Proposes EpG and OOI metrics showing agentic workflows use 4.33x more energy per successful goal than linear baselines due to orchestration structure.
hub Mixed citations
Carbon Emissions and Large Neural Network Training
Mixed citation behavior. Most common role is background (69%).
70
Pith papers citing it
Background
69% of classified citations
abstract
The computation demand for machine learning (ML) has grown rapidly recently, which comes with a number of costs. Estimating the energy cost helps measure its environmental impact and finding greener strategies, yet it is challenging without detailed information. We calculate the energy use and carbon footprint of several recent large models-T5, Meena, GShard, Switch Transformer, and GPT-3-and refine earlier estimates for the neural architecture search that found Evolved Transformer. We highlight the following opportunities to improve energy efficiency and CO2 equivalent emissions (CO2e): Large but sparsely activated DNNs can consume <1/10th the energy of large, dense DNNs without sacrificing accuracy despite using as many or even more parameters. Geographic location matters for ML workload scheduling since the fraction of carbon-free energy and resulting CO2e vary ~5X-10X, even within the same country and the same organization. We are now optimizing where and when large models are trained. Specific datacenter infrastructure matters, as Cloud datacenters can be ~1.4-2X more energy efficient than typical datacenters, and the ML-oriented accelerators inside them can be ~2-5X more effective than off-the-shelf systems. Remarkably, the choice of DNN, datacenter, and processor can reduce the carbon footprint up to ~100-1000X. These large factors also make retroactive estimates of energy cost difficult. To avoid miscalculations, we believe ML papers requiring large computational resources should make energy consumption and CO2e explicit when practical. We are working to be more transparent about energy use and CO2e in our future research. To help reduce the carbon footprint of ML, we believe energy usage and CO2e should be a key metric in evaluating models, and we are collaborating with MLPerf developers to include energy usage during training and inference in this industry standard benchmark.
hub tools
citation-role summary
background 12
method 1
citation-polarity summary
claims ledger
- abstract The computation demand for machine learning (ML) has grown rapidly recently, which comes with a number of costs. Estimating the energy cost helps measure its environmental impact and finding greener strategies, yet it is challenging without detailed information. We calculate the energy use and carbon footprint of several recent large models-T5, Meena, GShard, Switch Transformer, and GPT-3-and refine earlier estimates for the neural architecture search that found Evolved Transformer. We highlight the following opportunities to improve energy efficiency and CO2 equivalent emissions (CO2e): Large
co-cited works
representative citing papers
Develops the Inference-Cost Phillips Curve linking AI inference costs to inflation dynamics, derives structural slopes and optimal monetary policy, and reports empirical estimates from US and G7 data that align with theoretical predictions.
QuBD extends algorithmic complexity estimation to quantized DNN weights, revealing that complexity decreases during learning, increases with overfitting, follows grokking patterns, and correlates with generalization.
Introduces the Amortized Efficiency Threshold (AET) to identify the deployment volume at which neural combinatorial optimization solvers achieve lower total energy use than heuristic baselines after accounting for training costs.
DUET improves RLVR by allocating tokens across both prompt selection and rollout length, outperforming full-budget baselines even when using only half the tokens.
TokenArena is a continuous benchmark for AI inference endpoints that measures output speed, time to first token, blended price, effective context, quality, and modeled energy to produce composites of joules per correct answer, dollars per correct answer, and endpoint fidelity.
SAT trains multi-LLM teams with sequential block updates to deliver monotonic gains and plug-and-play model swaps that provably improve performance bounds.
Single-electron and single-photon stochastic physical neural networks achieve over 97% MNIST test accuracy when trained with empirical outputs in the backward pass using few trials per layer.
In Kuramoto networks at equilibrium, weak nudging makes phase displacement the exact gradient of loss w.r.t. natural frequencies, enabling frequency learning that beats weight learning and resolves convergence via spectral initialization.
Short input phrases can irreversibly overwrite hidden states in Mamba models, impairing information retrieval on a new benchmark while leaving pure Transformer models unaffected.
DenseAMs show tradeoffs between entropy production, retrieval accuracy, and speed at intermediate loads, with a new failure mode in higher-order networks at finite temperature.
SAM 3 introduces promptable concept segmentation that doubles accuracy of prior systems on images and videos while improving standard SAM segmentation performance.
