src/train_char_gpt2_128.py

@@ -221,6 +221,8 @@ def perplexity(model, data, batch_size=32):
         # Progress update
         processed = min(i + batch_size, total_sequences)
         print(f"\rppl {processed}/{total_sequences} ({processed/total_sequences*100:.1f}%)", end="", flush=True)
+    
+    print()  # Final newline
     return np.exp(total_loss_sum / total_tokens_count)
 
 #################################### Model #########################################mo

src/train_char_gpt2_256.py

@@ -221,6 +221,8 @@ def perplexity(model, data, batch_size=32):
         # Progress update
         processed = min(i + batch_size, total_sequences)
         print(f"\rppl {processed}/{total_sequences} ({processed/total_sequences*100:.1f}%)", end="", flush=True)
+    
+    print()  # Final newline
     return np.exp(total_loss_sum / total_tokens_count)
 
 #################################### Model #########################################mo

src/train_char_gpt2_512.py

@@ -221,6 +221,8 @@ def perplexity(model, data, batch_size=32):
         # Progress update
         processed = min(i + batch_size, total_sequences)
         print(f"\rppl {processed}/{total_sequences} ({processed/total_sequences*100:.1f}%)", end="", flush=True)
+    
+    print()  # Final newline
     return np.exp(total_loss_sum / total_tokens_count)
 
 #################################### Model #########################################mo

src/train_char_gpt2_64.py

@@ -221,6 +221,8 @@ def perplexity(model, data, batch_size=32):
         # Progress update
         processed = min(i + batch_size, total_sequences)
         print(f"\rppl {processed}/{total_sequences} ({processed/total_sequences*100:.1f}%)", end="", flush=True)
+    
+    print()  # Final newline
     return np.exp(total_loss_sum / total_tokens_count)
 
 #################################### Model #########################################mo

src/train_char_gpt2_koef_128.py

@@ -223,6 +223,8 @@ def perplexity(model, data, batch_size=32):
         # Progress update
         processed = min(i + batch_size, total_sequences)
         print(f"\rppl {processed}/{total_sequences} ({processed/total_sequences*100:.1f}%)", end="", flush=True)
+    
+    print()  # Final newline
     return np.exp(total_loss_sum / total_tokens_count)
 
 #################################### Model #########################################mo

src/train_char_gpt2_koef_256.py

@@ -223,6 +223,8 @@ def perplexity(model, data, batch_size=32):
         # Progress update
         processed = min(i + batch_size, total_sequences)
         print(f"\rppl {processed}/{total_sequences} ({processed/total_sequences*100:.1f}%)", end="", flush=True)
+    
+    print()  # Final newline
     return np.exp(total_loss_sum / total_tokens_count)
 
 #################################### Model #########################################mo

src/train_char_gpt2_koef_512.py

@@ -223,6 +223,8 @@ def perplexity(model, data, batch_size=32):
         # Progress update
         processed = min(i + batch_size, total_sequences)
         print(f"\rppl {processed}/{total_sequences} ({processed/total_sequences*100:.1f}%)", end="", flush=True)
+    
+    print()  # Final newline
     return np.exp(total_loss_sum / total_tokens_count)
 
 #################################### Model #########################################mo

src/train_char_gpt2_koef_64.py

@@ -223,6 +223,8 @@ def perplexity(model, data, batch_size=32):
         # Progress update
         processed = min(i + batch_size, total_sequences)
         print(f"\rppl {processed}/{total_sequences} ({processed/total_sequences*100:.1f}%)", end="", flush=True)
+    
+    print()  # Final newline
     return np.exp(total_loss_sum / total_tokens_count)
 
 #################################### Model #########################################mo

src/train_optics_char_gpt2_128.py

@@ -327,6 +327,8 @@ def perplexity(model, data, batch_size=32):
         # Progress update
         processed = min(i + batch_size, total_sequences)
         print(f"\rppl {processed}/{total_sequences} ({processed/total_sequences*100:.1f}%)", end="", flush=True)
+    
+    print()  # Final newline
     return np.exp(total_loss_sum / total_tokens_count)
 
 #################################### Model #########################################mo

src/train_optics_char_gpt2_256.py

@@ -327,6 +327,8 @@ def perplexity(model, data, batch_size=32):
         # Progress update
         processed = min(i + batch_size, total_sequences)
         print(f"\rppl {processed}/{total_sequences} ({processed/total_sequences*100:.1f}%)", end="", flush=True)
+    
+    print()  # Final newline
     return np.exp(total_loss_sum / total_tokens_count)
 
 #################################### Model #########################################mo

src/train_optics_char_gpt2_512.py

@@ -327,6 +327,8 @@ def perplexity(model, data, batch_size=32):
         # Progress update
         processed = min(i + batch_size, total_sequences)
         print(f"\rppl {processed}/{total_sequences} ({processed/total_sequences*100:.1f}%)", end="", flush=True)
+    
+    print()  # Final newline
     return np.exp(total_loss_sum / total_tokens_count)
 
