Update app.py

app.py

@@ -1,70 +1,241 @@
 import gradio as gr
-from huggingface_hub import InferenceClient
+import spaces
+from transformers import AutoModelForCausalLM, AutoTokenizer
+import torch
 
+model_name = "rubenroy/Zurich-14B-GCv2-5m"
+model = AutoModelForCausalLM.from_pretrained(
+    model_name,
+    torch_dtype=torch.bfloat16,
+    device_map="auto"
+)
+tokenizer = AutoTokenizer.from_pretrained(model_name)
 
-def respond(
-    message,
-    history: list[dict[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-    hf_token: gr.OAuthToken,
-):
-    """
-    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(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(
+@spaces.GPU
+def generate(message, chat_history, temperature=0.7, top_p=0.9, top_k=50, max_new_tokens=512, repetition_penalty=1.1):
+    messages = [
+        {"role": "system", "content": "You are a helpul assistant named Zurich, a 14 billion parameter Large Language model, you were fine-tuned and trained by Ruben Roy. You have been trained with the GammaCorpus v2 dataset, a dataset filled with structured and filtered multi-turn conversations, this was also made by Ruben Roy."}, # Attribution to Qwen is not included to prevent hallucinations.
+        {"role": "user", "content": message}
+    ]
+    text = tokenizer.apply_chat_template(
         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
+        tokenize=False,
+        add_generation_prompt=True
+    )
+    model_inputs = tokenizer([text], return_tensors="pt").to(model.device)
+    generated_ids = model.generate(
+        **model_inputs,
+        temperature=float(temperature),
+        top_p=float(top_p),
+        top_k=int(top_k),
+        max_new_tokens=int(max_new_tokens),
+        repetition_penalty=float(repetition_penalty),
+        do_sample=True if float(temperature) > 0 else False
+    )
+    generated_ids = [
+        output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
+    ]
+    response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
+    return response
 
-        response += token
-        yield response
+TITLE_HTML = """
+<link rel="stylesheet" href="https://cdnjs.cloudflare.com/ajax/libs/font-awesome/6.0.0/css/all.min.css">
+<style>
+    .model-btn {
+        background: linear-gradient(135deg, #2563eb 0%, #1d4ed8 100%);
+        color: white !important;
+        padding: 0.75rem 1rem;
+        border-radius: 0.5rem;
+        text-decoration: none !important;
+        font-weight: 500;
+        transition: all 0.2s ease;
+        font-size: 0.9rem;
+        display: flex;
+        align-items: center;
+        justify-content: center;
+        box-shadow: 0 2px 4px rgba(0,0,0,0.1);
+    }
+    .model-btn:hover {
+        background: linear-gradient(135deg, #1d4ed8 0%, #1e40af 100%);
+        box-shadow: 0 4px 6px rgba(0,0,0,0.2);
+    }
+    .model-section {
+        flex: 1;
+        max-width: 450px;
+        background: rgba(255, 255, 255, 0.05);
+        padding: 1.5rem;
+        border-radius: 1rem;
+        border: 1px solid rgba(255, 255, 255, 0.1);
+        backdrop-filter: blur(10px);
+        transition: all 0.3s ease;
+    }
+    .info-link {
+        color: #60a5fa;
+        text-decoration: none;
+        transition: color 0.2s ease;
+    }
+    .info-link:hover {
+        color: #93c5fd;
+        text-decoration: underline;
+    }
+    .info-section {
+        margin-top: 0.5rem;
+        font-size: 0.9rem;
+        color: #94a3b8;
+    }
+    .settings-section {
+        background: rgba(255, 255, 255, 0.05);
+        padding: 1.5rem;
+        border-radius: 1rem;
+        margin: 1.5rem auto;
+        border: 1px solid rgba(255, 255, 255, 0.1);
+        max-width: 800px;
+    }
+    .settings-title {
+        color: #e2e8f0;
+        font-size: 1.25rem;
+        font-weight: 600;
+        margin-bottom: 1rem;
+        display: flex;
+        align-items: center;
+        gap: 0.7rem;
+    }
+    .parameter-info {
+        color: #94a3b8;
+        font-size: 0.8rem;
+        margin-top: 0.25rem;
+    }
+</style>
 
