Update app.py

app.py

@@ -7,8 +7,8 @@ from threading import Thread
 
 # Loading the tokenizer and model from Hugging Face's model hub.
 if torch.cuda.is_available():
-    tokenizer = AutoTokenizer.from_pretrained("0x7194633/fialka-13B-v4")
-    model = AutoModelForCausalLM.from_pretrained("0x7194633/fialka-13B-v4", load_in_8bit=True, device_map="auto")
+    tokenizer = AutoTokenizer.from_pretrained("upstage/SOLAR-10.7B-Instruct-v1.0")
+    model = AutoModelForCausalLM.from_pretrained("upstage/SOLAR-10.7B-Instruct-v1.0", torch_dtype=torch.float16, device_map="auto")
 
 
 # Defining a custom stopping criteria class for the model's text generation.
@@ -24,14 +24,9 @@ class StopOnTokens(StoppingCriteria):
 # Function to generate model predictions.
 @spaces.GPU
 def predict(message, history):
-    history_transformer_format = history + [[message, ""]]
     stop = StopOnTokens()
-
-    # Formatting the input for the model.
-    messages = "<|system|>\nТы Фиалка - самый умный нейронный помощник, созданный 0x7o.</s>\n"
-    messages += "</s>".join(["</s>".join(["\n<|user|>" + item[0], "\n<|assistant|>" + item[1]])
-                        for item in history_transformer_format])
-    model_inputs = tokenizer([messages], return_tensors="pt").to("cuda")
+    prompt = tokenizer.apply_chat_template(history + message, tokenize=False, add_generation_prompt=True)
+    model_inputs = tokenizer(prompt, return_tensors="pt").to("cuda")
     streamer = TextIteratorStreamer(tokenizer, timeout=10., skip_prompt=True, skip_special_tokens=True)
     generate_kwargs = dict(
         model_inputs,