 #################################### Model #########################################mo

src/train_optics_char_gpt2_64.py

@@ -327,6 +327,8 @@ def perplexity(model, data, batch_size=32):
         # Progress update
         processed = min(i + batch_size, total_sequences)
         print(f"\rppl {processed}/{total_sequences} ({processed/total_sequences*100:.1f}%)", end="", flush=True)
+    
+    print()  # Final newline
     return np.exp(total_loss_sum / total_tokens_count)
 
 #################################### Model #########################################mo

src/train_optics_char_gpt2_ff.py

@@ -326,6 +326,8 @@ def perplexity(model, data, batch_size=32):
         # Progress update
         processed = min(i + batch_size, total_sequences)
         print(f"\rppl {processed}/{total_sequences} ({processed/total_sequences*100:.1f}%)", end="", flush=True)
+    
+    print()  # Final newline
     return np.exp(total_loss_sum / total_tokens_count)
 
 #################################### Model #########################################mo

src/train_optics_char_gpt2_nokoef_128.py

@@ -325,6 +325,8 @@ def perplexity(model, data, batch_size=32):
         # Progress update
         processed = min(i + batch_size, total_sequences)
         print(f"\rppl {processed}/{total_sequences} ({processed/total_sequences*100:.1f}%)", end="", flush=True)
+    
+    print()  # Final newline
     return np.exp(total_loss_sum / total_tokens_count)
 
 #################################### Model #########################################mo

src/train_optics_char_gpt2_nokoef_256.py

@@ -325,6 +325,8 @@ def perplexity(model, data, batch_size=32):
         # Progress update
         processed = min(i + batch_size, total_sequences)
         print(f"\rppl {processed}/{total_sequences} ({processed/total_sequences*100:.1f}%)", end="", flush=True)
+    
+    print()  # Final newline
     return np.exp(total_loss_sum / total_tokens_count)
 
 #################################### Model #########################################mo

src/train_optics_char_gpt2_nokoef_512.py

@@ -325,6 +325,8 @@ def perplexity(model, data, batch_size=32):
         # Progress update
         processed = min(i + batch_size, total_sequences)
         print(f"\rppl {processed}/{total_sequences} ({processed/total_sequences*100:.1f}%)", end="", flush=True)
+    
+    print()  # Final newline
     return np.exp(total_loss_sum / total_tokens_count)
 
 #################################### Model #########################################mo

src/train_optics_char_gpt2_nokoef_64.py

@@ -325,6 +325,8 @@ def perplexity(model, data, batch_size=32):
         # Progress update
         processed = min(i + batch_size, total_sequences)
         print(f"\rppl {processed}/{total_sequences} ({processed/total_sequences*100:.1f}%)", end="", flush=True)
+    
+    print()  # Final newline
     return np.exp(total_loss_sum / total_tokens_count)
 
 #################################### Model #########################################mo

src/train_optics_trainable_focal_dist_lens_64.py

@@ -343,6 +343,8 @@ def perplexity(model, data, batch_size=32):
         # Progress update
         processed = min(i + batch_size, total_sequences)
         print(f"\rppl {processed}/{total_sequences} ({processed/total_sequences*100:.1f}%)", end="", flush=True)
+    
+    print()  # Final newline
     return np.exp(total_loss_sum / total_tokens_count)
 
 #################################### Model #########################################

src/train_optics_trainable_lens_128.py

@@ -341,6 +341,8 @@ def perplexity(model, data, batch_size=32):
         # Progress update
         processed = min(i + batch_size, total_sequences)
         print(f"\rppl {processed}/{total_sequences} ({processed/total_sequences*100:.1f}%)", end="", flush=True)
+    
+    print()  # Final newline
     return np.exp(total_loss_sum / total_tokens_count)
 
 #################################### Model #########################################mo

src/train_optics_trainable_lens_256.py

@@ -341,6 +341,8 @@ def perplexity(model, data, batch_size=32):
         # Progress update
         processed = min(i + batch_size, total_sequences)
         print(f"\rppl {processed}/{total_sequences} ({processed/total_sequences*100:.1f}%)", end="", flush=True)
+    
+    print()  # Final newline
     return np.exp(total_loss_sum / total_tokens_count)
 
 #################################### Model #########################################mo

src/train_optics_trainable_lens_512.py

@@ -341,6 +341,8 @@ def perplexity(model, data, batch_size=32):
         # Progress update
         processed = min(i + batch_size, total_sequences)
         print(f"\rppl {processed}/{total_sequences} ({processed/total_sequences*100:.1f}%)", end="", flush=True)
+    
+    print()  # Final newline
     return np.exp(total_loss_sum / total_tokens_count)
 
 #################################### Model #########################################mo

src/train_optics_trainable_lens_64.py

@@ -341,6 +341,8 @@ def perplexity(model, data, batch_size=32):
         # Progress update
         processed = min(i + batch_size, total_sequences)
         print(f"\rppl {processed}/{total_sequences} ({processed/total_sequences*100:.1f}%)", end="", flush=True)
+    
+    print()  # Final newline
     return np.exp(total_loss_sum / total_tokens_count)
 
 #################################### Model #########################################