+<div style="background: linear-gradient(135deg, #1e293b 0%, #0f172a 100%); padding: 1.5rem; border-radius: 1.5rem; text-align: center; margin: 1rem auto; max-width: 1200px; box-shadow: 0 4px 6px -1px rgba(0, 0, 0, 0.1);">
+    <div style="margin-bottom: 1.5rem;">
+        <div style="display: flex; align-items: center; justify-content: center; gap: 1rem;">
+            <h1 style="font-size: 2.5rem; font-weight: 800; margin: 0; background: linear-gradient(135deg, #60a5fa 0%, #93c5fd 100%); -webkit-background-clip: text; -webkit-text-fill-color: transparent;">Zurich</h1>
+            <div style="width: 2px; height: 2.5rem; background: linear-gradient(180deg, #3b82f6 0%, #60a5fa 100%);"></div>
+            <p style="font-size: 1.25rem; color: #94a3b8; margin: 0;">GammaCorpus v2-5m</p>
+        </div>
+        <div class="info-section">
+            <span>Fine-tuned from <a href="https://huggingface.co/Qwen/Qwen2.5-14B-Instruct" class="info-link">Qwen 2.5 14B Instruct</a> | Model: <a href="https://huggingface.co/rubenroy/Zurich-14B-GCv2-5m" class="info-link">Zurich-14B-GCv2-5m</a> | Training Dataset: <a href="https://huggingface.co/datasets/rubenroy/GammaCorpus-v2-5m" class="info-link">GammaCorpus v2 5m</a></span>
+        </div>
+    </div>
 
+    <div style="display: flex; gap: 1.5rem; justify-content: center; flex-wrap: wrap;">
+        <div class="model-section">
+            <h2 style="font-size: 1.25rem; color: #e2e8f0; margin-bottom: 1.4rem; margin-top: 1px; font-weight: 600; display: flex; align-items: center; justify-content: center; gap: 0.7rem;">
+                <i class="fas fa-microchip"></i>
+                1.5B Models
+            </h2>
+            <div style="display: grid; grid-template-columns: repeat(2, 1fr); gap: 0.75rem;">
+                <a href="https://huggingface.co/rubenroy/Zurich-1.5B-GCv2-5m" class="model-btn">Zurich 1.5B GCv2 5m</a>
+                <a href="https://huggingface.co/rubenroy/Zurich-1.5B-GCv2-1m" class="model-btn">Zurich 1.5B GCv2 1m</a>
+                <a href="https://huggingface.co/rubenroy/Zurich-1.5B-GCv2-500k" class="model-btn">Zurich 1.5B GCv2 500k</a>
+                <a href="https://huggingface.co/rubenroy/Zurich-1.5B-GCv2-100k" class="model-btn">Zurich 1.5B GCv2 100k</a>
+                <a href="https://huggingface.co/rubenroy/Zurich-1.5B-GCv2-50k" class="model-btn">Zurich 1.5B GCv2 50k</a>
+                <a href="https://huggingface.co/rubenroy/Zurich-1.5B-GCv2-10k" class="model-btn">Zurich 1.5B GCv2 10k</a>
+            </div>
+        </div>
+        <div class="model-section">
+            <h2 style="font-size: 1.25rem; color: #e2e8f0; margin-bottom: 1.4rem; margin-top: 1px; font-weight: 600; display: flex; align-items: center; justify-content: center; gap: 0.7rem;">
+                <i class="fas fa-brain"></i>
+                7B Models
+            </h2>
+            <div style="display: grid; grid-template-columns: repeat(2, 1fr); gap: 0.75rem;">
+                <a href="https://huggingface.co/rubenroy/Zurich-7B-GCv2-5m" class="model-btn">Zurich 7B GCv2 5m</a>
+                <a href="https://huggingface.co/rubenroy/Zurich-7B-GCv2-1m" class="model-btn">Zurich 7B GCv2 1m</a>
+                <a href="https://huggingface.co/rubenroy/Zurich-7B-GCv2-500k" class="model-btn">Zurich 7B GCv2 500k</a>
+                <a href="https://huggingface.co/rubenroy/Zurich-7B-GCv2-100k" class="model-btn">Zurich 7B GCv2 100k</a>
+                <a href="https://huggingface.co/rubenroy/Zurich-7B-GCv2-50k" class="model-btn">Zurich 7B GCv2 50k</a>
+                <a href="https://huggingface.co/rubenroy/Zurich-7B-GCv2-10k" class="model-btn">Zurich 7B GCv2 10k</a>
+            </div>
+        </div>
+        <div class="model-section">
+            <h2 style="font-size: 1.25rem; color: #e2e8f0; margin-bottom: 1.4rem; margin-top: 1px; font-weight: 600; display: flex; align-items: center; justify-content: center; gap: 0.7rem;">
+                <i class="fas fa-rocket"></i>
+                14B Models
+            </h2>
+            <div style="display: grid; grid-template-columns: repeat(2, 1fr); gap: 0.75rem;">
+                <a href="https://huggingface.co/rubenroy/Zurich-14B-GCv2-5m" class="model-btn">Zurich 14B GCv2 5m</a>
+                <a href="https://huggingface.co/rubenroy/Zurich-14B-GCv2-1m" class="model-btn">Zurich 14B GCv2 1m</a>
+                <a href="https://huggingface.co/rubenroy/Zurich-14B-GCv2-500k" class="model-btn">Zurich 14B GCv2 500k</a>
+                <a href="https://huggingface.co/rubenroy/Zurich-14B-GCv2-100k" class="model-btn">Zurich 14B GCv2 100k</a>
+                <a href="https://huggingface.co/rubenroy/Zurich-14B-GCv2-50k" class="model-btn">Zurich 14B GCv2 50k</a>
+                <a href="https://huggingface.co/rubenroy/Zurich-14B-GCv2-10k" class="model-btn">Zurich 14B GCv2 10k</a>
+            </div>
+        </div>
+    </div>
+</div>
 """
-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)",
-        ),
-    ],
-)
 
