Update app.py

app.py

@@ -8,7 +8,6 @@ import json
 import torch
 import re
 import nest_asyncio
-from hashlib import md5
 
 nest_asyncio.apply()
 
@@ -106,21 +105,6 @@ st.write("**It might take a while (~25s) to return an output on the first 'gener
 st.write("**For convenience, you can use chatgpt to copy text and evaluate model output.**")
 st.write("-" * 50)
 
-# async def generate_from_api(user_input, generation_config):
-#     url = "https://pauljeffrey--sabiyarn-fastapi-app.modal.run/predict"
-    
-#     payload = {
-#         "prompt": user_input,
-#         "config": generation_config
-#     }
-    
-#     headers = {
-#         'Content-Type': 'application/json'
-#     }
-    
-#     async with aiohttp.ClientSession() as session:
-#         async with session.post(url, headers=headers, json=payload) as response:
-#             return await response.text()
 
 async def generate_from_api(user_input, generation_config):
     urls = [
@@ -151,15 +135,6 @@ async def generate_from_api(user_input, generation_config):
 
     return "FAILED"
 
-def generate_cache_key(user_input, generation_config):
-    key_data = f"{user_input}_{json.dumps(generation_config, sort_keys=True)}"
-    return md5(key_data.encode()).hexdigest()
-
-@st.cache_data(show_spinner=False)
-def get_cached_response(user_input, generation_config):
-    return asyncio.run(generate_from_api(user_input, generation_config))
-
-
 # Sample texts
 sample_texts = {
     "select":"",
@@ -253,22 +228,9 @@ if st.button("Generate"):
         print("wrapped_input: ", wrapped_input)
         generation_config["max_new_tokens"]= min(max_new_tokens, 1024 - len(tokenizer.tokenize(wrapped_input)))
         start_time = time.time()
-        # try:
-            # Attempt the asynchronous API call
-        generation_config["max_new_tokens"] = min(max_new_tokens, 1024 - len(tokenizer.tokenize(wrapped_input)))
-        # generated_text = asyncio.run(generate_from_api(wrapped_input, generation_config))
-        cache_key = generate_cache_key(wrapped_input, generation_config)
-        generated_text = get_cached_response(wrapped_input, generation_config)
         
-        # loop = asyncio.new_event_loop()
-        # asyncio.set_event_loop(loop)
-        # generated_text = loop.run_until_complete(generate_from_api(wrapped_input, generation_config))
-        # except Exception as e:
-            # print(f"API call failed: {e}. Using local model for text generation.")                
-            # Use the locally loaded model for text generation
-            # input_ids = tokenizer(wrapped_input, return_tensors="pt")["input_ids"].to(device)
-            # output = model.generate(input_ids, **generation_config)
-            # generated_text = tokenizer.decode(output[0], skip_special_tokens=True)
+        generation_config["max_new_tokens"] = min(max_new_tokens, 1024 - len(tokenizer.tokenize(wrapped_input)))
+        generated_text = asyncio.run(generate_from_api(wrapped_input, generation_config))
             
         if generated_text == "FAILED":
             input_ids = tokenizer(wrapped_input, return_tensors="pt")["input_ids"].to(device)