Create Dockerfile

Dockerfile

@@ -0,0 +1,89 @@
+FROM python:3.11-slim
+
+WORKDIR /app
+
+# Install git
+RUN apt-get update && apt-get install -y git && rm -rf /var/lib/apt/lists/*
+
+# Clone YOUR repo
+RUN git clone https://github.com/unsloth/OpenMythos.git /app/repo
+WORKDIR /app/repo
+
+# Install repo dependencies (if you have requirements.txt or pyproject.toml)
+RUN pip install --no-cache-dir -e . || pip install --no-cache-dir -r requirements.txt || true
+
+# Install Gradio + common ML deps
+RUN pip install --no-cache-dir gradio torch transformers
+
+# Create the Gradio app inline
+RUN cat <<'EOF' > gradio_app.py
+import gradio as gr
+import torch
+from transformers import AutoTokenizer
+
+# Try importing from your repo (adjust if your module name differs)
+try:
+    from open_mythos.main import OpenMythos, MythosConfig
+except ImportError:
+    try:
+        from openmythos.main import OpenMythos, MythosConfig
+    except ImportError:
+        # Fallback: try finding any module with MythosConfig
+        import importlib, sys, os
+        for root, dirs, files in os.walk('/app/repo'):
+            for d in dirs:
+                if not d.startswith('.') and os.path.exists(os.path.join(root, d, '__init__.py')):
+                    sys.path.insert(0, root)
+            break
+        from open_mythos.main import OpenMythos, MythosConfig
+
+# Small CPU config
+CFG = MythosConfig(
+    vocab_size=50257,
+    dim=256,
+    n_heads=8,
+    max_seq_len=128,
+    max_loop_iters=4,
+    prelude_layers=1,
+    coda_layers=1,
+    n_experts=4,
+    n_shared_experts=1,
+    n_experts_per_tok=2,
+    expert_dim=64,
+    lora_rank=8,
+    attn_type="gqa",
+    n_kv_heads=2,
+)
+
+print("Loading model from your repo...")
+model = OpenMythos(CFG)
+model.eval()
+
+tokenizer = AutoTokenizer.from_pretrained("gpt2")
+tokenizer.pad_token = tokenizer.eos_token
+
+def generate(prompt: str, max_new_tokens: int = 20, n_loops: int = 4):
+    if not prompt.strip():
+        return "Please enter a prompt."
+    input_ids = tokenizer.encode(prompt, return_tensors="pt")
+    with torch.no_grad():
+        out = model.generate(input_ids, max_new_tokens=max_new_tokens, n_loops=n_loops)
+    return tokenizer.decode(out[0], skip_special_tokens=True)
+
+with gr.Blocks(title="OpenMythos") as demo:
+    gr.Markdown("# 🔄 OpenMythos — Your Repo")
+    with gr.Row():
+        with gr.Column():
+            prompt = gr.Textbox(label="Prompt", value="The future of AI is", lines=3)
+            max_new = gr.Slider(5, 100, value=20, step=1, label="Max New Tokens")
+            loops = gr.Slider(1, 8, value=4, step=1, label="Loop Iterations")
+            btn = gr.Button("Generate", variant="primary")
+        with gr.Column():
+            output = gr.Textbox(label="Output", lines=6)
+    btn.click(generate, inputs=[prompt, max_new, loops], outputs=output)
+
+demo.launch(server_name="0.0.0.0", server_port=7860)
+EOF
+
+EXPOSE 7860
+CMD ["python", "gradio_app.py"]