streamlit web demo

Author: greatzane

finetune.py

@@ -1,3 +1,5 @@
+# based on Baichuan2 code
+
 import os
 import math
 import pathlib

requirements.txt

@@ -2,3 +2,4 @@ transformers
 torch
 accelerate
 deepspeed
+streamlit

web_streamlit.py

@@ -0,0 +1,78 @@
+# based on Baichuan2 code
+
+import fire
+import json
+import torch
+import streamlit as st
+from transformers import AutoModelForCausalLM, AutoTokenizer
+from transformers.generation.utils import GenerationConfig
+
+
+st.set_page_config(page_title="LLM utils")
+st.title("LLM utils")
+
+
+@st.cache_resource
+def init_model(base_model):
+    model = AutoModelForCausalLM.from_pretrained(
+        base_model,
+        torch_dtype=torch.float16,
+        device_map="auto",
+        trust_remote_code=True
+    )
+    model.generation_config = GenerationConfig.from_pretrained(
+        base_model
+    )
+    tokenizer = AutoTokenizer.from_pretrained(
+        base_model,
+        use_fast=False,
+        trust_remote_code=True
+    )
+    return model, tokenizer
+
+
+def clear_chat_history():
+    del st.session_state.messages
+
+
+def init_chat_history():
+    if "messages" in st.session_state:
+        for message in st.session_state.messages:
+            avatar = '🧑' if message["role"] == "user" else '🤖'
+            with st.chat_message(message["role"], avatar=avatar):
+                st.markdown(message["content"])
+    else:
+        st.session_state.messages = []
+
+    return st.session_state.messages
+
+
+def main(
+        base_model: str = ""
+    ):
+
+    assert (
+        base_model
+    ), "Please specify a --base_model"
+
+    model, tokenizer = init_model(base_model)
+    messages = init_chat_history()
+
+    if prompt := st.chat_input("Shift + Enter 换行, Enter 发送"):
+        with st.chat_message("user", avatar='🧑'):
+            st.markdown(prompt)
+        messages.append({"role": "user", "content": prompt})
+        print(f"[user] {prompt}", flush=True)
+        with st.chat_message("assistant", avatar='🤖'):
+            placeholder = st.empty()
+            for response in model.chat(tokenizer, messages, stream=True):
+                placeholder.markdown(response)
+                if torch.backends.mps.is_available():
+                    torch.mps.empty_cache()
+        messages.append({"role": "assistant", "content": response})
+        print(json.dumps(messages, ensure_ascii=False), flush=True)
+
+        st.button("清空对话", on_click=clear_chat_history)
+
+if __name__ == "__main__":
+    fire.Fire(main)