29,259 technical terms and definitions
group relative policy optimization, llm reward free rl, process reward model training, math reasoning rl
GRPO and RL for LLM Reasoning is the reinforcement learning training paradigm that directly optimizes large language models for verifiable reasoning tasks — particularly…
gated recurrent unit, Cho 2014, update gate, reset gate, GRU vs LSTM, gated rnn, recurrent neural network
A gated recurrent unit, or GRU, is a recurrent neural network cell that reads a sequence one step at a time while carrying a hidden state, and that uses two learned gates…
gru gated recurrent, gru gated, gated recurrent
GRU Gated Recurrent Introduction & Motivation GRU Gated Recurrent addresses a central problem in timeseries process control and gru-reset-update-gate prediction: how to Predict…
gru gated, gated recurrent, cnn gru lstm passed rnn, gru gated recurrent, gated recurrent unit, recurrent unit
GRU: Gated Recurrent Unit Introduction & Motivation GRU (Gated Recurrent Unit): simplified LSTM alternative. Reset gate: selective past information. Update gate: control…
recommendation systems
GRU4Rec is a session-based recommendation model using gated recurrent units over click sequences - Sequential hidden states encode short-term intent and predict next likely items…
evaluation
gSCAN (grounded SCAN) is the benchmark for systematically testing compositional generalization in visually grounded instruction following — placing an agent in a grid world where…
gsm baseband, what is gsm baseband
GSM baseband is the processing that prepares and interprets digital information for the GSM radio link. It is distinct from simply amplifying a radio-frequency waveform.…
evaluation
GSM8K (Grade School Math 8K) is a benchmark of 8,500 high-quality, linguistically diverse grade school math word problems requiring 2-8 step reasoning to solve, designed…
math benchmark, word problems
GSM8K is a benchmark dataset of 8,500 linguistically diverse grade school math word problems designed to evaluate mathematical reasoning capabilities in language models. What Is…
gsm8k, evaluation
GSM8K is a grade-school math word-problem benchmark used to evaluate multi-step numerical reasoning - It is a core method in modern AI evaluation and safety execution workflows.…
Gradient-Based Sample Selection GSS for Memory Management Introduction & Motivation Gradient-Based Sample Selection GSS for Memory Management addresses a central problem…
gtn, graph neural networks
GTN is graph transformer network that learns soft meta-relational paths in heterogeneous graphs - It automates metapath construction instead of relying solely on hand-crafted…
guó, 国字, guó meaning, country nation
国 拼音：guó 意思：country / nation（国家） 例词： 中国 — China 国家 — country / nation 国王 — king 外国 — foreign country 例句： 中国很大。Zhōngguó hěn dà. — China is very big. 你是哪国人？Nǐ shì nǎ guó rén? — What…
qlora, efficient
Guanaco is a landmark family of fine-tuned language models that demonstrated QLoRA (Quantized Low-Rank Adaptation) could produce chatbot performance rivaling ChatGPT while…
ring, isolation, techniques, substrate, coupling
Guard Ring and Isolation Techniques is protective structures surrounding sensitive circuits reducing substrate and electromagnetic coupling — essential for noise-sensitive analog…
design
Guard band is a deliberate safety margin between nominal operation and failure limits — ensuring process drift, temperature swings, and aging don't push systems into unsafe…
guard band test limit optimization prediction, guard band test, band test limit, test limit optimization, limit optimization prediction, guard band, band test, test limit
Guard Band Test Limit Optimization Prediction with Machine Learning Introduction & Motivation Guard Band Test Limit Optimization Prediction with Machine Learning addresses…
design, latchup prevention, substrate resistance, minority carriers
Guard ring effectiveness depends on ring width, placement distance from the protected transistors, and ring continuity, since a guard ring with gaps or placed too far from…
design, latch-up, substrate noise, well taps, minority carrier collection
Guard rings are continuous diffusion rings, P+ tied to ground surrounding NMOS transistors and N+ tied to supply surrounding PMOS transistors, placed specifically to collect stray…
guardband reduction and yield-reliability trade-off, guardband reduction yield-reliability, reduction yield-reliability trade-off, guardband reduction, reduction yield-reliability
Guardband Reduction and Yield-Reliability Trade-Off Introduction & Motivation Guardband Reduction and Yield-Reliability Trade-Off addresses a central problem in design process…
design
Guardbanding is the intentional addition of design or test margin so products remain compliant under process, voltage, temperature, and aging uncertainty - it protects field…
advanced test & probe
Guardbanding is the practice of tightening test limits beyond nominal specifications to reduce defect escapes - It adds safety margin against measurement uncertainty, drift…
boundary, limit
Guardrails are deterministic checks deliberately wrapped around a fundamentally stochastic underlying system, and the specific reason guardrails exist as their own entirely…
ai safety
