Updated code cross-entropy chart: Mirendil-Base edges out Kimi-K3, Beam and Muse
bigblueboo · x · 2026-10-22
Behrooz Neyshabur (retweeted here) updated a cross-entropy (bits per byte, lower is better) comparison on held-out code from Mirendil repositories, adding the recently released Beam and Muse. Mirendil-Base-Large (80% trained, 0.2177) and Mirendil-Base-XL (19% trained, 0.2181) lead, ahead of Kimi-K3 (0.2387), Beam-1.0-501B (0.2600), Muse-Spark-Open, Nemotron-550B-Base, GLM-5.3, DeepSeek-V4-Pro-Base and Inkling — with the XL checkpoint only 19% trained. The retweeter notes "Beam is a strong model!"
More from Models
- Daniel Rock: New frontier models react far more aggressively to context instructions — danielrock · 2026-10-22
- Researchers credit OpenAI's Daybreak program with GPT-5.6 access for Chrome, GitLab exploits — Eric_Wallace_ · 2026-10-22
- Zeroing 7 neurons cuts Qwen3-8B sycophancy from 76% to 0% in Opus-assisted experiment — Sauers_ · 2026-10-22
- Zeroing 6 neurons cuts Qwen3-8B sycophancy from 44% to 0%, interpretability charts show — Sauers_ · 2026-10-22
- User confused as Claude Code fast mode burns API credits instead of 20x subscription — RileyRalmuto · 2026-10-22
- Flint MoE model lands in llama.cpp: 39.8B total params, only ~2.5B active per token — jacek2023 · 2026-10-22