gkrthk commited on
Commit
e769925
1 Parent(s): 341abc7
Files changed (1) hide show
  1. confluence_qa.py +3 -3
confluence_qa.py CHANGED
@@ -12,8 +12,8 @@ class ConfluenceQA:
12
  self.embeddings = HuggingFaceEmbeddings(model_name="multi-qa-MiniLM-L6-cos-v1")
13
 
14
  def define_model(self) -> None:
15
- tokenizer = T5Tokenizer.from_pretrained("google/flan-t5-large")
16
- model = T5ForConditionalGeneration.from_pretrained("google/flan-t5-large")
17
  pipe = pipeline("text2text-generation", model=model, tokenizer=tokenizer,max_new_tokens=1024)
18
  self.llm = HuggingFacePipeline(pipeline = pipe,model_kwargs={"temperature": 0.5})
19
 
@@ -28,7 +28,7 @@ class ConfluenceQA:
28
  url=confluence_url, username=username, api_key=api_key
29
  )
30
  documents = loader.load(include_attachments=include_attachment, limit=100, space_key=space_key)
31
- text_splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=10)
32
  documents = text_splitter.split_documents(documents)
33
  self.db = Chroma.from_documents(documents, self.embeddings)
34
  # question = "How do I make a space public?"
 
12
  self.embeddings = HuggingFaceEmbeddings(model_name="multi-qa-MiniLM-L6-cos-v1")
13
 
14
  def define_model(self) -> None:
15
+ tokenizer = T5Tokenizer.from_pretrained("google/flan-t5-xxl")
16
+ model = T5ForConditionalGeneration.from_pretrained("google/flan-t5-xxl")
17
  pipe = pipeline("text2text-generation", model=model, tokenizer=tokenizer,max_new_tokens=1024)
18
  self.llm = HuggingFacePipeline(pipeline = pipe,model_kwargs={"temperature": 0.5})
19
 
 
28
  url=confluence_url, username=username, api_key=api_key
29
  )
30
  documents = loader.load(include_attachments=include_attachment, limit=100, space_key=space_key)
31
+ text_splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=50)
32
  documents = text_splitter.split_documents(documents)
33
  self.db = Chroma.from_documents(documents, self.embeddings)
34
  # question = "How do I make a space public?"