<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Jingchen Ni</title><description>M.S. Student, Tsinghua University.</description><link>https://tonybotni.github.io/</link><item><title>[Publication] MG-RWKV: Multi-Grained Context-Aware RWKV for Temporal Forgery Localization</title><link>https://tonybotni.github.io/publications/mg-rwkv/</link><guid isPermaLink="true">https://tonybotni.github.io/publications/mg-rwkv/</guid><description>A multi-granularity RWKV framework for Temporal Forgery Localization that achieves state-of-the-art performance with linear O(T) complexity.</description><pubDate>Wed, 01 Jul 2026 00:00:00 GMT</pubDate></item><item><title>A Systems Design Comparison of LLM RL Post-training Frameworks</title><link>https://tonybotni.github.io/posts/llm-rl-post-training-frameworks/</link><guid isPermaLink="true">https://tonybotni.github.io/posts/llm-rl-post-training-frameworks/</guid><description>A deep systems comparison of verl, SLIME, AReaL, and ms-swift across rollout architecture, async control, partial rollout, weight sync, and platform trade-offs.</description><pubDate>Tue, 16 Jun 2026 00:00:00 GMT</pubDate></item><item><title>Why Prefill and Decode Should Not Share a GPU</title><link>https://tonybotni.github.io/posts/pd-disaggregation-llm-inference/</link><guid isPermaLink="true">https://tonybotni.github.io/posts/pd-disaggregation-llm-inference/</guid><description>A systems-oriented introduction to Prefill-Decode disaggregation in LLM inference, from KV cache physics and roofline intuition to vLLM and SGLang implementation trade-offs.</description><pubDate>Tue, 16 Jun 2026 00:00:00 GMT</pubDate></item><item><title>[Publication] DeskCraft: Benchmarking Desktop Agents on Professional Workflows and Human-in-the-Loop Collaboration</title><link>https://tonybotni.github.io/publications/deskcraft/</link><guid isPermaLink="true">https://tonybotni.github.io/publications/deskcraft/</guid><description>A desktop GUI agent benchmark targeting long-horizon professional workflows and proactive human-agent collaboration.</description><pubDate>Mon, 01 Jun 2026 00:00:00 GMT</pubDate></item><item><title>[Publication] Cognitive Kernel-Pro: A Framework for Deep Research Agents and Agent Foundation Models Training</title><link>https://tonybotni.github.io/publications/ck-pro/</link><guid isPermaLink="true">https://tonybotni.github.io/publications/ck-pro/</guid><description>A fully open-source multi-module agent framework for deep research agents and agent foundation models training.</description><pubDate>Wed, 01 Apr 2026 00:00:00 GMT</pubDate></item><item><title>[Publication] AdaMem: Adaptive User-Centric Memory for Long-Horizon Dialogue Agents</title><link>https://tonybotni.github.io/publications/adamem/</link><guid isPermaLink="true">https://tonybotni.github.io/publications/adamem/</guid><description>An adaptive user-centric memory framework for long-horizon dialogue agents.</description><pubDate>Sun, 01 Mar 2026 00:00:00 GMT</pubDate></item><item><title>[Publication] FCL-COD: Weakly Supervised Camouflaged Object Detection with Frequency-aware and Contrastive Learning</title><link>https://tonybotni.github.io/publications/fcl-cod/</link><guid isPermaLink="true">https://tonybotni.github.io/publications/fcl-cod/</guid><description>A frequency-aware and contrastive learning-based weakly-supervised camouflaged object detection framework.</description><pubDate>Sun, 01 Mar 2026 00:00:00 GMT</pubDate></item><item><title>[Publication] Natural-Language Agent Harnesses</title><link>https://tonybotni.github.io/publications/nlahs/</link><guid isPermaLink="true">https://tonybotni.github.io/publications/nlahs/</guid><description>Externalizing agent harness control logic as portable natural-language artifacts with a shared runtime.</description><pubDate>Sun, 01 Mar 2026 00:00:00 GMT</pubDate></item><item><title>[Publication] SAHSR: Semantic Alignment and Hard-Sample Retraining Framework for Cross-Modal Person Re-Identification</title><link>https://tonybotni.github.io/publications/sahsr/</link><guid isPermaLink="true">https://tonybotni.github.io/publications/sahsr/</guid><description>A framework addressing semantic mismatch and hard-sample discrimination in Visible-Infrared Person Re-Identification.</description><pubDate>Sat, 01 Mar 2025 00:00:00 GMT</pubDate></item><item><title>[Publication] IteRPrimE: Zero-shot Referring Image Segmentation with Iterative Grad-CAM Refinement and Primary Word Emphasis</title><link>https://tonybotni.github.io/publications/iterprime/</link><guid isPermaLink="true">https://tonybotni.github.io/publications/iterprime/</guid><description>A method for zero-shot referring image segmentation with iterative Grad-CAM refinement and primary word emphasis.</description><pubDate>Sat, 01 Feb 2025 00:00:00 GMT</pubDate></item></channel></rss>