Use PEFT or Full-parameter to CPT/SFT/DPO/GRPO 600+ LLMs (Qwen3.6, DeepSeek-V4, GLM-5.1, InternLM3, Llama4, ...) and 300+ MLLMs (Qwen3-VL, Qwen3-Omni, InternVL3.5, Ovis2.5, GLM4.5v, Gemma4, Llava, Phi4, ...) (AAAI 2025).
A Next-Generation Training Engine Built for Ultra-Large MoE Models
[ICCV 2025] Are VLMs Ready for Autonomous Driving? An Empirical Study from the Reliability, Data, and Metric Perspectives
Higher performance OpenAI LLM service than vLLM serve: A pure C++ high-performance OpenAI LLM service implemented with GPRS+TensorRT-LLM+Tokenizers.cpp, supporting chat and function call, AI agents, distributed multi-GPU inference, multimodal capabilities, and a Gradio chat interface.
[ICML 2026 Oral] DOUBT: Decoupled Object-level Understanding and Bridging via vMF-based Trustworthiness for Hallucination Detection in MLLMs
PriorTR (ECCV 2026): training-free, prior-corrected visual token reduction for accelerating multimodal LLMs — image & video.
[ICML 2026] RLSF-V: Mitigating Hallucinations in MLLMs via Fuzzy Semantic Self-Feedback
Streamlit App Combining Vision, Language, and Audio AI Models
VLM (InternVL3) inference on Jetson Orin Nano (8GB) using TensorRT-LLM.
Automate Microsoft Account OAuth registration and token imports with this Chrome extension for the Microsoft Account Manager API.
Unified evaluation harness for Vision-Language Models — VQAv2, TextVQA, COCO, MMBench, MMBench-CN
ACL 2025, SDP workshop Shared Task Submission
Description: Domain-shift failure analysis of LLaVA-1.5-7B and InternVL2-8B across photos, charts, medical, and screenshots. 945 probes, 6 failure categories. Chart OCR: LLaVA 0% vs InternVL2 71%. Domain-dependent failure confirmed via chi-square (p<0.0001).
Unofficial PyTorch reproduction for InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Models.
Research validation copy for D-MoLE and InternVL-based modulation recognition experiments.
Comparative analysis of six Vision-Language Models (InternVL, Mini-Gemini, Qwen-VL series) on architecture, benchmark performance (MME), and data pipeline evolution. Research repository featuring a one-page summary – full paper available upon request.