Skip to content

Claude Skills by VectorSpaceLab

github.com/VectorSpaceLab

6,028 skillsA× 6,018B× 8D× 213 installs7,823 views

Sparse Replay TrainerATrain an online text classifier with sparse replay updates from episodic memory.testingpythonVotes: 0GitHub stars: 247
Flan Heldout Instruction EvaluatorAEvaluate held-out direct and CoT instruction examples, compute accuracy delta, and emit FLAN recovery mechanism checks.testingpythonVotes: 0GitHub stars: 247
Flan Instruction Mixture BuilderABuild FLAN-style multi-source instruction-finetuning mixtures while excluding held-out evaluation tasks and auditing source/CoT coverage.testingpythonVotes: 0GitHub stars: 247
Flan Instruction Prompt FormatterARender FLAN-style direct, few-shot, and chain-of-thought instruction prompt/completion pairs without leaking hidden answers into prompts.toolspythonVotes: 0GitHub stars: 247
Flan Reduced Instruction Finetuning LoopAExecute a bounded optimizer-based proxy for FLAN instruction finetuning and record loss plus parameter-change evidence.testingpythonVotes: 0GitHub stars: 247
Instruction FormattingARender FLAN-style instruction examples with direct, exemplar, and chain-of-thought formatting controls.toolspythonVotes: 0GitHub stars: 247
Lora Parameter FreezingAApply LoRA parameter freezing and compact checkpoint filtering for reduced adaptation experiments and audits.testingpythonVotes: 0GitHub stars: 247
Lora Recovery EvaluationAAssemble and validate a soft-mode reduced LoRA recovery with executable mechanism evidence and source-boundary logs.researchpythonVotes: 0GitHub stars: 247
Lora Training StepARun a deterministic reduced LoRA training step that updates only low-rank factors and records loss evidence.researchpythonVotes: 0GitHub stars: 247
Low Rank Adapter LayerAConstruct LoRA low-rank linear adapters and verify merged inference equivalence without relying on the original implementation repository.testingpythonVotes: 0GitHub stars: 247
Generative Latent MirAScore latent replay candidates with KL drift, entropy confidence penalties, and diversity filtering for MIR.toolsVotes: 0GitHub stars: 247
Mir Recovery EvaluationARun bounded MIR recovery experiments and emit accuracy, forgetting, traces, and mechanism evidence.toolsVotes: 0GitHub stars: 247
Online Stream MemoryABuild bounded online continual-learning streams, replay memories, and forgetting ledgers for MIR-style experiments.testingpythonVotes: 0GitHub stars: 247
Virtual Update InterferenceARank replay candidates by loss increase under a virtual incoming-batch update for MIR selection.testingVotes: 0GitHub stars: 247
Mmlu Answer ScoringACompute MMLU option-label accuracy and confidence calibration gap from model predictions.toolspythonVotes: 0GitHub stars: 247
Mmlu Fewshot PromptingABuild MMLU zero-shot and five-shot prompts while hiding the test answer from the model query.toolspythonVotes: 0GitHub stars: 247
Mmlu Item SchemaAValidate and canonicalize MMLU-style four-option subject examples for evaluation workflows.toolspythonVotes: 0GitHub stars: 247
Mmlu Recovery HarnessARun a bounded MMLU proxy recovery using generated item, prompting, and scoring skills.toolspythonVotes: 0GitHub stars: 247
P3 MaterializationAMaterialize PromptSource/P3-style template collections into prompted dataset records with template identity, metadata propagation, coverage counts, and rendering diagnostics.researchpythonbashVotes: 0GitHub stars: 247
Prompt Iteration ViewerAApply PromptSource-style templates across multiple examples and summarize browse, sourcing, and review diagnostics for prompt iteration.researchpythonbashVotes: 0GitHub stars: 247
Prompt Metadata QualityAValidate PromptSource-style prompt metadata and community quality constraints for natural-language prompts, metrics, answer choices, and target format.researchpythonbashVotes: 0GitHub stars: 247
Prompt Template RenderingARender PromptSource-style Jinja prompt templates into natural-language input/target pairs with answer choices, deterministic choices, skip handling, and structured diagnostics.toolspythonbashVotes: 0GitHub stars: 247
Passnet FeedbackAFast, deterministic feedback for PassNet work: pre-flight pattern/match verification WITHOUT burning a GPU evaluation, per-node bottleneck analysis of a sample's graphs, and eval-log parsing into per-variant status + estimated score + failure classification. Use BEFORE every GPU evaluation (check_pattern), at the START of a sample (analyze_graph), and AFTER every evaluation (parse_eval_log).developmentpythonbashVotes: 0GitHub stars: 247
Passnet OrchestrateARound/batch planning and decision support for PassNet graph optimization. Use for multi-sample triage, worker allocation, eval-budget policy, and post-eval keep/revert decisions. For solving one concrete sample end to end, use passnet-solve as the entry skill; consult this skill only when you need broader planning or a specific decision gate.businesspythonrustVotes: 0GitHub stars: 247
Passnet Pattern FusionAAuthor PassNet pass files: write patterns that actually MATCH the FX graph, pick fusion regions (which ops to absorb into one Triton kernel), and structure multi-pass submissions with the mandatory shared-dispatch architecture. Use when creating or fixing pattern()/replacement_args()/replacement_func() files, when a pass "failed to match", or when deciding how to fuse multiple kernels/ops.developmentpythonrustVotes: 0GitHub stars: 247
Passnet SkillBPassNet GPU kernel optimization via compiler passes. Design and implement Triton-based optimization passes, create pass files under ./pass_dir/, self-evaluate with pass_evaluator, and iterate to maximize GPU speedup.developmentpythongoVotes: 0GitHub stars: 247
Passnet SolveBEND-TO-END playbook for solving one PassNet sample: analyze the computation graphs, decide the optimization strategy (what to fuse, what to replace, what to leave alone), drive the iteration loop, and maximize the sample score. This is the ENTRY skill — invoke it first for any PassNet optimization task; it tells you when to use passnet-pattern-fusion, passnet-triton-opt and passnet-feedback.ai-agentspythongoVotes: 0GitHub stars: 247
Passnet Triton OptAMake a single PassNet Triton kernel fast AND numerically faithful: performance model (when a replacement can win at all), block/grid/warp tuning, autotune policy, launch overhead, and per-op numeric recipes to pass the dtype baseline tolerances. Use when a pass matches and is correct but speedup below expected, or when correctness fails by small numeric margins.developmentpythonrustVotes: 0GitHub stars: 247