Upload README.md with huggingface_hub

README.md

@@ -1,27 +1,30 @@
 ---
-title: Olmo 3 7B RL-Zero-Math
-emoji: 🧮
+title: Olmo 3 7B Think
+emoji: 🧠
 colorFrom: blue
 colorTo: indigo
 sdk: gradio
 sdk_version: 5.49.1
 app_file: app.py
-short_description: Chat demo for allenai/Olmo-3-7B-RL-Zero-Math
+short_description: Chat demo for allenai/Olmo-3-7B-Think
 python_version: "3.12"
 startup_duration_timeout: 30m
 pinned: false
 license: apache-2.0
 models:
-  - allenai/Olmo-3-7B-RL-Zero-Math
+  - allenai/Olmo-3-7B-Think
 ---
 
-# 🧮 Olmo 3 · 7B · RL-Zero-Math
+# 🧠 Olmo 3 · 7B · Think
 
-A streaming chat demo for [`allenai/Olmo-3-7B-RL-Zero-Math`](https://huggingface.co/allenai/Olmo-3-7B-RL-Zero-Math),
-a 7B model from Ai2's Olmo 3 family trained with reinforcement learning from verifiable
-rewards on math queries.
+A streaming chat demo for [`allenai/Olmo-3-7B-Think`](https://huggingface.co/allenai/Olmo-3-7B-Think),
+a 7B reasoning model from Ai2's Olmo 3 family. It produces a step-by-step reasoning
+trace inside `<think>...</think>`, shown here as a collapsible **🧠 Thinking** bubble,
+followed by a final answer.
 
-Runs on **ZeroGPU** — the model is loaded in bf16 and generation is streamed token by token.
-Adjust max new tokens, temperature, and top-p in the generation settings accordion.
+Runs on **ZeroGPU** — the model is loaded in bf16 and generation is streamed token by
+token. Supports up to **32,768** new tokens (default 8,192; very large budgets can be
+slow on ZeroGPU). Recommended sampling settings (temperature 0.6, top-p 0.95) match the
+model's own `generation_config`.
 
 Built with [Gradio](https://gradio.app).