Update app.py

app.py

@@ -104,7 +104,7 @@ async def precompute_chunk(json_input, chunk_size, current_chunk):
         return None, None, None
 
 # Function to process and visualize a chunk of log probs with dynamic top_logprobs
-def visualize_logprobs(json_input, chunk=0, chunk_size=1000):
+def visualize_logprobs(json_input, chunk=0, chunk_size=100):
     try:
         # Parse the input (handles JSON only)
         data = parse_input(json_input)
@@ -152,7 +152,7 @@ def visualize_logprobs(json_input, chunk=0, chunk_size=1000):
         if not logprobs or not tokens:
             return (create_empty_figure("Log Probabilities of Generated Tokens"), None, "No tokens to display.", create_empty_figure("Top Token Log Probabilities"), create_empty_figure("Significant Probability Drops"), 1, 0)
 
-        # Paginate data for chunks of 1,000 tokens
+        # Paginate data for chunks of 100 tokens
         total_chunks = max(1, (len(logprobs) + chunk_size - 1) // chunk_size)
         start_idx = chunk * chunk_size
         end_idx = min((chunk + 1) * chunk_size, len(logprobs))
@@ -285,7 +285,7 @@ def visualize_logprobs(json_input, chunk=0, chunk_size=1000):
 with gr.Blocks(title="Log Probability Visualizer") as app:
     gr.Markdown("# Log Probability Visualizer")
     gr.Markdown(
-        "Paste your JSON log prob data below to visualize tokens in chunks of 1,000. Fixed filter ≥ -100000, dynamic number of top_logprobs, handles missing or null fields. Next chunk is precomputed proactively."
+        "Paste your JSON log prob data below to visualize tokens in chunks of 100. Fixed filter ≥ -100000, dynamic number of top_logprobs, handles missing or null fields. Next chunk is precomputed proactively."
     )
 
     with gr.Row():
@@ -326,7 +326,7 @@ with gr.Blocks(title="Log Probability Visualizer") as app:
     async def precompute_next_chunk(json_input, current_chunk, precomputed_next):
         if precomputed_next is not None:
             return precomputed_next  # Use cached precomputed chunk if available
-        next_tokens, next_logprobs, next_alternatives = await precompute_chunk(json_input, 1000, current_chunk)
+        next_tokens, next_logprobs, next_alternatives = await precompute_chunk(json_input, 100, current_chunk)
         if next_tokens is None or next_logprobs is None or next_alternatives is None:
             return None
         return (next_tokens, next_logprobs, next_alternatives)