Upload 18 files

app.py

@@ -195,7 +195,7 @@ with col1:
     )
 
 if decoder_model == "GPT-3.5 Turbo":
-    with col1:
+    with col2:
         with st.form("gpt_form"):
             openai_key = st.text_input(
                 "Enter OpenAI key",
@@ -208,23 +208,31 @@ if decoder_model == "GPT-3.5 Turbo":
                 openai.api_key = api_key
                 generated_text = gpt_turbo_model(edited_prompt)
 
+                st.subheader("Answer:")
+                regex_pattern_sentences = "(?<!\w\.\w.)(?<![A-Z][a-z]\.)(?<=\.|\?)\s"
+                generated_text_list = re.split(regex_pattern_sentences, generated_text)
+                for answer_text in generated_text_list:
+                    answer_text = f"""{answer_text}"""
+                    st.write(
+                        f"<ul><li><p>{answer_text}</p></li></ul>",
+                        unsafe_allow_html=True,
+                    )
+
 if decoder_model == "Vicuna-7B":
     with col2:
         st.write("The Vicuna Model is running: ...")
         st.write("The model takes 10-15 mins to generate the text.")
-    generated_text = vicuna_text_generate(prompt, vicuna_text_gen_model)
-
+        generated_text = vicuna_text_generate(prompt, vicuna_text_gen_model)
+        st.subheader("Answer:")
+        regex_pattern_sentences = "(?<!\w\.\w.)(?<![A-Z][a-z]\.)(?<=\.|\?)\s"
+        generated_text_list = re.split(regex_pattern_sentences, generated_text)
+        for answer_text in generated_text_list:
+            answer_text = f"""{answer_text}"""
+            st.write(
+                f"<ul><li><p>{answer_text}</p></li></ul>",
+                unsafe_allow_html=True,
+            )
 
-with col2:
-    st.subheader("Answer:")
-    regex_pattern_sentences = "(?<!\w\.\w.)(?<![A-Z][a-z]\.)(?<=\.|\?)\s"
-    generated_text_list = re.split(regex_pattern_sentences, generated_text)
-    for answer_text in generated_text_list:
-        answer_text = f"""{answer_text}"""
-        st.write(
-            f"<ul><li><p>{answer_text}</p></li></ul>",
-            unsafe_allow_html=True,
-        )
 
 
 tab1, tab2 = st.tabs(["Retrieved Text", "Retrieved Documents"])

utils/entity_extraction.py

@@ -7,20 +7,21 @@ from nltk.stem import PorterStemmer, WordNetLemmatizer
 
 def generate_ner_docs_prompt(query):
     prompt = """USER: Extract the company names and time duration mentioned in the question. The entities should be extracted in the following format: {"companies": list of companies mentioned in the question,"start-duration": ("start-quarter", "start-year"), "end-duration": ("end-quarter", "end-year")}. Return {"companies": None, "start-duration": (None, None), "end-duration": (None, None)} if the entities are not found.
+
 Examples:
-What did analysts ask about the Wearables during AAPL's earnings call?
-{"companies": ["AAPL"], "start-duration": (None, None), "end-duration": (None, None)}
 What is Intel's update on the server chip roadmap and strategy for Q1 2019?
 {"companies": ["Intel"], "start-duration": ("Q1", "2019"), "end-duration": ("Q1", "2019")}
 What are the opportunities and challenges in the Indian market for Amazon in 2016?
 {"companies": ["Amazon"], "start-duration": ("Q1", "2016"), "end-duration": ("Q4", "2016")}
+What did analysts ask about the Cisco's Webex?
+{"companies": ["Cisco"], "start-duration": (None, None), "end-duration": (None, None)}
 What is the comparative performance analysis between Intel and AMD in key overlapping segments such as PC, Gaming, and Data Centers in Q2 to Q3 2018?
 {"companies": ["Intel", "AMD"], "start-duration": ("Q2", "2018"), "end-duration": ("Q3", "2018")}
 How did Microsoft and Amazon perform in terms of reliability and scalability of cloud for the years 2016 and 2017?
 {"companies": ["Microsoft", "Amazon"], "start-duration": ("Q1", "2016"), "end-duration": ("Q4", "2017")}"""
     input_prompt = f"""###Input: {query}
 ASSISTANT:"""
-    final_prompt = prompt + "\n" + input_prompt
+    final_prompt = prompt + "\n\n" + input_prompt
     return final_prompt