Spaces:
Paused
Paused
ramortegui
commited on
Commit
•
869e15e
1
Parent(s):
59277db
Remove blank lines
Browse files
app.py
CHANGED
@@ -8,24 +8,17 @@ from transformers import AutoTokenizer
|
|
8 |
|
9 |
bshtml_dir_loader = DirectoryLoader('./data/', loader_cls=BSHTMLLoader)
|
10 |
data = bshtml_dir_loader.load()
|
11 |
-
print("loading documents")
|
12 |
|
13 |
bloomz_tokenizer = AutoTokenizer.from_pretrained("bigscience/bloomz-1b7")
|
14 |
|
15 |
-
print("add tokenizer")
|
16 |
|
17 |
text_splitter = CharacterTextSplitter.from_huggingface_tokenizer(bloomz_tokenizer,
|
18 |
chunk_size=100,
|
19 |
chunk_overlap=0,
|
20 |
separator="\n")
|
21 |
|
22 |
-
|
23 |
-
print("Add text spliters")
|
24 |
-
|
25 |
documents = text_splitter.split_documents(data)
|
26 |
|
27 |
-
print("Getting HF embeddings")
|
28 |
-
|
29 |
embeddings = HuggingFaceEmbeddings()
|
30 |
|
31 |
llm = HuggingFacePipeline.from_model_id(
|
@@ -33,24 +26,17 @@ llm = HuggingFacePipeline.from_model_id(
|
|
33 |
task="text-generation",
|
34 |
model_kwargs={"temperature" : 0, "max_length" : 500})
|
35 |
|
36 |
-
print("Adding LLM hugginFacePipeline with bigscience bloomz")
|
37 |
|
38 |
vectordb = Chroma.from_documents(documents=documents, embedding=embeddings)
|
39 |
|
40 |
-
print("Getting vectors")
|
41 |
-
|
42 |
doc_retriever = vectordb.as_retriever()
|
43 |
|
44 |
-
print("Creating Retreiver")
|
45 |
-
|
46 |
|
47 |
shakespeare_qa = RetrievalQA.from_chain_type(llm=llm, chain_type="stuff", retriever=doc_retriever)
|
48 |
|
49 |
-
print("Add shakespeare qa")
|
50 |
|
51 |
def query(query):
|
52 |
shakespeare_qa.run(query)
|
53 |
-
|
54 |
|
55 |
iface = gr.Interface(fn=query, inputs="text", outputs="text")
|
56 |
iface.launch()
|
|
|
8 |
|
9 |
bshtml_dir_loader = DirectoryLoader('./data/', loader_cls=BSHTMLLoader)
|
10 |
data = bshtml_dir_loader.load()
|
|
|
11 |
|
12 |
bloomz_tokenizer = AutoTokenizer.from_pretrained("bigscience/bloomz-1b7")
|
13 |
|
|
|
14 |
|
15 |
text_splitter = CharacterTextSplitter.from_huggingface_tokenizer(bloomz_tokenizer,
|
16 |
chunk_size=100,
|
17 |
chunk_overlap=0,
|
18 |
separator="\n")
|
19 |
|
|
|
|
|
|
|
20 |
documents = text_splitter.split_documents(data)
|
21 |
|
|
|
|
|
22 |
embeddings = HuggingFaceEmbeddings()
|
23 |
|
24 |
llm = HuggingFacePipeline.from_model_id(
|
|
|
26 |
task="text-generation",
|
27 |
model_kwargs={"temperature" : 0, "max_length" : 500})
|
28 |
|
|
|
29 |
|
30 |
vectordb = Chroma.from_documents(documents=documents, embedding=embeddings)
|
31 |
|
|
|
|
|
32 |
doc_retriever = vectordb.as_retriever()
|
33 |
|
|
|
|
|
34 |
|
35 |
shakespeare_qa = RetrievalQA.from_chain_type(llm=llm, chain_type="stuff", retriever=doc_retriever)
|
36 |
|
|
|
37 |
|
38 |
def query(query):
|
39 |
shakespeare_qa.run(query)
|
|
|
40 |
|
41 |
iface = gr.Interface(fn=query, inputs="text", outputs="text")
|
42 |
iface.launch()
|