Agent Skills by 傅聖祐
Showing 99 vetted skills indexed across 1 GitHub repositories.
box755autoresearch
Orchestrate autonomous AI research workflows with two-loop experiment and synthesis cycles.
box755serving-llms-vllm
Deploy OpenAI-compatible LLM inference endpoints using vLLM with PagedAttention and continuous batching.
box755tensorrt-llm
Optimize LLM inference on NVIDIA GPUs with TensorRT-LLM runtime.
box755sglang
Serve structured LLM generations with RadixAttention prefix caching.
box755llama-cpp
Run llama.cpp with GGUF models for local LLM inference on CPU and non-NVIDIA hardware.
box755academic-plotting
Generate publication-quality ML paper figures from research context and experiment data.
box755presenting-conference-talks
Generate Beamer LaTeX and editable PPTX conference slides with speaker notes from a compiled research paper.
box755ml-paper-writing
Draft ML/AI paper sections for major conferences with verified citations.
box755systems-paper-writing
Generate reviewer-aligned section and paragraph blueprints for systems conference papers.
box755lambda-labs-gpu-cloud
Launch and operate Lambda Labs GPU cloud instances for ML training and inference.
box755skypilot-multi-cloud-orchestration
Orchestrate ML training and batch workloads across multiple cloud providers with SkyPilot task YAMLs.
box755modal-serverless-gpu
Deploy GPU-based machine learning functions and endpoints on Modal.
box755nemo-curator
Curate multimodal web-scale datasets with quality filtering, deduplication, and PII redaction.
box755ray-data
Stream dataset reading and distributed transformations for scalable ML data processing.
box755dspy
Build declarative language-model pipelines with DSPy Signatures and automatic optimization.
box755guidance
Constrain LLM generation with regex and grammar constraints for structured outputs.
box755outlines
Constrain LLM token decoding to schema-derived grammars for valid structured outputs.
box755instructor
Extract structured data from LLM outputs using Pydantic schemas.
box755evaluating-cosmos-policy
Evaluate NVIDIA Cosmos Policy on LIBERO and RoboCasa simulations.
box755audiocraft-audio-generation
Generate WAV music or sound effects from text prompts using Meta AudioCraft.
box755fine-tuning-openvla-oft
Fine-tune and evaluate OpenVLA-OFT/OFT+ for robot action generation with visual-language inputs.
box755blip-2-vision-language
Generate image captions and visual question answers with BLIP-2 models.
box755segment-anything-model
Segment arbitrary images via point, box, or mask prompts.
box755clip
Embed images and text into a shared space for zero-shot matching.