A promptable model trained on 1B masks achieves competitive zero-shot segmentation performance across tasks and is released publicly with its dataset.
MEMIT scales direct memory editing in transformers from single facts to thousands of associations by optimizing MLP weight updates.
OPT releases open decoder-only transformers up to 175B parameters that match GPT-3 performance at one-seventh the carbon cost, along with code and training logs.
Latent diffusion models achieve state-of-the-art inpainting and competitive results on unconditional generation, scene synthesis, and super-resolution by performing the diffusion process in the latent space of pretrained autoencoders with cross-attention conditioning, while cutting computational and
Multitask fine-tuning of an encoder-decoder model on prompted datasets produces zero-shot generalization that often beats models up to 16 times larger on standard benchmarks.
Proposes a textbook-based true/false QA task where PTLMs score ~50% closed-book even after pre-training on the text and ~60% open-book with retrieval.
Poisoning external knowledge bases with LLM-agent-crafted documents can increase RAG inference token consumption by up to 13.12 times at over 90% success rate while preserving answer quality.
DUET is a photonic tensor core paradigm that uses structural symmetry in VODICs to support arbitrary signed operands directly, experimentally tested on image classification, segmentation, and Transformer tasks.
AI data center waste heat upgraded by heat pumps can drive direct air capture to achieve net CO2 removal and offset operational emissions in several US states under current and 2030 scenarios.
Fixing the visual encoder in multilingual CLIP isolates text-branch deficits as the cause of lower visual grounding performance for low-resource languages, with model scaling widening some gaps but not others.
A memristor-array Hopfield network uses device nonlinearity to exceed classical memory capacity with K ~ 0.14N experimentally and superlinear K ~ 0.3 N^1.2 in simulations.
UniSD unifies self-distillation components for autoregressive LLMs and its full integrated version improves base models by 5.4 points and baselines by 2.8 points across six benchmarks.
citing papers explorer
-
Energy per Successful Goal: Goal-Level Energy Accounting for Agentic AI Systems
Proposes EpG and OOI metrics showing agentic workflows use 4.33x more energy per successful goal than linear baselines due to orchestration structure.
-
The Economics of AI Inference: Inflation Dynamics, Welfare Costs, and Optimal Monetary Policy under the Inference-Cost Phillips Curve
Develops the Inference-Cost Phillips Curve linking AI inference costs to inflation dynamics, derives structural slopes and optimal monetary policy, and reports empirical estimates from US and G7 data that align with theoretical predictions.
-
Characterizing Learning in Deep Neural Networks using Tractable Algorithmic Complexity Analysis
QuBD extends algorithmic complexity estimation to quantized DNN weights, revealing that complexity decreases during learning, increases with overfitting, follows grokking patterns, and correlates with generalization.
-
An Amortized Efficiency Threshold for Comparing Neural and Heuristic Solvers in Combinatorial Optimization
Introduces the Amortized Efficiency Threshold (AET) to identify the deployment volume at which neural combinatorial optimization solvers achieve lower total energy use than heuristic baselines after accounting for training costs.
-
DUET: Optimize Token-Budget Allocation for Reinforcement Learning with Verifiable Rewards
DUET improves RLVR by allocating tokens across both prompt selection and rollout length, outperforming full-budget baselines even when using only half the tokens.
-
Token Arena: A Continuous Benchmark Unifying Energy and Cognition in AI Inference
TokenArena is a continuous benchmark for AI inference endpoints that measures output speed, time to first token, blended price, effective context, quality, and modeled energy to produce composites of joules per correct answer, dollars per correct answer, and endpoint fidelity.
-
SAT: Sequential Agent Tuning for Coordinator Free Plug and Play Multi-LLM Training with Monotonic Improvement Guarantees
SAT trains multi-LLM teams with sequential block updates to deliver monotonic gains and plug-and-play model swaps that provably improve performance bounds.
-
Training single-electron and single-photon stochastic physical neural networks
Single-electron and single-photon stochastic physical neural networks achieve over 97% MNIST test accuracy when trained with empirical outputs in the backward pass using few trials per layer.
-
The Phase Is the Gradient: Equilibrium Propagation for Frequency Learning in Kuramoto Networks
In Kuramoto networks at equilibrium, weak nudging makes phase displacement the exact gradient of loss w.r.t. natural frequencies, enabling frequency learning that beats weight learning and resolves convergence via spectral initialization.
