Update app.py

app.py

@@ -1,84 +1,68 @@
-from transformers import AutoConfig, AutoTokenizer, AutoModelForSeq2SeqLM, AutoModelForCausalLM, MistralForCausalLM
-from peft import PeftModel, PeftConfig
+import os
+import math
+import transformers
+from transformers import AutoModelForCausalLM, AutoTokenizer
 import torch
 import gradio as gr
-import random
-from textwrap import wrap
+import sentencepiece
 
+title = "Welcome to Tonic's 🐋🐳Orca-2-13B (in 8bit)!"
+description = "You can use [🐋🐳microsoft/Orca-2-13b](https://huggingface.co/microsoft/Orca-2-13b) via API using Gradio by scrolling down and clicking Use 'Via API' or privately by [cloning this space on huggingface](https://huggingface.co/spaces/Tonic1/TonicsOrca2?duplicate=true) . [Join my active builders' server on discord](https://discord.gg/VqTxc76K3u). Big thanks to the HuggingFace Organisation for the Community Grant."
 
-title = "👋🏻토닉의 미스트랄메드 채팅에 오신 것을 환영합니다🚀👋🏻Welcome to Tonic's MistralMed Chat🚀"
-description = "이 공간을 사용하여 현재 모델을 테스트할 수 있습니다. [(Tonic/MistralMed)](https://huggingface.co/Tonic/MistralMed) 또는 이 공간을 복제하고 로컬 또는 🤗HuggingFace에서 사용할 수 있습니다. [Discord에서 함께 만들기 위해 Discord에 가입하십시오](https://discord.gg/VqTxc76K3u). You can use this Space to test out the current model [(Tonic/MistralMed)](https://huggingface.co/Tonic/MistralMed) or duplicate this Space and use it locally or on 🤗HuggingFace. [Join me on Discord to build together](https://discord.gg/VqTxc76K3u)."
-examples = [["[Question:] What is the proper treatment for buccal herpes?", "You are a medicine and public health expert, you will receive a question, answer the question, and provide a complete answer"]]
+# os.environ['PYTORCH_CUDA_ALLOC_CONF'] = 'max_split_size_mb:50'
+device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
+model_name = "microsoft/Orca-2-13b"
+# offload_folder = './model_weights'
 
-base_model_id = "mistralai/Mistral-7B-v0.1"
-model_directory = "Tonic/mistralmed"
-device = "cuda" if torch.cuda.is_available() else "cpu"
+# if not os.path.exists(offload_folder):
+#     os.makedirs(offload_folder)
 
-def wrap_text(text, width=90):
-    lines = text.split('\n')
-    wrapped_lines = [textwrap.fill(line, width=width) for line in lines]
-    wrapped_text = '\n'.join(wrapped_lines)
-    return wrapped_text
+tokenizer = AutoTokenizer.from_pretrained(model_name, use_fast=False)
+model = transformers.AutoModelForCausalLM.from_pretrained(model_name, device_map="auto", load_in_8bit=True)
 
-def multimodal_prompt(user_input, system_prompt="You are an expert medical analyst:"):
+class OrcaChatBot:
+    def __init__(self, model, tokenizer, system_message="You are Orca, an AI language model created by Microsoft. You are a cautious assistant. You carefully follow instructions. You are helpful and harmless and you follow ethical guidelines and promote positive behavior."):
+        self.model = model
+        self.tokenizer = tokenizer
+        self.system_message = system_message
 
-    formatted_input = f"<s>[INST]{system_prompt} {user_input}[/INST]</s>"
+    def predict(self, user_message, temperature=0.4, max_new_tokens=70, top_p=0.99, repetition_penalty=1.9):
+        prompt = f"<|im_start|>system\n{self.system_message}<|im_end|>\n<|im_start|>user\n{user_message}<|im_end|>\n<|im_start|>assistant"
+        inputs = self.tokenizer(prompt, return_tensors='pt', add_special_tokens=False)
+        input_ids = inputs["input_ids"].to(self.model.device)
 
-    encodeds = tokenizer(formatted_input, return_tensors="pt", add_special_tokens=False)
-    model_inputs = encodeds.to(device)
+        output_ids = self.model.generate(
+            input_ids,
+            max_length=input_ids.shape[1] + max_new_tokens,
+            temperature=temperature,
+            top_p=top_p,
+            repetition_penalty=repetition_penalty,
+            pad_token_id=self.tokenizer.eos_token_id,
+            do_sample=True 
+        )
 
-    output = model.generate(
-        **model_inputs,
-        max_length=max_length,
-        use_cache=True,
-        early_stopping=True,
-        bos_token_id=model.config.bos_token_id,
-        eos_token_id=model.config.eos_token_id,
-        pad_token_id=model.config.eos_token_id,
-        temperature=0.1,
-        do_sample=False
-    )
+        response = self.tokenizer.decode(output_ids[0], skip_special_tokens=True)
 
-    # Decode the response
-    response_text = tokenizer.decode(output[0], skip_special_tokens=True)
+        return response
 
-    return response_text
+Orca_bot = OrcaChatBot(model, tokenizer)
 
-
-tokenizer = AutoTokenizer.from_pretrained("mistralai/Mistral-7B-v0.1", trust_remote_code=True, padding_side="left")
-tokenizer.pad_token = tokenizer.eos_token
-tokenizer.padding_side = 'left'
-peft_config = PeftConfig.from_pretrained("Tonic/mistralmed")
-peft_model = MistralForCausalLM.from_pretrained("mistralai/Mistral-7B-v0.1")
-peft_model = PeftModel.from_pretrained(peft_model, "Tonic/mistralmed")
-peft_model = peft_model.to(torch.bfloat16)
-peft_model = peft_model.to(device)
-
-class ChatBot:
-    def __init__(self):
-        self.history = []
-
-class ChatBot:
-    def __init__(self):
-        self.history = []
-
-    def predict(self, user_input, system_prompt="You are an expert medical analyst:"):
-        formatted_input = f"<s>[INST]{system_prompt} {user_input}[/INST]"
-        user_input_ids = tokenizer.encode(formatted_input, return_tensors="pt")
-        user_input_ids = user_input_ids.to(device)
-        response = peft_model.generate(input_ids=user_input_ids, max_length=256, pad_token_id=tokenizer.eos_token_id)
-        response_text = tokenizer.decode(response[0], skip_special_tokens=True)
-        
-        return response_text  
-
-bot = ChatBot()
+def gradio_predict(user_message, system_message, max_new_tokens, temperature, top_p, repetition_penalty):
+    full_message = f"{system_message}\n{user_message}" if system_message else user_message
+    return Orca_bot.predict(full_message, temperature, max_new_tokens, top_p, repetition_penalty)
 
 iface = gr.Interface(
-    fn=bot.predict,
+    fn=gradio_predict,
     title=title,
     description=description,
-    examples=examples,
-    inputs=["text", "text"], 
+    inputs=[
+        gr.Textbox(label="Your Message", type="text", lines=3),
+        gr.Textbox(label="Introduce a Character Here or Set a Scene (system prompt)", type="text", lines=2),
+        gr.Slider(label="Max new tokens", value=125, minimum=25, maximum=256, step=1),
+        gr.Slider(label="Temperature", value=0.1, minimum=0.05, maximum=1.0, step=0.05),
+        gr.Slider(label="Top-p (nucleus sampling)", value=0.90, minimum=0.01, maximum=0.99, step=0.05),
+        gr.Slider(label="Repetition penalty", value=1.9, minimum=1.0, maximum=2.0, step=0.05)
+    ],
     outputs="text",
     theme="ParityError/Anime"
 )