model-pruning
davila7Reduce 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.
gguf-quantization
davila7GGUF format and llama.cpp quantization for efficient CPU/GPU inference. Use when deploying models on consumer hardware, Apple Silicon, or when needing flexible quantization from 2-8 bit without GPU requirements.
hqq-quantization
davila7Half-Quadratic Quantization for LLMs without calibration data. Use when quantizing models to 4/3/2-bit precision without needing calibration datasets, for fast quantization workflows, or when deploying with vLLM or HuggingFace Transformers.
knowledge-distillation
davila7Compress large language models using knowledge distillation from teacher to student models. Use when deploying smaller models with retained performance, transferring GPT-4 capabilities to open-source models, or reducing inference costs. Covers temperature scaling, soft targets, reverse KLD, logit distillation, and MiniLLM training strategies.
awq-quantization
davila7Activation-aware weight quantization for 4-bit LLM compression with 3x speedup and minimal accuracy loss. Use when deploying large models (7B-70B) on limited GPU memory, when you need faster inference than GPTQ with better accuracy preservation, or for instruction-tuned and multimodal models. MLSys 2024 Best Paper Award winner.
coreml
Use when deploying custom ML models on-device, converting PyTorch models, compressing models, implementing LLM inference, or optimizing CoreML performance. Covers model conversion, compression, stateful models, KV-cache, multi-function models, MLTensor.
axiom-ios-ml
Use when deploying ANY machine learning model on-device, converting models to CoreML, compressing models, or implementing speech-to-text. Covers CoreML conversion, MLTensor, model compression (quantization/palettization/pruning), stateful models, KV-cache, multi-function models, async prediction, SpeechAnalyzer, SpeechTranscriber.
coreml-diag
CoreML diagnostics - model load failures, slow inference, memory issues, compression accuracy loss, compute unit problems, conversion errors.
unity-importer
Besty0728Asset import settings. Use when users want to configure texture, audio, or model import settings. Triggers: import settings, texture settings, audio settings, model settings, compression, max size, Unity compression.
context
ktg-oneCONTEXT: Cognitive Order Normalized in Transformer EXtract Truncated. Cross-model context handoff via Progressive Density Layering, MLDoE expert compression, Japanese semantic density, and Negentropic Coherence Lattice validation. Creates portable carry-packets that transfer cognitive state between AI sessions. Use when context reaches 80%, switching models, ending sessions, user says save, quicksave, handoff, transfer, continue later, /qs, /context, or needs session continuity.
solana-compression
tenequmBuild with ZK Compression on Solana using Light Protocol. Use when creating compressed tokens, compressed PDAs, or integrating ZK compression into Solana programs. Covers compressed account model, state trees, validity proofs, and client integration with Helius/Photon RPC.
llm-cost-optimizer
TerminalSkillsTrack, analyze, and reduce LLM API costs — model routing, prompt caching, semantic caching, and budget alerts. Use when someone asks to "reduce AI costs", "track LLM spending", "optimize API costs", "set up model routing", "cache LLM responses", "compare model costs", "set budget limits for AI", or "my OpenAI bill is too high". Covers cost tracking per feature/user, smart model routing (expensive model for hard tasks, cheap for easy), semantic caching, prompt compression, and budget alerting.
prompt-enhancer
samhvw8Prompt engineering and optimization for AI/LLMs. Capabilities: transform unclear prompts, reduce token usage, improve structure, add constraints, optimize for specific models, backward-compatible rewrites. Actions: improve, enhance, optimize, refactor, compress prompts. Keywords: prompt engineering, prompt optimization, token efficiency, LLM prompt, AI prompt, clarity, structure, system prompt, user prompt, few-shot, chain-of-thought, instruction tuning, prompt compression, token reduction, prompt rewrite, semantic preservation. Use when: improving unclear prompts, reducing token consumption, optimizing LLM outputs, restructuring verbose requests, creating system prompts, enhancing prompt clarity.
curiosity-driven
plurigridSchmidhuber's curiosity-driven learning: Intrinsic motivation via compression progress. Seek states that improve world model.
axiom-ios-ml-coreml
megastepUse when deploying custom ML models on-device, converting PyTorch models, compressing models, implementing LLM inference, or optimizing CoreML performance. Covers model conversion, compression, stateful models, KV-cache, multi-function models, MLTensor.
ax-coreml
KasempiternalOn-device ML with CoreML -- model conversion (PyTorch to CoreML), compression (palettization/quantization/pruning), stateful KV-cache for LLMs, multi-function models, MLTensor, async prediction, and diagnostics
axiom-ios-ml-coreml-diag
megastepCoreML diagnostics - model load failures, slow inference, memory issues, compression accuracy loss, compute unit problems, conversion errors.
axiom-ios-ml
megastepUse when deploying ANY machine learning model on-device, converting models to CoreML, compressing models, or implementing speech-to-text. Covers CoreML conversion, MLTensor, model compression (quantization/palettization/pruning), stateful models, KV-cache, multi-function models, async prediction, SpeechAnalyzer, SpeechTranscriber.