Claude Skills by VectorSpaceLab

"Maintain decontamination hygiene, clean-training-data safety,

"Run lm-evaluation-harness evaluations through the lm-eval CLI,

"Choose, install, configure, extend, and troubleshoot LM Evaluation

"Understand, validate, summarize, compare, and safely route

"Create, validate, and debug lm-evaluation-harness YAML tasks, task

"Use LMDeploy to run offline LLM/VLM inference, serve

"Tune LMDeploy PyTorch/TurboMind backend configs and extend PyTorch

"Use LMDeploy pipeline APIs and chat CLI for offline text LLM

"Plan LMDeploy Lite quantization, KV-cache quantization, and

"Serve LMDeploy models through OpenAI-compatible,

"Use LMDeploy vision-language and multimodal APIs for media

"Routes LMFlow dataset, training, inference, evaluation, alignment,

"Helps with LMFlow dataset schemas, conversation templates,

"Helps with LMFlow generation, evaluation, benchmark, and optional

"Helps with LMFlow image-text datasets, visual chat flows,

"Helps with LMFlow reward modeling, DPO, DPOv2, iterative DPO,

"Helps build LMFlow fine-tuning and optimization commands for full

"Router for using lmms-eval to run evaluations, inspect models and

"Guide for running lmms-eval from the CLI or Python API, including

"Guide for resolving and extending lmms-eval model backends,

"Guide for the lmms-eval HTTP server, Python clients, MCP tooling,

"Guide for authoring and debugging lmms-eval task YAMLs, utils.py

"Route LMOps paper-code workflows for prompt optimization,

"Route AdaptLLM, Instruction Pre-Training, PDS data selection,

"MiniLLM, DPKD, and Tuna distillation and ranking-finetuning

"Operate LMOps example retrieval, many-shot prompting, and ICL

"Operate ProTeGi text prompt optimization and Promptist

"Operate CoRAG chain-of-retrieval augmented generation and LLMA

"Plan safe VeRL/Ray/vLLM experiential-learning workflows for OEL,

"Routes agents through Logparser log-template extraction, parser

"Guides Logparser benchmark runs, dataset evaluation, and

"Guides Logparser log-template extraction, ordinary parser

"Guides Logparser workflows that need import shims, compilers,

"Use the LoRA repository and loralib package to add low-rank

"Integrate loralib into PyTorch modules, select trainable adapter

"Prepare, train, decode, and evaluate the LoRA GPT-2 data-to-text

"Construct and troubleshoot LoRA fine-tuning and evaluation

"Guides LTP Chinese NLP package workflows for Python pipelines,

"Guides LTP legacy perceptron and ltp_extension workflows for fast

"Guides high-level Python LTP pipeline workflows for model loading,

"Guides Rust ltp crate and ltp-cffi workflows for legacy

"Guides optional ltp_core training, evaluation, Hydra

"Routes the repo's long-term forecasting, statistical baselines,

"Routes FEDformer training, comparison, and sweep workflows for the

"Train, test, predict, sweep, and visualize the core forecasting

"Routes Pyraformer long-range forecasting, single-step forecasting,

"Run and troubleshoot Naive, GBRT, ARIMA, and SARIMA statistical

"Routes LTX-2 audio-video generation, training, data preparation,

"Helps agents build custom LTX-2 core component code with ltx_core

"Prepares LTX training datasets by validating manifests, drafting