True,
+    half_precision_backend='auto',
     evaluation_strategy="steps",
     per_device_eval_batch_size=8,
     predict_with_generate=True,
     generation_max_length=225,
     save_steps=1000,
-    eval_steps=1000,
+    eval_steps=20,
     logging_steps=25,
     report_to=["tensorboard"],
     load_best_model_at_end=True,