Dev: OAI Dot/Astra ran an entire ML research project end-to-end, leaving him only scope decisions
andreisavu · x · 2026-10-30
Developer andreisavu says the AI capability overhang is "very real": his LittlePly open chess research notebook — a small 85.94M-parameter move-prediction model (CM-5) live on Lichess — was built and deployed end-to-end by OAI Dot/Astra while he just threw ideas at it.
- Against a fixed Sunfish benchmark, checkpoints improved from CM-3 (13W/2D/85L) to CM-5 (27W/9D/63L); CM-6 finished a training sanity run at 16.5% score over 100 games and remains research-only.
- His own role has shrunk to scope management and sequencing decisions.
More from coding & agent
- Anthropic launches Claude Duet, a proactive agent powered by Opus 5.5 — joellewenstein · 2026-10-30
- The Future of MCP: 6 Shifts That Will Shape Production AI Agents — MaryamMiradi · 2026-10-30
- Spring AI 2.1 adds VectorStore.upsert for apps that bring their own embeddings — therealdanvega · 2026-10-30
- Auditing an LLM-built knowledge graph: 0 of 186 facts matched verbatim, 1/3 of edges hit wrong entities — JeremyCMorgan · 2026-10-30
- Same agent harness, opposite rankings: TB4 benchmark drama exposes eval chaos — lunwang1996 · 2026-10-30
- The eternal eval debugging meme: ask the vendor if it's data or harness, it's data — dejavucoder · 2026-10-30