-
Hidden State Poisoning Attacks against Mamba-based Language Models
Short input phrases can irreversibly overwrite hidden states in Mamba models, impairing information retrieval on a new benchmark while leaving pure Transformer models unaffected.
-
Stochastic Thermodynamics of Associative Memory
DenseAMs show tradeoffs between entropy production, retrieval accuracy, and speed at intermediate loads, with a new failure mode in higher-order networks at finite temperature.
-
SAM 3: Segment Anything with Concepts
SAM 3 introduces promptable concept segmentation that doubles accuracy of prior systems on images and videos while improving standard SAM segmentation performance.
-
Segment Anything
A promptable model trained on 1B masks achieves competitive zero-shot segmentation performance across tasks and is released publicly with its dataset.
-
Mass-Editing Memory in a Transformer
MEMIT scales direct memory editing in transformers from single facts to thousands of associations by optimizing MLP weight updates.
-
OPT: Open Pre-trained Transformer Language Models
OPT releases open decoder-only transformers up to 175B parameters that match GPT-3 performance at one-seventh the carbon cost, along with code and training logs.
-
High-Resolution Image Synthesis with Latent Diffusion Models
Latent diffusion models achieve state-of-the-art inpainting and competitive results on unconditional generation, scene synthesis, and super-resolution by performing the diffusion process in the latent space of pretrained autoencoders with cross-attention conditioning, while cutting computational and
-
Multitask Prompted Training Enables Zero-Shot Task Generalization
Multitask fine-tuning of an encoder-decoder model on prompted datasets produces zero-shot generalization that often beats models up to 16 times larger on standard benchmarks.
-
Perhaps PTLMs Should Go to School -- A Task to Assess Open Book and Closed Book QA
Proposes a textbook-based true/false QA task where PTLMs score ~50% closed-book even after pre-training on the text and ~60% open-book with retrieval.
-
Inference Cost Attacks for Retrieval-Augmented Large Language Models
Poisoning external knowledge bases with LLM-agent-crafted documents can increase RAG inference token consumption by up to 13.12 times at over 90% success rate while preserving answer quality.
-
General-Purpose Photonic Computing Primitive for Contemporary Artificial Intelligence
DUET is a photonic tensor core paradigm that uses structural symmetry in VODICs to support arbitrary signed operands directly, experimentally tested on image classification, segmentation, and Transformer tasks.
-
Recasting AI Data Centers as Engines for Carbon Removal
AI data center waste heat upgraded by heat pumps can drive direct air capture to achieve net CO2 removal and offset operational emissions in several US states under current and 2030 scenarios.
-
Language-Conditioned Visual Grounding with CLIP Multilingual
Fixing the visual encoder in multilingual CLIP isolates text-branch deficits as the cause of lower visual grounding performance for low-resource languages, with model scaling widening some gaps but not others.
-
A Hardware-aware Hopfield Network with a Nonlinear Memristor Array for Robust Associative Memory with Superlinear Capacity
A memristor-array Hopfield network uses device nonlinearity to exceed classical memory capacity with K ~ 0.14N experimentally and superlinear K ~ 0.3 N^1.2 in simulations.
-
UniSD: Towards a Unified Self-Distillation Framework for Large Language Models
UniSD unifies self-distillation components for autoregressive LLMs and its full integrated version improves base models by 5.4 points and baselines by 2.8 points across six benchmarks.
-
OpenG2G: A Simulation Platform for AI Datacenter-Grid Runtime Coordination
OpenG2G is a new extensible simulation platform that lets users implement and compare classic, optimization, and learning-based controllers for AI datacenter power flexibility coordinated with the grid.
-
A Meta Reinforcement Learning Approach to Goals-Based Wealth Management
MetaRL pre-trained on GBWM problems delivers near-optimal dynamic strategies in 0.01s achieving 97.8% of DP optimal utility and handles larger problems where DP fails.
-
Are Large Language Models Economically Viable for Industry Deployment?
Small LLMs under 2B parameters achieve better economic break-even, energy efficiency, and hardware density than larger models on legacy GPUs for industrial tasks.
-
TRON: Trainable, architecture-reconfigurable random optical neural networks
TRON demonstrates a trainable and reconfigurable optical neural network that combines multi-scattering media with DMD-based matrix multiplication and performs in-situ optimization plus neural architecture search on the optical hardware itself.
