Add application file

app.py

@@ -0,0 +1,152 @@
+import gradio as gr
+import re
+
+
+def predict_house_price(description, location, area):
+    FastLanguageModel.for_inference(model)  # Enable native 2x faster inference
+    messages = [  # Change below!
+        {
+            "role": "user",
+            "content": f"Mô tả: {description}\nĐịa chỉ: {location}\nDiện tích: {area}",
+        },
+    ]
+    input_ids = tokenizer.apply_chat_template(
+        messages,
+        add_generation_prompt=True,
+        return_tensors="pt",
+    ).to("cuda")
+
+    # from transformers import TextStreamer
+    # text_streamer = TextStreamer(tokenizer, skip_prompt = True)
+    # _ = model.generate(input_ids, streamer = text_streamer, max_new_tokens = 128, pad_token_id = tokenizer.eos_token_id)
+    # return extract_price(_)
+    output = model.generate(
+        input_ids, max_new_tokens=256, pad_token_id=tokenizer.eos_token_id
+    )
+
+    decoded = tokenizer.decode(output[0], skip_special_tokens=True)
+    return f"{extract_price(decoded)} tỷ"
+
+
+def extract_price(decoded):
+    # match number like 6.950.000.000, or 6950000000, or 6.95B, etc.
+    price_pattern = r"[\d\.\,]+"
+    matches = re.findall(price_pattern, decoded)
+    last_item = matches[-1]
+    return last_item
+
+
+from unsloth import FastLanguageModel
+import torch
+
+max_seq_length = 2048  # Choose any! We auto support RoPE Scaling internally!
+dtype = (
+    None  # None for auto detection. Float16 for Tesla T4, V100, Bfloat16 for Ampere+
+)
+load_in_4bit = True  # Use 4bit quantization to reduce memory usage. Can be False.
+
+# 4bit pre quantized models we support for 4x faster downloading + no OOMs.
+fourbit_models = [
+    "unsloth/mistral-7b-v0.3-bnb-4bit",  # New Mistral v3 2x faster!
+    "unsloth/mistral-7b-instruct-v0.3-bnb-4bit",
+    "unsloth/llama-3-8b-bnb-4bit",  # Llama-3 15 trillion tokens model 2x faster!
+    "unsloth/llama-3-8b-Instruct-bnb-4bit",
+    "unsloth/llama-3-70b-bnb-4bit",
+    "unsloth/Phi-3-mini-4k-instruct",  # Phi-3 2x faster!
+    "unsloth/Phi-3-medium-4k-instruct",
+    "unsloth/mistral-7b-bnb-4bit",
+    "unsloth/gemma-7b-bnb-4bit",  # Gemma 2.2x faster!
+]  # More models at https://huggingface.co/unsloth
+
+model, tokenizer = FastLanguageModel.from_pretrained(
+    # model_name = "unsloth/llama-3-8b-bnb-4bit",
+    model_name="QuangHoDev/lora_model",
+    max_seq_length=max_seq_length,
+    dtype=dtype,
+    load_in_4bit=load_in_4bit,
+    # token = "hf_...", # use one if using gated models like meta-llama/Llama-2-7b-hf
+)
+
+model = FastLanguageModel.get_peft_model(
+    model,
+    r=16,  # Choose any number > 0 ! Suggested 8, 16, 32, 64, 128
+    target_modules=[
+        "q_proj",
+        "k_proj",
+        "v_proj",
+        "o_proj",
+        "gate_proj",
+        "up_proj",
+        "down_proj",
+    ],
+    lora_alpha=16,
+    lora_dropout=0,  # Supports any, but = 0 is optimized
+    bias="none",  # Supports any, but = "none" is optimized
+    # [NEW] "unsloth" uses 30% less VRAM, fits 2x larger batch sizes!
+    use_gradient_checkpointing="unsloth",  # True or "unsloth" for very long context
+    random_state=3407,
+    use_rslora=False,  # We support rank stabilized LoRA
+    loftq_config=None,  # And LoftQ
+)
+
+from datasets import load_dataset
+
+dataset = load_dataset("QuangHoDev/house-prices-info", split="train")
+
+from unsloth import to_sharegpt
+
+dataset = to_sharegpt(
+    dataset,
+    merged_prompt="Tiêu đề: {title}\nMô tả: {description}\nĐịa chỉ: {location}\nDiện tích: {area}",
+    output_column_name="price",
+    conversation_extension=3,  # Select more to handle longer conversations
+)
+
+from unsloth import standardize_sharegpt
+
+dataset = standardize_sharegpt(dataset)
+
+chat_template = """Dưới đây là thông tin về các bất động sản và giá của chúng.
+Hãy đoán giá bất động sản theo mô tả.
+
+>>> Mô tả bất động sản:
+{INPUT}
+
+>>> Giá là:
+{OUTPUT}"""
+
+from unsloth import apply_chat_template
+
+dataset = apply_chat_template(
+    dataset,
+    tokenizer=tokenizer,
+    chat_template=chat_template,
+    # default_system_message = "You are a helpful assistant", << [OPTIONAL]
+)
+
+with gr.Blocks() as demo:
+    description = gr.Textbox(label="Mô tả nhà", lines=4)
+    location = gr.Textbox(label="Địa chỉ", lines=2)
+    area = gr.Number(label="Diện tích (m²)")
+    output = gr.Textbox(label="Giá nhà")
+    greet_btn = gr.Button("Đoán giá", variant="primary")
+    greet_btn.click(
+        fn=predict_house_price,
+        inputs=[description, location, area],
+        outputs=output,
+        api_name="greet",
+    )
+
+    gr.Examples(
+        examples=[
+            [
+                "Nhanh tay sở hữu căn nhà riêng (căn góc) 3 tầng tại Phường 10, Gò Vấp với thiết kế 3 phòng ngủ + 4toilet/PT + bếp...",
+                "261/, Đường Quang Trung, Phường 10, Quận Gò Vấp",
+                40,
+            ],
+        ],
+        inputs=[description, location, area],
+        label="Ví dụ mẫu",
+    )
+
+    demo.launch()