Open Source Community Still Debating Base Weights
kalomaze · x · 2026-07-16
Responding to claims that modern open source projects are increasingly reluctant to release true base weights, the poster argues: - In some cases, it's simply a **lack of interest or demand**; for example, Qwen seems inconsistent and not particularly strategic about this. - However, some organizations continue to release **base weights**, such as **Arcee** and **InclusionAI**. Another perspective suggests that some teams no longer strictly separate "pre-training" and "post-training" into distinct phases, treating them instead as parts of a single, unified training pipeline.
More from Models
- Grok Build CLI v0.2.108 adds cross-machine session resume and editor handoff — Scobleizer · 2026-07-21
- Gemini 3.6 Flash reportedly appears in Antigravity under `gemini-3.6-flash-tiered` — gaganghotra_ · 2026-07-21
- Grok 4.5 tops a long-horizon terminal benchmark, Elon Musk says — elonmusk · 2026-07-21
- K3 seems faster on the biggest coding plan than through OpenRouter, user says — xeophon · 2026-07-21
- Andrew Ng-style distillation joke turns model reuse into a theft punchline — pmddomingos · 2026-07-21
- Grok’s Twitter/X search quality appears to have regressed — ivan_bezdomny · 2026-07-21