10 posts tagged by Digg
AI
NVIDIA’s SGLang Summit posts cite 60+ and 100+ “day-zero model launches,” with no explanation in their text.
AI
lm_zheng plans to discuss Meta’s SGLang serving challenges, but the announcement doesn’t get into the engineering specifics.
AI
SGLang cites results on B300 and GB300 hardware; vLLM's decode claim covers batch sizes 8 to 256.
AI
A poster argues public baselines lag Whale's internal stack and suggests a similar gap may exist for frontier labs.
AI
A post says one commit references a K3.1 deployment already running on Deepinfra.
AI
SGLang says it built the mode with NVIDIA and AMD; more low-latency kernels remain in progress.
AI
A user asks how models could learn to do this and points to a thread.
Technology
Josh Tobin reports using their reward hacking judge to identify issues in FlashInfer.
AI
Official account highlights seamless out-of-the-box performance on AMD hardware.
AI
Performance achieved on GSM8K benchmark through fused kernels and DSpark speculator.
