update

app.py

@@ -666,14 +666,12 @@ def debug_chat_response_echo(
 
 
 # ============ CONSTANT ============
+MODEL_NAME = "DAMO-SeaL-13B"
 MODEL_TITLE = "DAMO-SeaL-13B - An Assistant for South East Asian Languages"
 MODEL_DESC = """
 This is a 13B DAMO-SeaL-Chat assistant model built by DAMO Academy, Alibaba Group. It can produce helpful responses in English, Vietnamese, Indonesian and Thai.
-""".strip()
-
-
-cite_markdown = """
-## Citation
+<br>
+#### Citation
 If you find our project useful, hope you can star our repo and cite our paper as follows:
 ```
 @article{damonlpsg2023seallm,
@@ -681,6 +679,11 @@ If you find our project useful, hope you can star our repo and cite our paper as
   title = {SeaL: A language model for South East Asian Languages},
   year = 2023,
 }
+```
+""".strip()
+
+
+cite_markdown = """
 """
 #   journal = {arXiv preprint arXiv:2306.02858}
 #   url = {https://arxiv.org/abs/2306.02858}
@@ -699,7 +702,6 @@ def launch():
     global demo, llm, DEBUG
     model_desc = MODEL_DESC
     model_path = MODEL_PATH
-    assert os.path.exists(model_path), f'{model_path} not found'
     model_title = MODEL_TITLE
     tensor_parallel = TENSOR_PARALLEL
     assert tensor_parallel > 0 , f'{tensor_parallel} invalid'
@@ -708,10 +710,11 @@ def launch():
     max_tokens = 4096
 
     if DEBUG:
-        model_desc += "<br>!!!!! This is in debug mode, responses will be copy original"
+        model_desc += "\n<br>!!!!! This is in debug mode, responses will be copy original"
         response_fn = debug_chat_response_echo
     else:
         # ! load the model
+        assert os.path.exists(model_path), f'{model_path} not found'
         llm = LLM(model=model_path, dtype=dtype, tensor_parallel_size=tensor_parallel)
 
         print(f'Use system prompt:\n{sys_prompt}')
@@ -723,6 +726,7 @@ def launch():
     demo = gr.ChatInterface(
         response_fn,
         chatbot=ChatBot(
+            # value=MODEL_NAME,
             bubble_full_width=False,
             latex_delimiters=[
                 { "left": "$", "right": "$", "display": False},
@@ -742,7 +746,32 @@ def launch():
             gr.Textbox(value=sys_prompt, label='System prompt', lines=8)], 
     )
 
-    gr.Markdown(cite_markdown)
+    # with gr.Blocks() as demo:
+    #     gr.ChatInterface(
+    #         response_fn,
+    #         chatbot=ChatBot(
+    #             bubble_full_width=False,
+    #             latex_delimiters=[
+    #                 { "left": "$", "right": "$", "display": False},
+    #                 { "left": "$$", "right": "$$", "display": True},
+    #             ]
+    #         ),
+    #         textbox=gr.Textbox(placeholder='Type message', lines=8, max_lines=128, min_width=200),
+    #         submit_btn=gr.Button(value='Submit', variant="primary", scale=0),
+    #         # stop_btn=None,
+    #         title=f"{model_title}",
+    #         description=f"{model_desc}",
+    #         # ! decide if can change the system prompt.
+    #         additional_inputs=[
+    #             gr.Number(value=0, label='Temperature (higher -> more random)'), 
+    #             gr.Number(value=max_tokens, label='Max generated tokens (increase if want more generation)'), 
+    #             gr.Number(value=0.4, label='Frequency penalty (> 0 encourage new tokens)'), 
+    #             gr.Textbox(value=sys_prompt, label='System prompt', lines=8)
+    #         ], 
+    #     )
+
+    #     gr.Markdown(cite_markdown)
+
     demo.queue()
     # demo.launch(server_port=args.port)
     demo.launch()