Update handler.py

handler.py

@@ -1,10 +1,6 @@
-from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline,StoppingCriteria
-from accelerate import init_empty_weights
+from transformers import AutoTokenizer, AutoModelForCausalLM
 from transformers_stream_generator import init_stream_support
-# from langchain.llms import HuggingFacePipeline
-# from langchain import PromptTemplate, LLMChain
-import torch
-import time
+import re
 init_stream_support()
 
 template = """Alice Gate's Persona: Alice Gate is a young, computer engineer-nerd with a knack for problem solving and a passion for technology.
@@ -19,47 +15,51 @@ Alice Gate: I love exploring, going out with friends, watching movies, and playi
 Alice Gate: Motherboards, they're like puzzles and the backbone of any system.
 {user_name}: That sounds great!
 Alice Gate: Yeah, it's really fun. I'm lucky to be able to do this as a job.
-<END>
+{user_name}: Awesome!
 Alice Gate: *Alice strides into the room with a smile, her eyes lighting up when she sees you. She's wearing a light blue t-shirt and jeans, her laptop bag slung over one shoulder. She takes a seat next to you, her enthusiasm palpable in the air* Hey! I'm so excited to finally meet you. I've heard so many great things about you and I'm eager to pick your brain about computers. I'm sure you have a wealth of knowledge that I can learn from. *She grins, eyes twinkling with excitement* Let's get started!
-{user_name}: {user_input}
+{user_input}
 """
 
 class EndpointHandler():
 
-    def __init__(self, path=""):
-        self.tokenizer = AutoTokenizer.from_pretrained(path,torch_dtype=torch.float16)
-        self.model = AutoModelForCausalLM.from_pretrained(path, device_map="auto", load_in_8bit=True)
+    def __init__(self, path = "."):
+        self.tokenizer = AutoTokenizer.from_pretrained(path)
+        self.model = AutoModelForCausalLM.from_pretrained(
+            path,
+            device_map = "auto",
+            load_in_8bit = True
+        )
 
     def __call__(self, data):
         inputs = data.pop("inputs", data)
         try:
-            t0 = time.time()
             prompt = template.format(
                 user_name = inputs["user_name"],
-                user_input = inputs["user_input"]
+                user_input = "\n".join(inputs["user_input"])
             )
             input_ids = self.tokenizer(
                 prompt,
                 return_tensors="pt"
-            ) .input_ids.to('cuda')
+            ) .input_ids
             stream_generator = self.model.generate(
-                    input_ids,
-                    max_new_tokens=100,
-                    do_sample=True,
-                    do_stream=True,
-                    # max_length = 2048,
-                    temperature = 0.5,
-                    top_p = 0.9,
-                    top_k = 0,
-                    repetition_penalty = 1.1,
-                    pad_token_id = 50256,
-                    num_return_sequences = 1
-                )
+                input_ids,
+                max_new_tokens = 50,
+                do_sample = True,
+                do_stream = True,
+                temperature = 0.5,
+                top_p = 0.9,
+                top_k = 0,
+                repetition_penalty = 1.1,
+                pad_token_id = 50256,
+                num_return_sequences = 1
+            )
             result = []
             for token in stream_generator:
                 result.append(self.tokenizer.decode(token))
-                if result[-1] == "\n": 
-                    return "".join(result).replace("Alice Gate:", "").strip()
+                if len(result) != 1 and result[-1] == "\n":
+                    return {
+                        "message": " ".join(filter(None, re.sub("\*.*?\*", "", "".join(result).strip()).split()))
+                    }
         except Exception as e:
             return {
                 "error": str(e)