Spaces:
Sleeping
Sleeping
Commit
•
8534ef7
1
Parent(s):
328d783
Update app.py
Browse files
app.py
CHANGED
@@ -89,8 +89,17 @@ def main():
|
|
89 |
|
90 |
# loading remote llama model
|
91 |
#llm = LlamaForCausalLM.from_pretrained("meta-llama/Llama-2-7b-hf")
|
92 |
-
llm = AutoModelForCausalLM.from_pretrained("google/gemma-1.1-2b-it")
|
93 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
94 |
# custom prompt
|
95 |
custom_template="""
|
96 |
You are a smart personal assistant and your task is to provide the answer of the given question based only on the given context. \n
|
@@ -111,7 +120,7 @@ Helpful Answer:
|
|
111 |
qa = ConversationalRetrievalChain.from_llm(llm,
|
112 |
#chain_type = "stuff",
|
113 |
chain_type = "stuff",
|
114 |
-
verbose=True,
|
115 |
#retriever=docsearch.as_retriever()
|
116 |
retriever=docsearch.as_retriever(search_kwargs = {"k" : 4, "search_type" : "similarity"}),
|
117 |
combine_docs_chain_kwargs={"prompt": QA_PROMPT}
|
|
|
89 |
|
90 |
# loading remote llama model
|
91 |
#llm = LlamaForCausalLM.from_pretrained("meta-llama/Llama-2-7b-hf")
|
92 |
+
#llm = AutoModelForCausalLM.from_pretrained("google/gemma-1.1-2b-it")
|
93 |
|
94 |
+
token = os.environ["HF_TOKEN"]
|
95 |
+
llm = AutoModelForCausalLM.from_pretrained(
|
96 |
+
"google/gemma-7b-it",
|
97 |
+
# torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32,
|
98 |
+
torch_dtype=torch.float16,
|
99 |
+
token=token,
|
100 |
+
)
|
101 |
+
|
102 |
+
|
103 |
# custom prompt
|
104 |
custom_template="""
|
105 |
You are a smart personal assistant and your task is to provide the answer of the given question based only on the given context. \n
|
|
|
120 |
qa = ConversationalRetrievalChain.from_llm(llm,
|
121 |
#chain_type = "stuff",
|
122 |
chain_type = "stuff",
|
123 |
+
#verbose=True,
|
124 |
#retriever=docsearch.as_retriever()
|
125 |
retriever=docsearch.as_retriever(search_kwargs = {"k" : 4, "search_type" : "similarity"}),
|
126 |
combine_docs_chain_kwargs={"prompt": QA_PROMPT}
|