Deploy Lexi OBLITERATED ZeroGPU smoke test

README.md

@@ -1,14 +1,14 @@
 ---
-title: Test
-emoji: 🏃
-colorFrom: red
-colorTo: pink
+title: Lexi OBLITERATED Smoke Test
+emoji: 🧪
+colorFrom: gray
+colorTo: indigo
 sdk: gradio
-sdk_version: 6.26.0
-python_version: '3.12'
+python_version: "3.12.12"
 app_file: app.py
 pinned: false
-short_description: B
 ---
 
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# Lexi OBLITERATED Smoke Test
+
+Testing Llama-3.1-8B-Lexi-Uncensored-V2-OBLITERATED on Hugging Face ZeroGPU.

app.py

@@ -0,0 +1,91 @@
+import spaces
+import torch
+import gradio as gr
+from transformers import AutoTokenizer, AutoModelForCausalLM
+
+MODEL_ID = "OBLITERATUS/Llama-3.1-8B-Lexi-Uncensored-V2-OBLITERATED"
+
+SYSTEM_PROMPT = """You are Lexi. Respond directly and naturally to the user."""
+
+print("Loading tokenizer...")
+tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
+
+print("Loading model...")
+model = AutoModelForCausalLM.from_pretrained(
+    MODEL_ID,
+    torch_dtype=torch.bfloat16,
+    low_cpu_mem_usage=True,
+).to("cuda")
+
+model.eval()
+print("Model ready.")
+
+@spaces.GPU(duration=45)
+def chat(message, history):
+    messages = [
+        {
+            "role": "system",
+            "content": SYSTEM_PROMPT
+        }
+    ]
+
+    for item in history or []:
+        if isinstance(item, dict):
+            role = item.get("role")
+            content = item.get("content")
+
+            if role in ("user", "assistant") and content is not None:
+                messages.append(
+                    {
+                        "role": role,
+                        "content": str(content)
+                    }
+                )
+
+    messages.append(
+        {
+            "role": "user",
+            "content": message
+        }
+    )
+
+    input_ids = tokenizer.apply_chat_template(
+        messages,
+        tokenize=True,
+        add_generation_prompt=True,
+        return_tensors="pt",
+    ).to("cuda")
+
+    with torch.inference_mode():
+        output = model.generate(
+            input_ids,
+            max_new_tokens=192,
+            do_sample=True,
+            temperature=0.75,
+            top_p=0.90,
+            repetition_penalty=1.05,
+            use_cache=True,
+        )
+
+    new_tokens = output[0, input_ids.shape[-1]:]
+
+    response = tokenizer.decode(
+        new_tokens,
+        skip_special_tokens=True
+    )
+
+    return response.strip()
+
+demo = gr.ChatInterface(
+    fn=chat,
+    type="messages",
+    title="Lexi 8B — OBLITERATED",
+    description="Direct FP16/BF16 ZeroGPU smoke test",
+    examples=[
+        "Who are you?",
+        "Explain quantum entanglement in plain English.",
+        "Write a strange short story about a machine waking up."
+    ],
+)
+
+demo.launch()

requirements.txt

@@ -0,0 +1,3 @@
+transformers>=4.56.0,<5
+accelerate>=1.8.0
+safetensors>=0.4.5