Stripped newline from text and convert image to RGB.

src/model/train.py

@@ -163,27 +163,15 @@ def validation_step(self, batch, batch_idx):
             candidates = []
             references = []
 
-            columns = [
-                "prediction",
-                "references",
-            ]
-            log_data = []
-
             for pred, captions in zip(predictions, labels):
-                predicted_text = (
-                    parts[1] if len(parts := pred.split("\n", 1)) > 1 else pred
-                )
+                # predicted_text = (
+                #     parts[1] if len(parts := pred.split("\n", 1)) > 1 else pred
+                # )
+                predicted_text = pred.replace("caption en", "").replace("\n", "")
+                captions = [caption.replace("\n", "") for caption in captions]
                 candidates.append(predicted_text)
                 references.append(captions)
 
-                if self.config.train_wandb_logger is not None:
-                    log_data.append([predicted_text, captions])
-
-            if len(log_data) > 0:
-                self.config.train_wandb_logger.log_text(
-                    key=f"log_samples_{batch_idx}", columns=columns, data=log_data
-                )
-
             corpus_scores = self.calculate_corpus_scores(references, candidates)
 
             bleu_1_scores.append(corpus_scores["bleu_1"].item())

src/utils/train_utils.py

@@ -10,7 +10,7 @@
 
 
 def train_collate_fn(examples, processor, device):
-    images = [example["image"] for example in examples]
+    images = [example["image"].convert("RGB") for example in examples]
     texts = [PROMPT for _ in examples]
     captions = [example["caption"][0] for example in examples]
 
@@ -41,7 +41,7 @@ def train_collate_fn(examples, processor, device):
 
 
 def eval_collate_fn(examples, processor, device):
-    images = [example["image"] for example in examples]
+    images = [example["image"].convert("RGB") for example in examples]
     texts = [PROMPT for _ in examples]
     captions = [example["caption"] for example in examples]