Update app.py

app.py

@@ -1,62 +1,77 @@
+import spaces
+import torch
+
 import gradio as gr
-from huggingface_hub import InferenceClient
+from huggingface_hub import snapshot_download
+from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig
 
+
+model = None
+model_id = "nazimali/Mistral-Nemo-Kurdish-Instruct"
+
+infer_prompt = """Li jêr rêwerzek heye ku peywirek rave dike, bi têketinek ku çarçoveyek din peyda dike ve tê hev kirin. Bersivek ku daxwazê ​​bi guncan temam dike binivîsin.
+### Telîmat:
+{}
+### Têketin:
+{}
+### Bersiv:
 """
-For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
-"""
-client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
+
+snapshot_download("nazimali/Mistral-Nemo-Kurdish")
+snapshot_download(repo_id=model_id)
 
 
+@spaces.GPU
 def respond(
     message,
     history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
 ):
-    messages = [{"role": "system", "content": system_message}]
+    global model
 
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
+    if model is None:
+        bnb_config = BitsAndBytesConfig(
+            load_in_4bit=True,
+            bnb_4bit_use_double_quant=True,
+            bnb_4bit_quant_type="nf4",
+            bnb_4bit_compute_dtype=torch.bfloat16,
+        )
 
-    messages.append({"role": "user", "content": message})
+        model = AutoModelForCausalLM.from_pretrained(
+            model_id,
+            quantization_config=bnb_config,
+            device_map="auto",
+        )
+        tokenizer = AutoTokenizer.from_pretrained(model_id)
 
-    response = ""
+        model.eval()
 
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
+    prompt = infer_prompt.format("tu arîkarek alîkar î", message)
 
-        response += token
-        yield response
+    input_ids = tokenizer(
+        prompt,
+        return_tensors="pt",
+        add_special_tokens=False,
+        return_token_type_ids=False,
+    ).to("cuda")
 
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
-demo = gr.ChatInterface(
-    respond,
-    additional_inputs=[
-        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
-        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
-        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
-        gr.Slider(
-            minimum=0.1,
-            maximum=1.0,
-            value=0.95,
-            step=0.05,
-            label="Top-p (nucleus sampling)",
-        ),
-    ],
-)
+    with torch.inference_mode():
+        generated_ids = model.generate(
+            **input_ids,
+            max_new_tokens=120,
+            do_sample=True,
+            temperature=0.7,
+            top_p=0.7,
+            num_return_sequences=1,
+            pad_token_id=tokenizer.pad_token_id,
+            eos_token_id=tokenizer.eos_token_id,
+        )
+
+    decoded_output = tokenizer.batch_decode(generated_ids)[0]
+
+    return decoded_output.replace(prompt, "").replace("</s>", "")
+
+
+demo = gr.ChatInterface(respond, examples=["سڵاو ئەلیکوم، چۆنیت؟", "Selam alikum, tu çawa yî?"], title="Mistral Nemo Kurdish Instruct")
 
 
 if __name__ == "__main__":