This skill provides reference guidance for citation verification in academic writing. Use when the user asks about "citation verification best practices", "how to verify references", "preventing fake citations", or needs guidance on citation accuracy. This skill supports ml-paper-writing by providing detailed verification principles and common error patterns.

Guides the user through an interactive conversation to define their research project, then generates research_brief.json and tasks.json. Use when starting a new project, when no research_brief.json exists, when the user wants to start from a specific pipeline stage, or when the user wants to redefine their research pipeline.

Structured manuscript/grant review with checklist-based evaluation. Use when writing formal peer reviews with specific criteria methodology assessment, statistical validity, reporting standards compliance (CONSORT/STROBE), and constructive feedback. Best for actual review writing, manuscript revision. For evaluating claims/evidence quality use scientific-critical-thinking; for quantitative scoring frameworks use scholar-evaluation.

Facilitates structured brainstorming sessions, conducts comprehensive research, and generates creative solutions using proven frameworks. Trigger keywords - brainstorm, ideate, research, SCAMPER, SWOT, mind map, creative, explore ideas, market research, competitive analysis, innovation, problem solving, feature generation

Help professors and researchers write, revise, adapt, and polish grant proposals for US agencies (NSF, NIH, DOE, DARPA, NASA) and Chinese agencies (NSFC 国自然). Use this skill whenever the user mentions grants, proposals, funding applications, 基金申请, 本子, R01, R21, CAREER, 面上, 青年基金, specific aims, 立项依据, broader impacts, or wants to plan, draft, review, or resubmit any research funding proposal — even if they don't explicitly say "grant". Also use this skill when the user wants to adapt a previous proposal for a new submission. Six-phase workflow: profiling → planning → drafting → quality review → simulated peer review → submission prep.

Creates implementation plan, writes project code with judge feedback loop, and submits final experiment run. Use after code-survey in both Idea and Plan branches.

This skill should be used when the user asks to "analyze experimental results", "generate results section", "statistical analysis of experiments", "compare model performance", "create results visualization", or mentions connecting experimental data to paper writing. Provides comprehensive guidance for analyzing ML/AI experimental results and generating paper-ready content.

Comprehensive research assistant that synthesizes information from multiple sources with citations. Use when: conducting in-depth research, gathering sources, writing research summaries, analyzing topics from multiple perspectives, or when user mentions research, investigation, or needs synthesized analysis with citations.

Acquires missing code repositories for the selected idea (Phase A) and conducts comprehensive code survey mapping academic concepts to implementations (Phase B). Outputs acquired_code_repos, updated_prepare_res, and model_survey for downstream use by inno-implementation-plan.

Serverless GPU cloud platform for running ML workloads. Use when you need on-demand GPU access without infrastructure management, deploying ML models as APIs, or running batch jobs with automatic scaling.

Serves LLMs with high throughput using vLLM's PagedAttention and continuous batching. Use when deploying production LLM APIs, optimizing inference latency/throughput, or serving models with limited GPU memory. Supports OpenAI-compatible endpoints, quantization (GPTQ/AWQ/FP8), and tensor parallelism.

Optimizes LLM inference with NVIDIA TensorRT for maximum throughput and lowest latency. Use for production deployment on NVIDIA GPUs (A100/H100), when you need 10-100x faster inference than PyTorch, or for serving models with quantization (FP8/INT4), in-flight batching, and multi-GPU scaling.

Fast structured generation and serving for LLMs with RadixAttention prefix caching. Use for JSON/regex outputs, constrained decoding, agentic workflows with tool calls, or when you need 5× faster inference than vLLM with prefix sharing. Powers 300,000+ GPUs at xAI, AMD, NVIDIA, and LinkedIn.

Runs LLM inference on CPU, Apple Silicon, and consumer GPUs without NVIDIA hardware. Use for edge deployment, M1/M2/M3 Macs, AMD/Intel GPUs, or when CUDA is unavailable. Supports GGUF quantization (1.5-8 bit) for reduced memory and 4-10× speedup vs PyTorch on CPU.

Expert guidance for fine-tuning LLMs with LLaMA-Factory - WebUI no-code, 100+ models, 2/3/4/5/6/8-bit QLoRA, multimodal support

Evaluates LLMs across 100+ benchmarks from 18+ harnesses (MMLU, HumanEval, GSM8K, safety, VLM) with multi-backend execution. Use when needing scalable evaluation on local Docker, Slurm HPC, or cloud platforms. NVIDIA's enterprise-grade platform with container-first architecture for reproducible benchmarking.

Accelerate LLM inference using speculative decoding, Medusa multiple heads, and lookahead decoding techniques. Use when optimizing inference speed (1.5-3.6× speedup), reducing latency for real-time applications, or deploying models with limited compute. Covers draft models, tree-based attention, Jacobi iteration, parallel token generation, and production deployment strategies.

Train Mixture of Experts (MoE) models using DeepSpeed or HuggingFace. Use when training large-scale models with limited compute (5× cost reduction vs dense models), implementing sparse architectures like Mixtral 8x7B or DeepSeek-V3, or scaling model capacity without proportional compute increase. Covers MoE architectures, routing mechanisms, load balancing, expert parallelism, and inference optimization.

Reduce LLM size and accelerate inference using pruning techniques like Wanda and SparseGPT. Use when compressing models without retraining, achieving 50% sparsity with minimal accuracy loss, or enabling faster inference on hardware accelerators. Covers unstructured pruning, structured pruning, N:M sparsity, magnitude pruning, and one-shot methods.

Merge multiple fine-tuned models using mergekit to combine capabilities without retraining. Use when creating specialized models by blending domain-specific expertise (math + coding + chat), improving performance beyond single models, or experimenting rapidly with model variants. Covers SLERP, TIES-Merging, DARE, Task Arithmetic, linear merging, and production deployment strategies.