Cedlijh
Community@ced3-han · New York / Beijing / Xian
PhD @ NYU, BS @ PKU, work on Generative Modeling
Agent Skills by Cedlijh
Showing 52 vetted skills indexed across 1 GitHub repositories.
Ced3-hanusing-superpowers
Invoke relevant skills before responding to any user message.
Ced3-hantdd-workflow
Enforce a Test-Driven Development workflow with unit, integration, and end-to-end tests.
Ced3-hanpaper-reading
Summarize research papers and extract key figures into a structured template.
Ced3-hanhuggingface-papers
Fetches Hugging Face paper markdown and structured metadata via API.
Ced3-hanplanning-with-files
Plan and track tasks using structured markdown files for progress and findings.
Ced3-hanopenrlhf-training
Run distributed RLHF training for large language models with Ray and vLLM.
Ced3-hanpython-testing
Implement test-driven development and automated testing in Python with pytest.
Ced3-hantest-driven-development
Implement test-driven development for machine learning code with targeted tests and minimal changes.
Ced3-hanproject-guidelines-example
Generate a project guideline template covering architecture, file structure, code patterns, testing, and deployment.
Ced3-haniterative-retrieval
Automate iterative context retrieval for multi-agent workflows.
Ced3-hansystematic-debugging
Guide a 4-phase debugging process for identifying and resolving software bugs.
Ced3-hanllama-factory
Guide fine-tuning of large language models with LLaMA-Factory WebUI.
Ced3-hansimpo-training
Optimize LLM preferences with reference-free SimPO training.
Ced3-hancoding-standards
Guide TypeScript, JavaScript, React, and Node.js development with universal coding standards.
Ced3-hanpytorch-fsdp2
Implement PyTorch FSDP2 sharding, mixed precision, and distributed checkpointing.
Ced3-hantraining-llms-megatron
Train large language models with NVIDIA Megatron-Core parallelism strategies.
Ced3-haneval-harness
Evaluate Claude Code sessions with capability and regression evals.
Ced3-hanpaper-finder
Search AI research papers across multiple sources and organize them into a knowledge base.
Ced3-hancreative-thinking-for-research
Generate novel research ideas using cognitive science frameworks for computer science and AI.
Ced3-hanfine-tuning-with-trl
Fine-tune LLMs with TRL reinforcement learning and HuggingFace Transformers.
Ced3-hangrpo-rl-training
Guide GRPO training with the TRL library and custom reward functions.
Ced3-hangguf-quantization
Convert HuggingFace models to GGUF format and quantize them for llama.cpp inference.
Ced3-hanawq-quantization
Automate 4-bit activation-aware quantization for large language models.
Ced3-hanxlsx
Automates Excel file operations and data analysis using pandas and LibreFormula.