add max_tokens

app.py

@@ -32,7 +32,7 @@ def gen_conversation(conversations):
     return messages
 
 
-def predict(inputs, top_p, temperature, openai_api_key, enable_index,
+def predict(inputs, top_p, temperature, openai_api_key, enable_index, max_tokens,
             chat_counter, chatbot=[], history=[]):  # repetition_penalty, top_k
     headers = {
         "Content-Type": "application/json",
@@ -44,7 +44,7 @@ def predict(inputs, top_p, temperature, openai_api_key, enable_index,
     if enable_index:
         # Faiss 检索最近的embedding
         docsearch = FAISS.load_local(faiss_store, CohereEmbeddings(cohere_api_key=embeddings_key))
-        llm = ChatOpenAI(openai_api_key=openai_api_key)
+        llm = ChatOpenAI(openai_api_key=openai_api_key, max_tokens=max_tokens)
         messages_combine = [
             SystemMessagePromptTemplate.from_template(MyTemplate['chat_combine_template']),
             HumanMessagePromptTemplate.from_template("{question}")
@@ -141,9 +141,11 @@ with gr.Blocks(css="""#col_container {width: 1000px; margin-left: auto; margin-r
         run = gr.Button("Run")
 
         # inputs, top_p, temperature, top_k, repetition_penalty
-        with gr.Accordion("Parameters", open=False):
+        with gr.Accordion("Parameters", open=True):
             top_p = gr.Slider(minimum=-0, maximum=1.0, value=1.0, step=0.05, interactive=True,
                               label="Top-p (nucleus sampling)", )
+            max_tokens = gr.Slider(minimum=512, maximum=3000, value=3000, step=100, interactive=True,
+                              label="Max Tokens", )
             temperature = gr.Slider(minimum=-0, maximum=5.0, value=1.0, step=0.1, interactive=True,
                                     label="Temperature", )
             # top_k = gr.Slider( minimum=1, maximum=50, value=4, step=1, interactive=True, label="Top-k",)
@@ -153,9 +155,9 @@ with gr.Blocks(css="""#col_container {width: 1000px; margin-left: auto; margin-r
             # 后续考虑加入搜索结果
             enable_search = gr.Checkbox(label='是', info='是否使用搜索结果')
 
-    inputs.submit(predict, [inputs, top_p, temperature, openai_api_key, enable_index, chat_counter, chatbot, state],
+    inputs.submit(predict, [inputs, top_p, temperature, openai_api_key, enable_index, max_tokens, chat_counter, chatbot, state],
                   [chatbot, state, chat_counter], )
-    run.click(predict, [inputs, top_p, temperature, openai_api_key, enable_index, chat_counter, chatbot, state],
+    run.click(predict, [inputs, top_p, temperature, openai_api_key, enable_index, max_tokens, chat_counter, chatbot, state],
               [chatbot, state, chat_counter], )
 
     # 每次对话结束都重置对话

indexer/index.pkl

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:0e6006bd06e30b017ce77e5859f1c7b0abcad6c69ac81c9067e1adeb448ac273
-size 17619
+oid sha256:1e1dfd6979bc9bf972281e3ec2a29aa93731dc16dcf84f9777c20b58d942983c
+size 245756

requirements.txt

@@ -2,5 +2,4 @@ openai==0.27.2
 gradio==3.21.0
 langchain==0.0.113
 cohere==3.4.0
-faiss-cpu==1.7.3
-sentence-transformers==2.2.2
+faiss-cpu==1.7.3