Update app.py

app.py

@@ -1,19 +1,32 @@
 from fastapi import FastAPI
 import time
+time = time.time()
 import torch
 
 from transformers import AutoModelForCausalLM, AutoTokenizer
 device = "cpu" # the device to load the model onto
-
+time1 = time.time()
 model = AutoModelForCausalLM.from_pretrained(
     "Qwen/Qwen2-0.5B-Instruct",
     torch_dtype="auto",
     device_map="auto"
 )
+time2 = time.time()
+print(time2-time1)
 tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen2-0.5B-Instruct")
-
+time3 = time.time()
+print(time3-time1)
+model1 = AutoModelForCausalLM.from_pretrained(
+    "Qwen/Qwen2-1.5B-Instruct",
+    torch_dtype="auto",
+    device_map="auto"
+)
+tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen2-1.5B-Instruct")
+time4 = time.time()
+print(time4-time3)
 app = FastAPI()
-
+time5 = time.time()
+print(time5-time4)
 @app.get("/")
 async def read_root():
     return {"Hello": "World!"}
@@ -22,25 +35,35 @@ messages = [
         {"role": "system", "content": "You are a helpful assistant, Sia, developed by Sushma. You will response in polity and brief."},
         {"role": "user", "content": "I'm Alok. Who are you?"},
         {"role": "assistant", "content": "I am Sia, a small language model created by Sushma."},
-        {"role": "user", "content": "Hi, How are you?"}
+        {"role": "user", "content": "How are you?"}
 ]
+time1 = time.time()
 text = tokenizer.apply_chat_template(
     messages,
     tokenize=False,
     add_generation_prompt=True
 )
+time2 = time.time()
+print(time2-time1)
 model_inputs = tokenizer([text], return_tensors="pt").to(device)
-    
+time3 = time.time()
+print(time3-time2)    
 generated_ids = model.generate(
     model_inputs.input_ids,
     max_new_tokens=64
 )
+time4 = time.time()
+print(time4-time3)
 generated_ids = [
     output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
 ]
-    
+time5 = time.time()
+print(time5-time4)    
 response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
 print(response)
+time6 = time.time()
+print(time6-time5)
+
 end_time = time.time()
 time_taken = end_time - start_time 
 print(time_taken)
@@ -53,7 +76,7 @@ async def read_droot():
         {"role": "system", "content": "You are a helpful assistant, Sia, developed by Sushma. You will response in polity and brief."},
         {"role": "user", "content": "I'm Alok. Who are you?"},
         {"role": "assistant", "content": "I am Sia, a small language model created by Sushma."},
-        {"role": "user", "content": "Hi, How are you?"}
+        {"role": "user", "content": "How are you?"}
     ]
     text = tokenizer.apply_chat_template(
         messages,
@@ -77,3 +100,34 @@ async def read_droot():
     print(time_taken)
     return {"Hello": "World!"}
 
+@app.get("/text")
+async def read_droot():
+    starttime = time.time()
+    messages = [
+        {"role": "system", "content": "You are a helpful assistant, Sia, developed by Sushma. You will response in polity and brief."},
+        {"role": "user", "content": "I'm Alok. Who are you?"},
+        {"role": "assistant", "content": "I am Sia, a small language model created by Sushma."},
+        {"role": "user", "content": "How are you?"}
+    ]
+    text = tokenizer.apply_chat_template(
+        messages,
+        tokenize=False,
+        add_generation_prompt=True
+    )
+    model_inputs = tokenizer1([text], return_tensors="pt").to(device)
+        
+    generated_ids = model1.generate(
+        model_inputs.input_ids,
+        max_new_tokens=64
+    )
+    generated_ids = [
+        output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
+    ]
+        
+    response = tokenizer1.batch_decode(generated_ids, skip_special_tokens=True)[0]
+    print(response)
+    end_time = time.time()
+    time_taken = end_time - starttime 
+    print(time_taken)
+    return {"Hello": "World!"}
+    #return {response: time}