882 technical terms and definitions
chain rule of probability, causal language model, teacher forcing, left to right generation
An autoregressive model predicts each element of a sequence conditioned on every element that came before it, and the whole approach rests on one piece of exact, lossless algebra…
autoregressive model sequential generation and factorization order, autoregressive model sequential, model sequential generation, sequential generation factorization
Autoregressive Model Sequential Generation and Factorization Order Introduction & Motivation Autoregressive Model Sequential Generation and Factorization Order addresses a central…
Autoregressive Model Fitting and Selection Introduction & Motivation Autoregressive Model Fitting and Selection addresses a central problem in timeseries forecasting…
neural architecture
AutoSlim is an automated approach to finding optimal channel configurations for slimmable networks — instead of using uniform width multipliers (0.25×, 0.5×, etc.), AutoSlim…
autotuning systems for ml, autotuning systems ml, autotuning systems, systems ml
Autotuning Systems for ML Introduction & Motivation Autotuning Systems for ML addresses a central problem in ml-hardware process control and tuning-time-overhead prediction: how…
model optimization
AutoTVM is a TVM module that searches operator schedule configurations to maximize backend performance - It replaces manual schedule tuning with data-driven optimization. What Is…
Auxiliary Task Learning Introduction & Motivation Auxiliary Task Learning addresses a central problem in multitask learning across several related prediction objectives: how…
Auxiliary Task Learning (Extended Study) Introduction & Motivation Auxiliary Task Learning (Extended Study) addresses a central problem in multitask learning across several…
Auxiliary Task Learning (Extended Study) Introduction & Motivation Auxiliary Task Learning (Extended Study) addresses a central problem in multitask learning across several…
autonomous vehicle compute architecture, autonomous vehicle compute, vehicle compute architecture, autonomous vehicle, vehicle compute, compute architecture
Autonomous Vehicle Compute Architecture Introduction & Motivation Autonomous Vehicle Compute Architecture addresses a central problem in application process control…
autonomous vehicle computing, autonomous vehicle, vehicle computing
Autonomous Vehicle Computing Introduction & Motivation Autonomous Vehicle Computing addresses a central problem in application process control and latency prediction: how…
av computing
AV Computing Introduction & Motivation AV Computing addresses a central problem in application process control and latency prediction: how to Predict latency from av-comp process…
production
Availability is the percentage of time equipment is in a ready-to-run state, excluding periods when it is down for failures or planned service - it reflects mechanical…
manufacturing operations
Availability is the proportion of total time a system is capable of operating when required - It combines reliability and maintainability into an operational readiness metric.…
ha high availability, reliability high availability, fault tolerance
High availability (HA) is the design of a system to ensure it remains operational and accessible for a very high percentage of time, minimizing downtime even during hardware…
manufacturing operations
Availability Rate is the proportion of planned production time during which equipment is actually running - It captures downtime impact on usable capacity. What Is Availability…
avalanche breakdown multiplication factor prediction, avalanche breakdown multiplication, breakdown multiplication factor, multiplication factor prediction
Avalanche Breakdown Multiplication Factor Prediction with Machine Learning Introduction & Motivation Avalanche Breakdown Multiplication Factor Prediction with Machine Learning…
avalanche photodiode gain breakdown prediction, avalanche photodiode gain, photodiode gain breakdown, gain breakdown prediction, avalanche photodiode, photodiode gain
Avalanche Photodiode Gain Breakdown Prediction with Machine Learning Introduction & Motivation Avalanche Photodiode Gain Breakdown Prediction with Machine Learning addresses…
avalanche photodiode gain drift prediction machine learning, avalanche photodiode, gain drift prediction, apd, avalanche photodiode gain drift prediction, photodiode gain
Avalanche Photodiode Gain Drift Prediction with Machine Learning Introduction & Motivation Avalanche Photodiode Gain Drift Prediction with Machine Learning addresses a central…
unclamped inductive switching uis, edge termination avalanche current, avalanche hot spot failure
Avalanche Ruggedness Breakdown voltage tells you where avalanche breakdown starts; it says nothing about whether a device actually survives going through it. Avalanche ruggedness…
Average Treatment Effect Estimation Introduction & Motivation Average Treatment Effect Estimation addresses a central problem in causal inference and {target} analysis: how…
avl, supply chain & logistics
AVL is approved vendor list defining suppliers authorized for specific materials or components - Controlled vendor entries ensure purchases come from qualified and compliant…
avl tree strict balance, avl tree strict, tree strict balance, avl tree, tree strict, strict balance
AVL Tree Strict Balance Introduction & Motivation AVL Tree Strict Balance addresses a central problem in data-structures process control and height-limit-factor prediction: how…
aws trainium, what is aws trainium
A family of AWS-designed AI accelerators. AWS Trainium is Amazon Web Services’ custom accelerator family for machine-learning workloads, delivered as part of AWS computing…
flops, training compute, model flops utilization, compute budget
Compute, in AI, means floating-point operations — FLOPs — the raw arithmetic a training run or inference request actually consumes, and it functions as the field's basic unit…
machine intelligence, ai, ai systems, data analytics, machine learning
artificial intelligence is the broad field of building machines that perceive, reason, learn, generate, plan, and act toward goals. AI spans symbolic systems, machine learning…
sacrificial oxide removal, bitline parasitic capacitance reduction, DRAM air gap isolation, selective wet etch bitline
Air-gap spacer removal does the opposite of almost every other spacer step in this flow — instead of depositing and shaping a material that stays in the finished chip, it…
grown junction 1951 attach bar ends, solder thermal budget junction diffusion, cumulative thermal budget dopant diffusion, grown junction transistor solder joint
Attach the Two Bar Ends: Soldering Without Re-Grading a Junction Already Grown Sharp Every junction in this bar was already formed, with whatever sharpness the melt-mixing…
hybrid attention-ssm, hybrid attention ssm, attention state space hybrid, attention-ssm hybrid, jamba, zamba, samba, hybrid transformer-ssm
Mamba-2 is the 2024 successor to the Mamba selective state-space model, and its importance is less about a bigger benchmark number than about a unifying idea: it shows that…
sram six transistor step 14 mismatch margining statistical design, sram mismatch statistical design margin
Address Variation and Mismatch in the Cross-Coupled Pair at Scale 14. Address Variation and Mismatch in the Cross-Coupled Pair at Scale Design teams respond to the mismatch…
value alignment, rlhf, rlaif, constitutional ai, scalable oversight, model safety
AI alignment is the effort to make AI-system behavior reliably reflect intended goals, constraints, and human values under realistic deployment conditions. Capability alone does…
subtractive interconnect process, aluminum vs copper interconnect, chlorine plasma etch aluminum, metal wiring fabrication, aluminum etch profile
Aluminum metal etch is the process step that patterns aluminum wiring into the precise interconnect shapes a chip design calls for, and it represents an older interconnect…