Wednesday, August 5, 2026

1. TACT: Taxonomy-Aligned Post-Training for Pedagogically Adaptive English Tutoring
   T2arXiv AI/CL/LG（论文）

   TACT presents a human-grounded framework for post-training and evaluating pedagogically adaptive ESL tutors. It introduces two complementary taxonomies: a Tutor-Strategy Taxonomy with 13 response strategies and a Student-Move Taxonomy characterizing learner behavior. The TACTCorpus enriches 260 authentic teacher-student conversations with 32,379 annotations. By applying supervised fine-tuning and taxonomy-aligned Group Relative Policy Optimization to Qwen3.5-4B, the resulting TACTutor improves 20.30% over its backbone on TACTBench, outperforming proprietary baselines while maintaining performance on external benchmarks. In a blinded study with 50 learners, it received the highest mean rating. Data, benchmark, and model weights are released.

2. ParVL: Parallel Scaling and Expandable Compute Allocation for Multimodal LLMs
   T2arXiv AI/CL/LG（论文）

   Existing scaling strategies for multimodal large language models typically expand model parameters or sequential inference computation, incurring high memory or latency overhead, and fail to adjust the rigid compute allocation between vision transformers and language models. ParVL introduces a parallel scaling framework that reuses existing ViT and LLM backbone parameters across multiple vision and language branches, enabling flexible compute allocation between modalities under a fixed backbone budget. After full-parameter supervised fine-tuning on roughly 13B tokens, ParVL outperforms same-recipe single-branch baselines in overall multimodal performance, with optimal vision-language allocation varying across tasks. Code is available on GitHub.

—

1. Anthropic Builds AI Chip Design Team to Co-Optimize Hardware and Models
   T2TechCrunch：AI（RSS）

   Anthropic, the company behind Claude, is assembling an in-house AI chip design team to develop custom silicon. The company announced that it will co-design hardware and models to achieve faster and more efficient AI performance, a move that could reduce its dependence on Nvidia GPUs and cloud providers. This strategic pivot into infrastructure reflects a broader industry trend among frontier AI labs to vertically integrate their compute stacks. By customizing chips to their specific workloads, Anthropic aims to improve training and inference speeds, lower costs, and gain more control over its AI development roadmap. However, chip design is capital-intensive and time-consuming, requiring specialized engineering talent and long development cycles. The team is still in its early stages, with no details on chip architecture, manufacturing partners, or deployment timelines. This story is developing.

—

—