Update app.py

app.py

@@ -1,70 +1,127 @@
+# app.py inside your Hugging Face Space
+
+import os
+import re
 import gradio as gr
-from huggingface_hub import InferenceClient
-
-
-def respond(
-    message,
-    history: list[dict[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-    hf_token: gr.OAuthToken,
-):
+import torch
+from transformers import AutoTokenizer, AutoModelForCausalLM
+
+# 1. SeaLLM model focused on SEA languages (includes Lao)
+MODEL_NAME = "SeaLLMs/SeaLLMs-v3-1.5B-Chat"  # 1.5B chat model:contentReference[oaicite:6]{index=6}
+
+tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
+model = AutoModelForCausalLM.from_pretrained(
+    MODEL_NAME,
+    torch_dtype=torch.float32,   # CPU on free tier → float32
+)
+
+# 2. Load Laos history knowledge
+DATA_PATH = "data/laos_history.txt"
+if os.path.exists(DATA_PATH):
+    with open(DATA_PATH, "r", encoding="utf-8") as f:
+        RAW_KNOWLEDGE = f.read()
+else:
+    RAW_KNOWLEDGE = "ຍັງບໍ່ມີຂໍ້ມູນປະຫວັດສາດຖືກໂຫຼດ."
+
+# Split into paragraphs for simple retrieval
+PARAGRAPHS = [p.strip() for p in RAW_KNOWLEDGE.split("\n\n") if p.strip()]
+
+
+def retrieve_context(question: str, max_paragraphs: int = 5) -> str:
     """
-    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
+    VERY simple keyword-based retrieval over PARAGRAPHS.
+    Good enough for first prototype; later you can replace with embeddings.
     """
-    client = InferenceClient(token=hf_token.token, model="openai/gpt-oss-20b")
-
-    messages = [{"role": "system", "content": system_message}]
-
-    messages.extend(history)
-
-    messages.append({"role": "user", "content": message})
-
-    response = ""
-
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
-    ):
-        choices = message.choices
-        token = ""
-        if len(choices) and choices[0].delta.content:
-            token = choices[0].delta.content
-
-        response += token
-        yield response
-
-
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
-chatbot = gr.ChatInterface(
-    respond,
-    type="messages",
-    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)",
-        ),
-    ],
+    if not PARAGRAPHS:
+        return RAW_KNOWLEDGE
+
+    # tokenize question into simple lowercased words
+    terms = [w for w in re.split(r"\s+", question.lower()) if len(w) > 2]
+    if not terms:
+        return "\n\n".join(PARAGRAPHS[:max_paragraphs])
+
+    scored = []
+    for p in PARAGRAPHS:
+        p_lower = p.lower()
+        score = sum(p_lower.count(t) for t in terms)
+        if score > 0:
+            scored.append((score, p))
+
+    scored.sort(key=lambda x: x[0], reverse=True)
+    top = [p for _, p in scored[:max_paragraphs]]
+    if not top:
+        top = PARAGRAPHS[:max_paragraphs]
+
+    return "\n\n".join(top)
+
+
+SYSTEM_PROMPT = (
+    "ທ່ານແມ່ນຜູ້ຊ່ວຍເຫຼືອດ້ານປະຫວັດສາດຂອງປະເທດລາວ. "
+    "ຕອບແຕ່ພາສາລາວ, ອະທິບາຍໃຫ້ເຂົ້າໃຈງ່າຍ ແລະສັ້ນກະທັດຮັດ. "
+    "ໃຫ້ອີງຈາກຂໍ້ມູນຂ້າງລຸ່ມນີ້ເທົ່ານັ້ນ. "
+    "ຖ້າຂໍ້ມູນບໍ່ພຽງພໍ ຫຼືບໍ່ຊັດເຈນ ໃຫ້ບອກວ່າບໍ່ແນ່ໃຈ."
 )
 
-with gr.Blocks() as demo:
-    with gr.Sidebar():
-        gr.LoginButton()
-    chatbot.render()
 
+def build_prompt(question: str) -> str:
+    context = retrieve_context(question)
+    prompt = f"""{SYSTEM_PROMPT}
+
+ຂໍ້ມູນອ້າງອີງ:
+{context}
+
+ຄຳຖາມ: {question}
+
+ຄຳຕອບດ້ວຍພາສາລາວ:"""
+    return prompt
+
+
+def generate_answer(question: str) -> str:
+    prompt = build_prompt(question)
+    inputs = tokenizer(prompt, return_tensors="pt")
+    with torch.no_grad():
+        outputs = model.generate(
+            **inputs,
+            max_new_tokens=256,
+            do_sample=True,
+            temperature=0.7,
+            top_p=0.9,
+        )
+
+    # slice off the prompt part
+    generated_ids = outputs[0][inputs["input_ids"].shape[1]:]
+    answer = tokenizer.decode(generated_ids, skip_special_tokens=True)
+    # clean up a bit
+    return answer.strip()
+
+
+# 3. Gradio chat function
+def laos_history_bot(message: str, history: list):
+    """
+    Gradio ChatInterface expects (message, history) and returns a string.
+    We ignore history for now (you can later use it in the prompt).
+    """
+    if not message.strip():
+        return "ກະລຸນາພິມຄຳຖາມກ່ອນ."
+
+    try:
+        answer = generate_answer(message)
+    except Exception as e:
+        # simple fallback if something goes wrong
+        return f"ລະບົບມີບັນຫາ: {e}"
+
+    return answer
+
+
+demo = gr.ChatInterface(
+    fn=laos_history_bot,
+    title="Laos History Chatbot (Lao language)",
+    description="ຖາມຂໍ້ມູນກ່ຽວກັບປະຫວັດສາດຂອງປະເທດລາວ",
+    examples=[
+        "ອານາຈັກລ້ານຊ້າງເກີດຂຶ້ນໃນປີໃດ?",
+        "ເມືອງຫວຽງຈັນເຄີຍເປັນນະຄອນຫຼວງຂອງອານາຈັກໃດ?",
+    ],
+)
 
 if __name__ == "__main__":
     demo.launch()