update app.py

app.py

@@ -28,29 +28,22 @@ st.sidebar.title("Instructions: How to use")
 st.sidebar.write("""
 1. **Write Text or Select Sample:**
    - Enter text in the text area or use the dropdown to choose a sample.
-
 2. **Select a Task:**
    - Choose a task from the **task dropdown** if using your own text.
    - **Important:** This ensures correct model response.
-
 3. **Avoid Conflicts:**
    - Don't select a sample text if using your own text.
-
 4. **Select Nigerian Language:**
    - If prompted, choose the Nigerian language (it represents the input/base language for diacritization/cleaning, target language for translation).
-
 5. **Generate Output:**
    - Click the Generate button.
-
 6. **Translation Tips:**
    - English as the target language gives the best results.
    - You can also test inter-language translation i.e yoruba to igbo
-
 7. **Performance Note:**
    - The model's performance varies due to its size and training data. It performs best on text generation and translation.
    - For other tasks, try multiple times if model's output is not optimal (This is due to the generator's sampling parameter settings).
    - It's best to read/understand/translate the model's output completely first. Model can sometimes fail to stop generation after providing correct answers.
-
 8. **Other Tips:**
    - Use simple instructions for instruction following.
    - For question answering and generation, follow the structure in the corresponding sample text.
@@ -96,12 +89,10 @@ st.title("SabiYarn-125M : Generates text in multiple Nigerian languages.")
 
 st.write("**Supported Languages: English, Yoruba, Igbo, Hausa, Pidgin, Efik, Urhobo, Fulfulde, Fulah. \nResults may not be coherent for less represented languages (i.e Efik, \
 Urhobo, Fulfulde, Fulah).**")
-st.write("**It takes a while (~25s) to return an output on the first 'generate' click. Avg response time: 1-2s
+st.write("**It takes a while (~25s) to return an output on the first 'generate' click. Avg response time: 1-2s on GPU, 40s on CPU**")
 st.write("**Model outputs 50 tokens as default. Adjust in the side bar.**")
 st.write("**For convenience, you can use chatgpt to provide input text and translate/evaluate model output.**")
 st.write("-" * 50)
-
-
 popular_topics = [
     "Sports", "Entertainment", "Politics", "Travel", "Technology",
     "Health", "Business", "Science", "Education", "Lifestyle",
@@ -109,14 +100,12 @@ popular_topics = [
     "History", "Law", "Literature", "Music", "News","Africa",
     "Philosophy", "Religion", "Society", "World"
 ]
-
 async def assign_topic(generated_text, topic_list=popular_topics):
     lower_generated_text = generated_text.lower()
     for topic in topic_list:
         if topic.lower() in lower_generated_text:
             return topic.title()
     return generated_text
-
     
