Agent Skills by 踹斯
Showing 86 vetted skills indexed across 1 GitHub repositories.
choice5346dspy
Program language model calls declaratively with DSPy for automatic prompt optimization.
choice5346openrlhf-training
Trains LLMs with RLHF algorithms (PPO, GRPO, RLOO, and DPO using Ray and vLLM.
choice5346nemo-guardrails
Implements NVIDIA's NeMo Guardrails for runtime LLM application safety using programmable rails in the CoLang DSL.
choice5346llamaindex
Build LLM applications with Retrieval-Augmented Generation over ingested documents.
choice5346deepspeed
Guide distributed training with DeepSpeed ZeRO stages and pipeline parallelism.
choice5346crewai-multi-agent
Create and manage multi-agent systems for autonomous AI collaboration.
choice5346moe-training
Train Mixture of Experts models using DeepSpeed and HuggingFace Transformers.
choice5346nnsight-remote-interpretability
Run neural network interpretability experiments on large PyTorch models via NDIF remote execution.
choice5346langsmith-observability
Trace, evaluate, and monitor LLM applications using the LangSmith SDK.
choice5346audiocraft-audio-generation
Generate music and sound effects from text prompts using AudioCraft.
choice5346nemo-curator
Curate multimodal LLM training data with GPU-accelerated deduplication and filtering.
choice5346huggingface-accelerate
Simplify distributed PyTorch training with a unified API for DDP, DeepSpeed, and FSDP.
choice5346ml-paper-writing
Generate academic paper drafts from research repositories with LaTeX formatting and citation verification.
choice5346chroma
Manage vector embeddings and metadata for semantic search applications.
choice5346prompt-guard
Detect prompt injections and jailbreak attempts in LLM inputs.
choice5346evaluating-llms-harness
Evaluate LLMs against over 60 academic benchmarks including MMLU and HumanEval.
choice5346llama-factory
Fine-tune LLMs with LLaMA-Factory WebUI and QLoRA quantization.
choice5346knowledge-distillation
Compress Llama-2-70b to Llama-2-7b knowledge via softmax temperature distillation.
choice5346simpo-training
Optimize LLM alignment with reference-free SimPO preference training.
choice5346slime-rl-training
Train LLMs with RL using Megatron-LM and SGLang.
choice5346blip-2-vision-language
Generate image captions and answer visual questions with BLIP-2.
choice5346pytorch-fsdp2
Implement PyTorch FSDP2 distributed training with parameter sharding and mixed precision.
choice5346implementing-llms-litgpt
Implement and fine-tune LLMs using Lightning AI's LitGPT framework.
choice5346segment-anything-model
Segment objects in images using point or box prompts.