Upload 4 files

app.py

@@ -2,11 +2,11 @@ import gradio as gr
 from confidence import run_nli
 
 DESCRIPTION = """\
-# Llama-2 13B Chat
-This Space demonstrates model [Llama-2-13b-chat](https://huggingface.co/meta-llama/Llama-2-13b-chat) by Meta, a Llama 2 model with 13B parameters fine-tuned for chat instructions. Feel free to play with it, or duplicate to run generations without a queue! If you want to run your own service, you can also [deploy the model on Inference Endpoints](https://huggingface.co/inference-endpoints).
-🔎 For more details about the Llama 2 family of models and how to use them with `transformers`, take a look [at our blog post](https://huggingface.co/blog/llama2).
-🔨 Looking for an even more powerful model? Check out the large [**70B** model demo](https://huggingface.co/spaces/ysharma/Explore_llamav2_with_TGI).
-🐇 For a smaller model that you can run on many GPUs, check our [7B model demo](https://huggingface.co/spaces/huggingface-projects/llama-2-7b-chat).
+# Llama Chatbot with confidence scores 🩺
+This space shows that we can teach LLMs to express how confident they are in their answers.
+Since we can only access free CPUs, we use a tiny Llama ([TinyLlama-1.1B](https://huggingface.co/PY007/TinyLlama-1.1B-Chat-v0.3)) as the chatbot and an [NLI model](https://github.com/potsawee/selfcheckgpt) to get scores. <br/>
+💯   There will be a score between 0 and 1 after each sentence, and a higher value means the sentence is more factual.<br/>
+⏳ It takes 150s-300s to process each query, and we limit the token numbers of answers for saving time.
 """
 
 def greet(query, history):
@@ -18,13 +18,12 @@ def greet(query, history):
 sample_list = [
     "Tell me something about Albert Einstein, e.g., a short bio with birth date and birth place",
     "Tell me something about Lihu Chen, e.g., a short bio with birth date and birth place",
+    "How tall is the Eiffel Tower?"
 ]
 
 iface = gr.ChatInterface(
     fn=greet,
     stop_btn=None,
-    # inputs="text",
-    # outputs="text",
     examples=sample_list,
     cache_examples=True
 )
@@ -35,4 +34,4 @@ with gr.Blocks() as demo:
     #gr.Markdown(LICENSE)
 
 
-iface.launch()
+demo.launch()

confidence.py

@@ -70,7 +70,7 @@ def run_nli(query, sample_size=5):
     final_content = ''
     for index, sent in enumerate(sentences):
         final_content += sent.strip() + ' ({a}) '.format(a=scores[index])
-    final_content += '\nThe confidence score of this answer is {a}'.format(a=sum(scores)/len(scores))
+    final_content += '\nThe confidence score of this answer is {a}'.format(a=round(sum(scores)/len(scores), 4))
     return final_content