 async def generate_from_api(user_input, generation_config):
     urls = [
@@ -124,7 +113,6 @@ async def generate_from_api(user_input, generation_config):
         "https://daveokpare--sabiyarn-fastapi-app.modal.run/predict",
         "https://damilojohn--sabiyarn-fastapi-app.modal.run/predict"
     ]
-
     payload = {
         "prompt": user_input,
         "config": generation_config
@@ -133,7 +121,6 @@ async def generate_from_api(user_input, generation_config):
     headers = {
         'Content-Type': 'application/json'
     }
-
     async with aiohttp.ClientSession() as session:
         for url in urls:
             try:
@@ -144,9 +131,7 @@ async def generate_from_api(user_input, generation_config):
                         print(f"Failed to fetch from {url} with status code {response.status}")
             except Exception as e:
                 print(f"Error fetching from {url}: {e}")
-
     return "FAILED"
-
 # Sample texts
 sample_texts = {
     "select":"",
@@ -174,7 +159,6 @@ sample_texts = {
     # "clean this text": "Abin mamaki ne aikin da shugabaZn HNajeriya ybake yi. kCiF 39gaba Tda haRkGa sir!",
     "headline of this text": '** Sylvain Itté French ambassador don comot Niger Republic **. Sylvain Itté, di French ambassador for Niger don comot Niamey and currently e dey for flight from Ndjamena to Paris. Sylvain Itté, di French ambassador for Niger don comot Niamey very early dis morning and currently e dey for flight from Ndjamena to Paris.\n\nDi military detain Bazoum and im family for di presidential palace. Niger na former French colony, and France still get 1,500 sojas for di African country.\n\n"France don decide to withdraw dia ambassador. In di next hours our ambassador and several diplomats go return to France," Oga Macron tok.\n\nE add say di military co-operation dey "over" and French troops go leave in "di months to come".\n\n"Dis Sunday we celebrate one new step towards di sovereignty of Niger," di junta tok, for one statement wey AFP news agency quote.\n\nDi decision by Paris dey come afta months of hostility and protest against di presence of French for di kontri, wit regular demonstrations for di capital Niamey.\n\nDi move don scata France operations against Islamist militants for di wider Sahel region and Paris influence for there. But oga Macron tok say "putschists no go hold France hostage,"'
 }
-
 instruction_wrap = {
     # "Translate 'Often, all Yoruba children...' to Yoruba":"<translate> Often, all Yoruba children take pride in speaking the Yoruba language. <yor>",
     "Me ya nuna?":"<prompt> Me ya nuna? <response>:",
@@ -190,14 +174,13 @@ instruction_wrap = {
     "Kai found one for sale online but it was too much money for her. Keeping the provided context in mind, please answer the subsequent question: What does Kai need to do before this? A. cheaper B. Open up her laptop C. save money":"<prompt> Kai found one for sale online but it was too much money for her. Keeping the provided context in mind, please answer the subsequent question: What does Kai need to do before this? A. cheaper B. Open up her laptop C. save money <response>:",
     "Speak Yoruba": "<prompt> Speak Yoruba <response>:",
     "Please narrate a short story in yoruba":"<prompt> Please narrate a short story in yoruba <response>:",
-    "Afghanistan has around 150 radio stations and over 50 television stations, which includes the state-owned RTA TV and various private channels such as TOLO and Shamshad TV. The first Afghan newspaper was published in 1906 and there are hundreds of print outlets today. By the 1920s, Radio Kabul was broadcasting local radio services. Television programs began airing in the early 1970s. Voice of America, BBC, and Radio Free Europe/Radio Liberty (RFE/RL) broadcast in both of Afghanistan's official languages.\n Considering this context, what question would you ask?":"<prompt> Afghanistan has around 150 radio stations and over 50 television stations, which includes the state-owned RTA TV and various private channels such as TOLO and Shamshad TV. The first Afghan newspaper was published in 1906 and there are hundreds of print outlets today. By the 1920s, Radio Kabul was broadcasting local radio services. Television programs began airing in the early 1970s. Voice of America, BBC, and Radio Free Europe/Radio Liberty (RFE/RL) broadcast in both of Afghanistan's official languages.\n Considering this context, what question would you ask? <response>:"
+    "Afghanistan has around 150 radio stations and over 50 television stations, which includes the state-owned RTA TV and various private channels such as TOLO and Shamshad TV. The first Afghan newspaper was published in 1906 and there are hundreds of print outlets today. By the 1920s, Radio Kabul was broadcasting local radio services. Television programs began airing in the early 1970s. Voice of America, BBC, and Radio Free Europe/Radio Liberty (RFE/RL) broadcast in both of Afghanistan's official languages.\n Considering this context, what question would you ask?":"<prompt> Afghanistan has around 150 radio stations and over 50 television stations, which includes the state-owned RTA TV and various private channels such as TOLO and Shamshad TV. The first Afghan newspaper was published in 1906 and there are hundreds of print outlets today. By the 1920s, Radio Kabul was broadcasting local radio services. Television programs began airing in the early 1970s. Voice of America, BBC, and Radio Free Europe/Radio Liberty (RFE/RL) broadcast in both of Afghanistan's official languages.\n Considering this context, what question would you ask? <response>:",
     "Anyi na-echefu oke ike." : "<classify> Anyi na-echefu oke ike. <sentiment>",
     "Abin mamaki ne aikin da shugabaZn HNajeriya ybake yi. kCiF 39gaba Tda haRkGa sir!": "<clean> Abin mamaki ne aikin da shugabaZn HNajeriya ybake yi. kCiF 39gaba Tda haRkGa sir! <pcm>",
     "E sun, Alaga, fun ise amalayi ti e n se ni Naijiria. E maa ba a lo, egbon!": "<diacritize> E sun, Alaga, fun ise amalayi ti e n se ni Naijiria. E maa ba a lo, egbon!  <yor>",
     "Africa Free Trade Zone: Kò sí ìdènà láti kó ọjà láti orílẹ̀èdè kan sí òmíràn": "<classify> Africa Free Trade Zone: Kò sí ìdènà láti kó ọjà láti orílẹ̀èdè kan sí òmíràn <topic>",
     '** Sylvain Itté French ambassador don comot Niger Republic **. Sylvain Itté, di French ambassador for Niger don comot Niamey and currently e dey for flight from Ndjamena to Paris. Sylvain Itté, di French ambassador for Niger don comot Niamey very early dis morning and currently e dey for flight from Ndjamena to Paris.\n\nDi military detain Bazoum and im family for di presidential palace. Niger na former French colony, and France still get 1,500 sojas for di African country.\n\n"France don decide to withdraw dia ambassador. In di next hours our ambassador and several diplomats go return to France," Oga Macron tok.\n\nE add say di military co-operation dey "over" and French troops go leave in "di months to come".\n\n"Dis Sunday we celebrate one new step towards di sovereignty of Niger," di junta tok, for one statement wey AFP news agency quote.\n\nDi decision by Paris dey come afta months of hostility and protest against di presence of French for di kontri, wit regular demonstrations for di capital Niamey.\n\nDi move don scata France operations against Islamist militants for di wider Sahel region and Paris influence for there. But oga Macron tok say "putschists no go hold France hostage,"': '<title> ** Sylvain Itté French ambassador don comot Niger Republic **. Sylvain Itté, di French ambassador for Niger don comot Niamey and currently e dey for flight from Ndjamena to Paris. Sylvain Itté, di French ambassador for Niger don comot Niamey very early dis morning and currently e dey for flight from Ndjamena to Paris.\n\nDi military detain Bazoum and im family for di presidential palace. Niger na former French colony, and France still get 1,500 sojas for di African country.\n\n"France don decide to withdraw dia ambassador. In di next hours our ambassador and several diplomats go return to France," Oga Macron tok.\n\nE add say di military co-operation dey "over" and French troops go leave in "di months to come".\n\n"Dis Sunday we celebrate one new step towards di sovereignty of Niger," di junta tok, for one statement wey AFP news agency quote.\n\nDi decision by Paris dey come afta months of hostility and protest against di presence of French for di kontri, wit regular demonstrations for di capital Niamey.\n\nDi move don scata France operations against Islamist militants for di wider Sahel region and Paris influence for there. But oga Macron tok say "putschists no go hold France hostage,"  <headline>',
 }
-
 # instruction_wrap = {
 #     # "Translate 'Often, all Yoruba children...' to Yoruba":"<translate> Often, all Yoruba children take pride in speaking the Yoruba language. <yor>",
 #     "Tell me a story in pidgin": "<prompt> Tell me a story in pidgin <response>:",
@@ -211,7 +194,6 @@ instruction_wrap = {
 #     "what is the topic of this text": "<classify> Africa Free Trade Zone: Kò sí ìdènà láti kó ọjà láti orílẹ̀èdè kan sí òmíràn <topic>",
 #     'headline of this text': '<title> ** Sylvain Itté French ambassador don comot Niger Republic **. Sylvain Itté, di French ambassador for Niger don comot Niamey and currently e dey for flight from Ndjamena to Paris. Sylvain Itté, di French ambassador for Niger don comot Niamey very early dis morning and currently e dey for flight from Ndjamena to Paris.\n\nDi military detain Bazoum and im family for di presidential palace. Niger na former French colony, and France still get 1,500 sojas for di African country.\n\n"France don decide to withdraw dia ambassador. In di next hours our ambassador and several diplomats go return to France," Oga Macron tok.\n\nE add say di military co-operation dey "over" and French troops go leave in "di months to come".\n\n"Dis Sunday we celebrate one new step towards di sovereignty of Niger," di junta tok, for one statement wey AFP news agency quote.\n\nDi decision by Paris dey come afta months of hostility and protest against di presence of French for di kontri, wit regular demonstrations for di capital Niamey.\n\nDi move don scata France operations against Islamist militants for di wider Sahel region and Paris influence for there. But oga Macron tok say "putschists no go hold France hostage,"  <headline>',
 # }
-
 # Task options
 task_options = {
     "select": "{}",
@@ -226,7 +208,6 @@ task_options = {
     "Question-Answering" : "<prompt> {} <response>:",
     "Text Cleaning": "<clean> {} "
 }
-
 # Language options for diacritize, translation and clean tasks
 language_options = {
     "select": "",
@@ -239,38 +220,31 @@ language_options = {
     "Urhobo": "<urh>",
     "Fulah": "<ful>"
 }
-
         
