庄表伟
Community@zhuangbiaowei · Shanghai, China
Agent Skills by 庄表伟
Showing 99 vetted skills indexed across 2 GitHub repositories.
zhuangbiaoweimaster-control-expert
Classify requests into governance-backed workflows with control plans and decision logs.
zhuangbiaoweidspy
Create declarative language model programs with automatic prompt optimization in DSPy.
zhuangbiaoweiopenrlhf-training
Train large language models with RLHF algorithms using Ray and vLLM.
zhuangbiaoweinemo-guardrails
Implement runtime safety guardrails for LLM applications using NeMo Guardrails and Colang 2.0 DSL.
zhuangbiaoweillamaindex
Build LLM applications with RAG using data connectors and vector indices.
zhuangbiaoweideepspeed
Optimize distributed training and inference for large-scale AI models.
zhuangbiaoweicrewai-multi-agent
Orchestrate role-based multi-agent systems for sequential and hierarchical task execution.
zhuangbiaoweimoe-training
Train Mixture of Experts models using DeepSpeed and HuggingFace Transformers.
zhuangbiaoweitest_script_type
Run Python scripts and capture their output and arguments.
zhuangbiaoweinnsight-remote-interpretability
Inspect and modify PyTorch model activations with optional NDIF remote execution.
zhuangbiaoweilangsmith-observability
Trace, evaluate, and monitor LLM applications with LangSmith.
zhuangbiaoweiaudiocraft-audio-generation
Generate music and sound effects from text using AudioCraft.
zhuangbiaoweinemo-curator
Curate multimodal LLM training datasets with GPU-accelerated deduplication and filtering.
zhuangbiaoweihuggingface-accelerate
Simplify distributed PyTorch training with a unified API for DDP, DeepSpeed, and FSDP.
zhuangbiaoweiml-paper-writing
Automate writing and formatting ML papers for NeurIPS, ICML, and ICLR.
zhuangbiaoweichroma
Manage an open-source embedding database for semantic search and retrieval.
zhuangbiaoweiprompt-guard
Detect prompt injections and jailbreak attempts in LLM inputs.
zhuangbiaoweievaluating-llms-harness
Evaluate LLMs across academic benchmarks like MMLU and GSM8K.
zhuangbiaoweillama-factory
Fine-tune LLMs with LLaMA-Factory WebUI and QLoRA quantization.
zhuangbiaoweiyoutube-downloader
Download YouTube videos and audio with yt-dlp in configurable formats.
zhuangbiaoweiknowledge-distillation
Compress large language models into smaller student models using knowledge distillation.
zhuangbiaoweisimpo-training
Optimize LLM alignment with reference-free SimPO preference optimization.
zhuangbiaoweislime-rl-training
Trains large language models with GRPO reinforcement learning using Megatron-LM and SGLang.
zhuangbiaoweiblip-2-vision-language
Bridges frozen image encoders with LLMs for vision-language tasks like image captioning, VQA, and image-text retrieval.