Update app.py

app.py

@@ -9,8 +9,7 @@ from   threading import Thread
 from   transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer
 
 # Model
-model_name = "TinyLlama/TinyLlama-1.1B-Chat-v0.3"
-model_name = "mistralai/Mistral-7B-Instruct-v0.1"
+model_name = "augmxnt/shisa-7b-v1"
 
 # UI Settings
 title = "Shisa 7B"
@@ -18,15 +17,14 @@ description = "Test out Shisa 7B in either English or Japanese."
 placeholder = "Type Here / ここに入力してください" 
 examples = [
     "What's the best ramen in Tokyo?",
-    "東京でおすすめのラーメン屋さんを教えていただけますか。",
+    "あなたは熱狂的なポケモンファンです。",
     "東京でおすすめのラーメン屋ってどこ？",
 ]
 
 # LLM Settings
-system_prompt = 'You are a helpful, friendly assistant.'
+system_prompt = 'あなたは役に立つアシスタントです。'
 chat_history = [{"role": "system", "content": system_prompt}]
 tokenizer = AutoTokenizer.from_pretrained(model_name)
-tokenizer.chat_template = "{%- for idx in range(0, messages|length) -%}\n{%- if messages[idx]['role'] == 'user' -%}\n{%- if idx > 1 -%}\n{{- bos_token + '[INST] ' + messages[idx]['content'] + ' [/INST]' -}}\n{%- else -%}\n{{- messages[idx]['content'] + ' [/INST]' -}}\n{%- endif -%}\n{% elif messages[idx]['role'] == 'system' %}\n{{- '[INST] <<SYS>>\\n' + messages[idx]['content'] + '\\n<</SYS>>\\n\\n' -}}\n{%- elif messages[idx]['role'] == 'assistant' -%}\n{{- ' '  + messages[idx]['content'] + ' ' + eos_token -}}\n{% endif %}\n{% endfor %}\n"
 model = AutoModelForCausalLM.from_pretrained(
     model_name,
     torch_dtype=torch.bfloat16,
@@ -78,6 +76,6 @@ chat_interface = gr.ChatInterface(
 # https://huggingface.co/spaces/ysharma/Explore_llamav2_with_TGI/blob/main/app.py#L219 - we use this with construction b/c Gradio barfs on autoreload otherwise
 with gr.Blocks() as demo:
     chat_interface.render()
-    gr.Markdown("You can try asking this question in English, formal Japanese, and informal Japanese. You might need to ask it to reply informally with something like もっと友達みたいに話そうよ。あんまり堅苦しくなくて。to get informal replies. We limit output to 200 tokens.")
+    gr.Markdown("You can try asking this question in Japanese or English. We limit output to 200 tokens.")
 
 demo.queue().launch()