Adding some funny stuff #2

Merged
ryan merged 3 commits from data-preprocessing into main 2025-07-30 20:30:34 -04:00
2 changed files with 2 additions and 3 deletions
Showing only changes of commit 0a88a03c90 - Show all commits

View File

@@ -32,7 +32,7 @@ def chunk_data(texts: list[str], collection):
chunker = Chunker(collection)
print(f"chunking {len(texts)} documents")
for text in texts[: len(texts) // 2]:
for text in texts:
chunker.chunk_document(document=text)
@@ -40,7 +40,6 @@ def consult_oracle(input: str, collection):
# Ask
embeddings = Chunker.embedding_fx(input=[input])
results = collection.query(query_texts=[input], query_embeddings=embeddings)
print(results)
# Generate
output = ollama.generate(

View File

@@ -21,4 +21,4 @@ class PaperlessNGXService:
if __name__ == "__main__":
pp = PaperlessNGXService()
print(pp.get_data()[0].keys())
pp.get_data()