TinyDecide
An ultra-lightweight decision model with 10 million parameters and approximately 6 MB in size, designed for structured decision tasks such as classification, scoring, and routing. It supports deployment across multiple platforms including browsers, Node.js, Python, Rust, and ESP32.
Timeline (last 90 days)
TinyDecide is a decision model with only 10 million parameters and a size of approximately 6 MB.
TinyDecide is open-sourced and hosted on Hugging Face under the repository TheREZOR/TinyDecide.
The author claims that TinyDecide is the smallest model on the Decision Index leaderboard.
TinyDecide can run across multiple environments including browsers, Node.js, Python, Rust, and ESP32.
Running TinyDecide on an ESP32 means that its quantized inference weights and runtime memory footprint must be kept at the hundreds-of-KB level, typically relying on INT8 or even INT4 integer quantization and deeply trimmed inference engines.
TinyDecide's performance claims come from the author's own statements; the specific rankings and comparison baselines on Decision Index need to be verified against empirical data on the Hugging Face page.
TinyDecide's leaderboard ranking cannot be directly compared to the capabilities of GPT-4o or Claude, because the two target non-overlapping task domains.
The developer shared the TinyDecide model on Reddit.
All Facts (8)
TinyDecide is a decision model with only 10 million parameters and a size of approximately 6 MB.
50%UnverifiedTinyDecide's performance claims come from the author's own statements; the specific rankings and comparison baselines on Decision Index need to be verified against empirical data on the Hugging Face page.
50%UnverifiedTinyDecide's leaderboard ranking cannot be directly compared to the capabilities of GPT-4o or Claude, because the two target non-overlapping task domains.
50%UnverifiedThe author claims that TinyDecide is the smallest model on the Decision Index leaderboard.
50%UnverifiedTinyDecide can run across multiple environments including browsers, Node.js, Python, Rust, and ESP32.
50%UnverifiedRunning TinyDecide on an ESP32 means that its quantized inference weights and runtime memory footprint must be kept at the hundreds-of-KB level, typically relying on INT8 or even INT4 integer quantization and deeply trimmed inference engines.
50%UnverifiedTinyDecide is open-sourced and hosted on Hugging Face under the repository TheREZOR/TinyDecide.
50%UnverifiedThe developer shared the TinyDecide model on Reddit.
50%