Ropedia Academy · Models

One Space, many demos from Ropedia Academy — each tab is backed by a real model that fits its task:

• 💬 Language model — a real small instruct LLM (SmolLM2-360M-Instruct) answers your prompt
• 🎨 Diffusion (DDPM) — a real class-conditional diffusion model generates handwritten digits on demand
• 🔮 Action anticipation — the LLM predicts the next action in-context
• 🎮 CartPole policy — a REINFORCE / actor-critic agent solving Gymnasium CartPole-v1
• 🌍 World model — model-based planning with CEM
• 🛠️ Tool-use agent — a real ReAct loop: the LLM reasons, real tools execute (and fix its arithmetic)
• 🖼️ Gallery — every trained model's result figure + metrics

Hardware

Runs on free CPU (the LLM/DDPM are small; first LLM call loads the model and is slow). For snappy generation, set the Space hardware to ZeroGPU and add spaces to requirements.txt — the code already guards GPU sections with @spaces.GPU, so it just works.

Deploy

hf upload cy0307/ropedia-demos spaces/ropedia-demos . --repo-type=space