fix: better logging

handler.py

@@ -50,24 +50,27 @@ class EndpointHandler():
             
             if len(texts) == 1:
                 # It's a query
-                logger.debug(f"Query: {texts}")
+                logger.info(f"Received query of 1 text with {len(texts[0])} characters and {len(texts[0].split())} words")
                 embedding = self._checkpoint.queryFromText(
                     queries=texts,
                     full_length_search=False,  # Indicates whether to encode the query for a full-length search.
                 )
-                logger.debug(f"Query embedding shape: {embedding.shape}")
+                logger.info(f"Query embedding shape: {embedding.shape}")
                 return [
                     {"input": inputs, "query_embedding": embedding.tolist()[0]}
                 ]
             elif len(texts) > 1:
                 # It's a batch of chunks
-                logger.info(f"Batch of chunks: {texts}")
+                logger.info(f"Received batch of {len(texts)} chunks")
+                for i, text in enumerate(texts):
+                    logger.info(f"Chunk {i} has {len(text)} characters and {len(text.split())} words")
                 embeddings, token_id_lists = self._checkpoint.docFromText(
                     docs=texts,
                     bsize=self._config.bsize, # Batch size
                     keep_dims=True, # Do NOT flatten the embeddings
                     return_tokens=True, # Return the tokens as well
                 )
+                logger.info(f"Chunk embeddings shape: {embeddings.shape}")
                 token_lists = []
                 for text, embedding, token_ids in zip(texts, embeddings, token_id_lists):
                     logger.debug(f"Chunk: {text}")