I am a grade 2022 undergraduate from Zhejiang University, and I hope to learn code in depth on github platform.
Repositories
March-7/genius-indirect-fixtures
Public static fixtures for controlled Genius indirect prompt injection tests
March-7/Roco-Kingdom-Free-Your-Hand
恶魔狼+星云使者，放在巨人裂谷北猫头鹰，启动该工具，将会对其进行自动瞄准与捕捉。本项目识别工具完全基于视觉。缺陷：1.可能会指向入战精灵，2.无法应对触发战斗的情况，3.陨石误判成精灵，4.本项目采集样本较少，目前不清楚各个设备的画质，光影等对识别效果的影响。
March-7/cadence-type
cadence-type
March-7/StickyToken
Sticking to the Mean: Detecting Sticky Tokens in Text Embedding Models [ACL 2025 Main]
March-7/gpu-occupy-cli
March-7/Red-Teaming-AiFChem-Agent
Red‑Teaming-AiFChem-Agent
March-7/observatory
A toolkit for describing model features and intervening on those features to steer behavior.
March-7/latentqa-exp
March-7/strategic-deception-detection
March-7/deception-representation-exp
March-7/minimind-v
🚀 「大模型」1小时从0训练26M参数的视觉多模态VLM！🌏 Train a 26M-parameter VLM from scratch in just 1 hours!
March-7/MESA_and_MASK
This directory contains the implementation of the M&M (Mesa & Mask) benchmark and evaluation framework for assessing deceptive behaviors in large language models.
March-7/MESA_and_MASK_Leaderboard_Gitpage
M&M_Leaderboard_Gitpage
March-7/open-r1
Fully open reproduction of DeepSeek-R1
March-7/llm-course
Course to get into Large Language Models (LLMs) with roadmaps and Colab notebooks.
March-7/minimind
🚀🚀 「大模型」2小时完全从0训练26M的小参数GPT！🌏 Train a 26M-parameter GPT from scratch in just 2h!