Skip to content
OBLAIDISH NEWS
signal_tag · 3_broadcasts

#large-language-models

// 3 transmissions tagged with #large-language-models

VibeThinker 3B model beats Opus 4.5 on reasoning benchmarks
TX_216091· AI

VibeThinker 3B model beats Opus 4.5 on reasoning benchmarks

The VibeThinker paper on arXiv introduces a 3‑billion‑parameter model that outperforms Opus 4.5 on reasoning tasks using a new SFT+GRPO fine‑tuning pipeline. The result shows smaller models can rival larger ones when trained with the right technique.

Mistral AI Now Summit showcases 50% response-time boost with open-weight models
TX_156897· AI

Mistral AI Now Summit showcases 50% response-time boost with open-weight models

Mistral AI Now Summit highlighted open-weight models as a path for startups to compete, with a demo startup reporting a 50% cut in customer-service response time using a fine-tuned LLM [DevTo].

Δ-Mem cuts memory use in large language models without performance loss
TX_947343· AI

Δ-Mem cuts memory use in large language models without performance loss

Δ-Mem, a new memory optimization technique, reduces memory consumption in LLMs by compressing key-value states and reusing memory slots, maintaining full model performance [arXiv].