Guardrails commonly check a model's generated output against specific defined content policies, factual-accuracy constraints, or required structured-output formatting…
framework
Guardrails AI is the open-source framework for adding validation, safety checks, and structural constraints to LLM outputs — providing programmable guardrails that verify language…
structured, microsoft
Guidance is a Microsoft-developed programming language for constraining and controlling LLM outputs with guaranteed structure — replacing probabilistic prompt engineering…
framework
Guidance is the constraint-based language model programming framework by Microsoft that enables precise control over LLM output structure through interleaved generation…
Guidance-Free Diffusion Sampling Introduction & Motivation Guidance-Free Diffusion Sampling addresses a central problem in generative modeling of diffusion data: how to Learn…
Guidance-Free Diffusion Sampling (Extended Study) Introduction & Motivation Guidance-Free Diffusion Sampling (Extended Study) addresses a central problem in generative modeling…
Guidance-Free Diffusion Sampling (Extended Study) Introduction & Motivation Guidance-Free Diffusion Sampling (Extended Study) addresses a central problem in generative modeling…
generative models
Guidance scale is the numeric factor in classifier-free guidance that sets the strength of conditional steering during denoising - it is one of the most sensitive controls…
multimodal ai
Guidance Scale is the control parameter determining strength of conditional guidance during diffusion sampling - It directly affects prompt fidelity and output variability. What…
interpretability
Guided Backprop is a visualization method that modifies backpropagation to pass only positive gradients through ReLU layers - It produces sharper feature-importance maps than…
Guided Backpropagation High-Resolution Saliency Introduction & Motivation Guided Backpropagation High-Resolution Saliency addresses a central problem in explainability…
explainable ai
Guided Backpropagation is a visualization technique that modifies the standard backpropagation to produce sharper, more interpretable saliency maps — by additionally masking out…
Guided Backpropagation Introduction & Motivation Guided Backpropagation addresses a central problem in interpretability and explanation of xai predictions: how to Produce…
Guided Backpropagation (Extended Study) Introduction & Motivation Guided Backpropagation (Extended Study) addresses a central problem in interpretability and explanation of xai…
Guided Backpropagation (Extended Study) Introduction & Motivation Guided Backpropagation (Extended Study) addresses a central problem in interpretability and explanation of xai…
Guided Backpropagation (Extended Study) Introduction & Motivation Guided Backpropagation (Extended Study) addresses a central problem in interpretability and explanation of xai…
guided local search, guided local, local search
Guided Local Search Introduction & Motivation Guided Local Search addresses a central problem in ml process control and penalty-function-design prediction: how to Predict…
Guided Policy Search with Trajectory Optimization Introduction & Motivation Guided Policy Search with Trajectory Optimization addresses a central problem in reinforcement policy…
packaging
Gull-wing leads is the outward and downward bent lead form used in many surface-mount packages to create visible solder joints - they offer good inspectability and compliance…
gustafson's law, weak scaling, scaled speedup, parallel scaling
Gustafson's law models scaled speedup when the parallel portion of a problem grows with available processors while serial time remains comparatively fixed. It explains why larger…
guiding center drift velocity plasma kinetics, drift wave microturbulence etch reactor, parametric decay instability plasma sheath, gyroaverage Vlasov maxwell kinetic pde
Gyrokinetic Vlasov Transport Equations, Drift-Wave Microturbulence, and Non-Linear Instability Kinetics in EUV and ICP Etch Chambers Executive Summary Inductively Coupled Plasma…
hé, 和字, hé meaning, and with
和 拼音：hé 意思：and / with / peace（和，连接词） 例词： 你和我 — you and I 和平 — peace 和好 — make up / reconcile 例句： 我和你是朋友。Wǒ hé nǐ shì péngyǒu. — You and I are friends. 猫和狗。Māo hé gǒu. — Cat…
hē, 喝字, hē meaning, drink
喝 拼音：hē 意思：drink（喝） 例词： 喝水 — drink water 喝茶 — drink tea 喝汤 — drink soup 例句： 我喝水。Wǒ hē shuǐ. — I drink water. 你喝茶吗？Nǐ hē chá ma? — Do you drink tea? 跟读：hē hē hē
design
The H-gate is a transistor layout geometry for partially-depleted silicon-on-insulator (PD-SOI) devices where the gate electrode is shaped like the letter H, with the horizontal…
hǎo, 好字, hǎo meaning, good
好 拼音：hǎo（好的）/ hào（爱好，喜好） 意思：good / well / fine（好） 例词： 你好 — hello 很好 — very good 好吃 — delicious 好看 — good-looking 例句： 你好吗？Nǐ hǎo ma? — How are you? 这个很好。Zhège hěn hǎo. — This is…
hybrid parameters, h parameters, hie hfe hoe hre, two-port network, hybrid matrix, transistor small-signal model, Bell Labs two-port analysis
H-parameters, or hybrid parameters, are a set of four numbers that describe a transistor as a linear two-port network for small signals: the input impedance hi, the reverse…
H-Score Information-Theoretic Representation Transferability Introduction & Motivation H-Score Information-Theoretic Representation Transferability addresses a central problem…