-with gr.Blocks() as demo:
-    with gr.Sidebar():
-        gr.LoginButton()
-    chatbot.render()
+examples = [
+    ["Explain quantum computing in simple terms"],
+    ["Write a short story about a time traveler"],
+    ["Explain the process of photosynthesis"],
+    ["Tell me an interesting fact about Palm trees"]
+]
 
+with gr.Blocks() as demo:
+    gr.HTML(TITLE_HTML)
+    
+    with gr.Accordion("Generation Settings", open=False):
+        with gr.Row():
+            with gr.Column():
+                temperature = gr.Slider(
+                    minimum=0.0,
+                    maximum=2.0,
+                    value=0.7,
+                    step=0.1,
+                    label="Temperature",
+                    info="Higher values make the output more random, lower values make it more deterministic",
+                    interactive=True
+                )
+                top_p = gr.Slider(
+                    minimum=0.0,
+                    maximum=1.0,
+                    value=0.9,
+                    step=0.05,
+                    label="Top P",
+                    info="Controls the cumulative probability threshold for nucleus sampling",
+                    interactive=True
+                )
+                top_k = gr.Slider(
+                    minimum=1,
+                    maximum=100,
+                    value=50,
+                    step=1,
+                    label="Top K",
+                    info="Limits the number of tokens to consider for each generation step",
+                    interactive=True
+                )
+            with gr.Column():
+                max_new_tokens = gr.Slider(
+                    minimum=1,
+                    maximum=2048,
+                    value=512,
+                    step=1,
+                    label="Max New Tokens",
+                    info="Maximum number of tokens to generate in the response",
+                    interactive=True
+                )
+                repetition_penalty = gr.Slider(
+                    minimum=1.0,
+                    maximum=2.0,
+                    value=1.1,
+                    step=0.1,
+                    label="Repetition Penalty",
+                    info="Higher values stop the model from repeating the same info",
+                    interactive=True
+                )
+    
+    chatbot = gr.ChatInterface(
+        fn=generate,
+        additional_inputs=[
+            temperature,
+            top_p,
+            top_k,
+            max_new_tokens,
+            repetition_penalty
+        ],
+        examples=examples
+    )
 
-if __name__ == "__main__":
-    demo.launch()
+demo.launch(share=True)