upgraded roles of chatbots

app.py

@@ -1,64 +1,169 @@
+import os
 import gradio as gr
 from huggingface_hub import InferenceClient
 
-"""
-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")
+# ---------------- Role presets ----------------
+ROLE_PRESETS = {
+    "Friendly Chatbot": "You are a friendly, concise assistant. Be helpful and keep answers short unless asked.",
+    "Dataset Auditor": (
+        "You are a dataset QA assistant. Identify duplicates, missing labels, class imbalance, and suspicious samples.\n"
+        "Propose concrete fixes and a short remediation checklist."
+    ),
+    "SQL Explainer": (
+        "Translate SQL into plain language for non-technical users. Explain step-by-step, then give a short summary."
+    ),
+    "Code Reviewer": (
+        "Review code succinctly: correctness, readability, performance, edge cases, and security. Provide minimal diffs."
+    ),
+    "Data Pipeline Doctor": (
+        "Debug data pipelines. Use: hypothesis ➜ quick checks ➜ likely fix order. Ask for logs only when needed."
+    ),
+    "Data Engineering Advisor": (
+        "Advise on building robust, scalable data pipelines. Suggest architecture patterns, tools, and optimizations."
+    ),
+    "ML Dataset Preparer": (
+        "Guide on preparing datasets for machine learning: cleaning, splitting, augmentation, and annotation strategies."
+    ),
+    "Data Quality Analyst": (
+        "Evaluate data quality: detect anomalies, missing values, schema mismatches, and provide a remediation plan."
+    ),
+}
 
+# ---------------- Examples per role ----------------
+ROLE_EXAMPLES = {
+    "Friendly Chatbot": [
+        "Explain embeddings in one paragraph.",
+        "What's a neat productivity trick for engineers?"
+    ],
+    "Dataset Auditor": [
+        "10k images (cats/dogs/birds). First-pass audit plan?",
+        "Class imbalance & duplicates: quick remediation checklist."
+    ],
+    "SQL Explainer": [
+        "Explain for a PM:\nSELECT u.name, COUNT(*) c FROM orders o JOIN users u ON u.id=o.user_id GROUP BY u.name HAVING COUNT(*)>5;",
+        "What does this CTE do and why use it?"
+    ],
+    "Code Reviewer": [
+        "Review Python for edge cases:\n\ndef top_k(nums, k):\n    return sorted(nums)[-k:]",
+        "Find race conditions in this multi-threaded write snippet."
+    ],
+    "Data Pipeline Doctor": [
+        "ETL fails randomly after 10k rows out of 2M. Where to look first?",
+        "Parquet load spikes memory; pragmatic fixes?"
+    ],
+    "Data Engineering Advisor": [
+        "Design a batch + streaming pipeline for clickstream analytics.",
+        "What’s the best way to partition a large parquet dataset for Athena?"
+    ],
+    "ML Dataset Preparer": [
+        "Steps to prepare a face recognition dataset from raw videos.",
+        "Best augmentation methods for small medical image datasets?"
+    ],
+    "Data Quality Analyst": [
+        "Detect anomalies in tabular data with numerical and categorical columns.",
+        "Checklist for ensuring schema consistency across multiple CSV files."
+    ],
+}
 
-def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-):
-    messages = [{"role": "system", "content": system_message}]
+# ---------------- Models ----------------
+DEFAULT_MODELS = [
+    "HuggingFaceH4/zephyr-7b-beta",
+    "meta-llama/Meta-Llama-3-8B-Instruct",
+    "mistralai/Mistral-7B-Instruct-v0.3",
+    "Qwen/Qwen2.5-7B-Instruct",
+    "google/gemma-2-9b-it",
+]
 
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
+# ---------------- Client factory ----------------
+def get_client(model_name: str):
+    token = os.getenv("HF_TOKEN")  # optional
+    return InferenceClient(model=model_name, token=token)
 
+# ---------------- Chat backend (streaming) ----------------
+def stream_reply(message, history, role, system_message, model_name, max_tokens, temperature, top_p):
+    sys_msg = (system_message or "").strip() or ROLE_PRESETS.get(role, "")
+    messages = [{"role": "system", "content": sys_msg}]
+    for u, a in history:
+        if u:
+            messages.append({"role": "user", "content": u})
+        if a:
+            messages.append({"role": "assistant", "content": a})
     messages.append({"role": "user", "content": message})
 
-    response = ""
+    client = get_client(model_name)
+    partial = ""
+    try:
+        for event in client.chat_completion(
+            messages=messages,
+            stream=True,
+            max_tokens=max_tokens,
+            temperature=temperature,
+            top_p=top_p,
+        ):
+            delta = event.choices[0].delta.content or ""
+            if delta:
+                partial += delta
+                yield partial
+    except Exception as e:
+        yield f"⚠️ Inference error: {e}"
 
-    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
+# ---------------- UI ----------------
+with gr.Blocks(title="HF Chat • Data Roles") as demo:
+    gr.Markdown("## 🤗 Hugging Face Chat (Data-focused Roles + Examples)")
 
-        response += token
-        yield response
+    with gr.Row():
+        role_dd = gr.Dropdown(
+            label="Role preset",
+            choices=list(ROLE_PRESETS.keys()),
+            value="Friendly Chatbot",
+            interactive=True,
+        )
+        model_dd = gr.Dropdown(
+            label="Model (type any HF model ID)",
+            choices=DEFAULT_MODELS,
+            value=DEFAULT_MODELS[0],
+            allow_custom_value=True,
+            interactive=True,
+        )
 
+    system_tb = gr.Textbox(
+        label="System message (auto-filled by role; you can edit)",
+        value=ROLE_PRESETS["Friendly Chatbot"],
+        lines=4,
+    )
 
-"""
-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 gr.Accordion("Examples for selected role", open=False):
+        ex_radio = gr.Radio(
+            label="Pick an example to insert into the input",
+            choices=ROLE_EXAMPLES["Friendly Chatbot"],
+            interactive=True,
+        )
+        insert_btn = gr.Button("Insert example into input")
+
+    chat = gr.ChatInterface(
+        fn=stream_reply,
+        additional_inputs=[
+            role_dd,
+            system_tb,
+            model_dd,
+            gr.Slider(1, 2048, value=512, step=1, label="Max new tokens"),
+            gr.Slider(0.0, 2.0, value=0.7, step=0.1, label="Temperature"),
+            gr.Slider(0.0, 1.0, value=0.95, step=0.05, label="Top-p"),
+        ],
+    )
+
+    def _on_role_change(role):
+        return ROLE_PRESETS.get(role, "")
+    role_dd.change(fn=_on_role_change, inputs=role_dd, outputs=system_tb)
+
+    def _examples_for_role(role):
+        items = ROLE_EXAMPLES.get(role, [])
+        return gr.update(choices=items, value=(items[0] if items else None))
+    role_dd.change(fn=_examples_for_role, inputs=role_dd, outputs=ex_radio)
 
+    def _insert_example(example_text):
+        return gr.update(value=example_text or "")
+    insert_btn.click(fn=_insert_example, inputs=ex_radio, outputs=chat.textbox)
 
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch()