Update app.py

app.py

@@ -1,6 +1,6 @@
 from fastapi import FastAPI
 import time
-
+import torch
 
 from transformers import AutoModelForCausalLM, AutoTokenizer
 device = "cpu" # the device to load the model onto
@@ -75,4 +75,39 @@ async def read_droot():
     end_time = time.time()
     time_taken = end_time - starttime 
     print(time_taken)
-    return {"Hello": "World!"}
+    return {"Hello": "World!"}
+
+
+@app.get("/text")
+async def readdroot():
+    starttime = time.time()
+    messages = [
+        {"role": "system", "content": "You are a helpful assistant, Sia. You are developed by Sushma. You will response in polity and brief."},
+        {"role": "user", "content": "Who are you?"},
+        {"role": "assistant", "content": "I am Sia, a small language model created by Sushma. I am here to assist you."},
+        {"role": "user", "content": "Hi, How are you?"}
+    ]
+    text = tokenizer.apply_chat_template(
+        messages,
+        tokenize=False,
+        add_generation_prompt=True
+    )
+    model_inputs = tokenizer([text], return_tensors="pt").to(device)
+
+    with torch.no_grad():  # Disable gradient calculation
+        generated_ids = model.generate(
+            model_inputs.input_ids,
+            max_new_tokens=64,  # Adjust this based on needs
+            use_cache=False # Use cached activations if applicable
+        )
+
+    generated_ids = [
+        output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
+    ]
+        
+    response = tokenizer.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!"}