Running Qwen2.5-Coder-3B on Android: Jupyter + Pydantic Evals on-device eval pipeline
alex_b3b · reddit · 2026-11-11
The author's PythonHere project lets you run Python on Android directly from Jupyter, now with LiteRT-LM support for exploring small on-device models.
For benchmarking, they combined Pydantic Evals with plain Python functions to run inference on the phone and write custom evaluators in Jupyter — a setup they find very convenient for on-device experiments.
Testing Qwen2.5-Coder-3B-Instruct on Python tasks, the 3B model followed instructions better than expected and sometimes produced useful code. Video and notebooks with code and results are linked.
More from Infra
- Baidu's ERNIE Evolve tops public bests on NanoGPT Speedrun, brings RSI to real inference optimization — xiaohu · 2026-11-11
- Baidu says RSI auto-optimized ERNIE inference, 6.24x decode speedup on Kunxin P800 — xiaohu · 2026-11-11
- Nvidia's supply commitments jump from $119B to $279B in one quarter, locking up 2027-28 memory — tengyanAI · 2026-11-11
- Train 27B models on a 24GB GPU: ternary optimizer cuts memory to 1 byte/weight — oguzeldereli · 2026-11-11
- NVIDIA expands Meta Muse partnership, invests $3B in Meta stock — armand_ruiz · 2026-11-11
- Did Opus 5.5 really cost $10B? Analysts argue final training run is under 10% of that — teortaxesTex · 2026-11-11