huggingface/daily-papers ai SLAI T-Rex: Full-Parameter Post-training of the DeepSeek-V4 Family on Ascend SuperPOD Dongfang Li, Xiaodong Luo, Ruoyu Sun, Xuhui Chen, Linyuan Qiu 2026-07-21 UTC 1 min read Related Articles github-trending/python OpenDCAI /DataFlow reddit/r/machinelearningnews Open source AI systems for fintech: code, architecture, docs medium Why Kubernetes Is Slowing Down Your GPUs (And How Topology-Aware Scheduling Fixes It) reddit/r/LocalLLM I got tired of re-explaining my setup to every new model, so I built a shared KB over MCP hackernews Running a 28.9M parameter LLM on an $8 microcontroller ← Back to homepage