show timing/speed for summarization
Browse files- summarize.py +8 -2
summarize.py
CHANGED
@@ -48,7 +48,7 @@ keep_page_info = (
|
|
48 |
|
49 |
sources = load_documents(source_pdfs_path, keep_page_info)
|
50 |
|
51 |
-
print(f"Splitting {len(sources)}
|
52 |
|
53 |
chunks = split_chunks(
|
54 |
sources, chunk_size=int(chunk_size), chunk_overlap=int(chunk_overlap)
|
@@ -66,7 +66,13 @@ result = summarize_chain.call_chain(
|
|
66 |
)
|
67 |
|
68 |
end = timer()
|
69 |
-
|
70 |
|
71 |
print("\n\n***Summary:")
|
72 |
print(result["output_text"])
|
|
|
|
|
|
|
|
|
|
|
|
|
|
48 |
|
49 |
sources = load_documents(source_pdfs_path, keep_page_info)
|
50 |
|
51 |
+
print(f"Splitting {len(sources)} documents in to chunks ...")
|
52 |
|
53 |
chunks = split_chunks(
|
54 |
sources, chunk_size=int(chunk_size), chunk_overlap=int(chunk_overlap)
|
|
|
66 |
)
|
67 |
|
68 |
end = timer()
|
69 |
+
total_time = end - start
|
70 |
|
71 |
print("\n\n***Summary:")
|
72 |
print(result["output_text"])
|
73 |
+
|
74 |
+
print(f"Total time used: {total_time:.3f} s")
|
75 |
+
print(f"Number of tokens generated: {llm_loader.streamer.total_tokens}")
|
76 |
+
print(
|
77 |
+
f"Average generation speed: {llm_loader.streamer.total_tokens / total_time:.3f} tokens/s"
|
78 |
+
)
|