 # Dropdown for sample text
 sample_text = st.selectbox("Select a sample text to test the model:", list(sample_texts.keys()))
-
 # Dropdown for tasks
 task = st.selectbox("Select a task for the model:", list(task_options.keys()))
-
 # Conditionally show language options dropdown for diacritize and clean tasks
 if task in ["Text Diacritization", "Text Cleaning", "Translation"]:
     language = st.selectbox("Select a Nigerian language:", list(language_options.keys()))
     task_value = f"{task_options[task]} {language_options[language]}"
 else:
     task_value = task_options[task]
-
 def wrap_text(text, task_value):
     tasks = ["<classify>", "<prompt>", "<clean>", "<title>", "<diacritize>", "<translate>"]
     if any(task in text for task in tasks):
         return text
     return task_value.format(text)
-
 # Text input 
 user_input = st.text_area("Enter text below **(PLEASE, FIRST READ ALL INSTRUCTIONS IN THE SIDEBAR CAREFULLY FOR THE BEST EXPERIENCE)**: ", sample_texts.get(sample_text, sample_text))
-
 if task == "select":
     user_input = instruction_wrap.get(user_input, user_input)
                                      
 print("Final user input: ", user_input)
-
 if st.button("Generate"):
     if user_input:
-        with st.spinner("Please wait..."):")
+        with st.spinner("Please wait..."):
             wrapped_input = wrap_text(user_input, task_value)
             print("wrapped_input: ", wrapped_input)
             generation_config["max_new_tokens"]= min(max_new_tokens, 1024 - len(tokenizer.tokenize(wrapped_input)))
@@ -303,7 +277,6 @@ if st.button("Generate"):
                 generated_text  = generated_text[:15]
                 print("split", generated_text.split(" ")[0],  re.split(r"\.|\n|\*\*|\*", generated_text)[0], generated_text.split(" "))
                 generated_text = re.split(r"\.|\n|\*\*|\*", generated_text)[0]
-
                 generated_text = asyncio.run(assign_topic(generated_text))
                 
             elif task == "Translation":