Sentence Transformers v6.2.0 adds mergekit model merging, faster loading, multi-GPU evals
tomaarsen · x · 2026-10-21
tomaarsen, maintainer of Sentence Transformers, released v6.2.0. Highlights:
- Model merging via arceeai's mergekit, with a new SentenceTransformer.merge() API supporting methods like linear with custom weights
- Faster model loading
- Evaluators automatically share work across multiple GPUs
- Lower-memory losses, multi-vector improvements, and plenty of bug fixes
More from Models
- GPT-4o and 4.1 die Jan 14: a team's task-by-task eval of OpenAI's cheaper Luna replacements — max_gladysh · 2026-10-21
- NVIDIA Nemotron Retriever models (Embed, Parse, Rerank) land on Baseten — baseten · 2026-10-21
- Two AI judges disagreed by 28 points grading the same 4 models — TibboPlayz · 2026-10-21
- 35B MoE at 20.5 tok/s on an iPhone: OpenBMB's Lookahead Expert Scheduling skips no experts — GasSmooth7439 · 2026-10-21
- ICA Lens in action: decomposing Qwen residuals reveals clean city/France features — furongh · 2026-10-21
- $200 subscription covers $655 of API usage — 3900% ROI on Claude Code/Codex — umeshonai · 2026-10-21