Spaces:
Paused
Paused
Commit
Β·
3fb7a02
1
Parent(s):
14baa02
Create overwrites.py
Browse files- overwrites.py +12 -0
overwrites.py
ADDED
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from llama_index import Prompt
|
2 |
+
from typing import List
|
3 |
+
import logging
|
4 |
+
|
5 |
+
def compact_text_chunks(self, prompt: Prompt, text_chunks: List[str]) -> List[str]:
|
6 |
+
logging.debug("Compacting text chunks...πππ")
|
7 |
+
combined_str = [c.strip() for c in text_chunks if c.strip()]
|
8 |
+
combined_str = [f"[{index+1}] {c}" for index, c in enumerate(combined_str)]
|
9 |
+
combined_str = "\n\n".join(combined_str)
|
10 |
+
# resplit based on self.max_chunk_overlap
|
11 |
+
text_splitter = self.get_text_splitter_given_prompt(prompt, 1, padding=1)
|
12 |
+
return text_splitter.split_text(combined_str)
|