-
Watt Counts: Energy-Aware Benchmark for Sustainable LLM Inference on Heterogeneous GPU Architectures
Watt Counts supplies over 5,000 energy measurements across 50 LLMs and 10 GPUs and shows that hardware-aware selection can reduce server-scenario energy use by up to 70 percent with little effect on user experience.
-
Minimum-Action Learning: Energy-Constrained Symbolic Model Selection for Physical Law Identification from Noisy Data
MAL recovers correct symbolic force laws like Kepler gravity from noisy data by minimizing trajectory reconstruction, sparsity, and energy violation, reaching 100% identification via energy criterion on benchmarks.
-
When Should Users Check? Modeling Confirmation Frequency inMulti-Step Agentic AI Tasks
A decision-theoretic model based on the observed Confirmation-Diagnosis-Correction-Redo user pattern places intermediate confirmations in AI agent tasks, yielding 81% user preference and 13.54% faster completion versus confirm-at-end.
-
Optimizing LLM Inference: Fluid-Guided Online Scheduling with Memory Constraints
The paper develops fluid-guided online scheduling algorithms (WAIT and Nested WAIT) for LLM inference that handle endogenous KV-cache memory growth and improve stability and latency over baselines in simulations.
-
AI Failures in the Eyes of the Downstream Developer: A First Look at Concerns, Practices, and Challenges
Mixed-methods study maps downstream developers' concerns, practices, and challenges with AI failures in PTM-based software.
-
SAM 2: Segment Anything in Images and Videos
SAM 2 delivers more accurate video segmentation with 3x fewer user interactions and 6x faster image segmentation than the original SAM by training a streaming-memory transformer on the largest video segmentation dataset collected to date.
-
BLOOM: A 176B-Parameter Open-Access Multilingual Language Model
BLOOM is a 176B-parameter open-access multilingual language model trained on the ROOTS corpus that achieves competitive performance on benchmarks, with improved results after multitask prompted finetuning.
-
PaLM: Scaling Language Modeling with Pathways
PaLM 540B demonstrates continued scaling benefits by setting new few-shot SOTA results on hundreds of benchmarks and outperforming humans on BIG-bench.
-
ST-MoE: Designing Stable and Transferable Sparse Expert Models
ST-MoE introduces stability techniques for sparse expert models, allowing a 269B-parameter model to achieve state-of-the-art transfer learning results across reasoning, summarization, and QA tasks at the compute cost of a 32B dense model.
-
LaMDA: Language Models for Dialog Applications
LaMDA shows that fine-tuning on human-value annotations and consulting external knowledge sources significantly improves safety and factual grounding in large dialog models beyond what scaling alone achieves.
-
Ethical and social risks of harm from Language Models
The authors provide a detailed taxonomy of 21 risks associated with language models, covering discrimination, information leaks, misinformation, malicious applications, interaction harms, and societal impacts like job loss and environmental costs.
-
Deduplicating Training Data Makes Language Models Better
Deduplicating training datasets reduces language model verbatim memorization by 10x, improves training efficiency, and enables more accurate evaluation by cutting train-test overlap.
-
Optimized Federated Knowledge Distillation with Distributed Neural Architecture Search
FedKDNAS combines client-side neural architecture search with knowledge distillation from aggregated server predictions to improve accuracy and efficiency in heterogeneous federated learning.
-
GraphRAG on Consumer Hardware: Benchmarking Local LLMs for Healthcare EHR Schema Retrieval
GraphRAG with 7-8B local LLMs on 8GB VRAM hardware builds knowledge graphs from EHR docs and answers queries, with Llama 3.1 creating the largest graph, Qwen 2.5 scoring highest on quality, and models below ~7B failing to complete the pipeline.
-
The Thermodynamic Costs of Simple Linear Regression
Thermodynamic lower bounds are approximated for exact and SGD linear regression, producing energy-aware scaling laws for optimal training dataset size given a target generalization error.
-
Position: LLM Inference Should Be Evaluated as Energy-to-Token Production
LLM inference should be reframed and evaluated as energy-to-token production with a Token Production Function that accounts for power, cooling, and efficiency ceilings.
-
ELAS: Efficient Pre-Training of Low-Rank Large Language Models via 2:4 Activation Sparsity
ELAS pre-trains low-rank LLMs by applying 2:4 activation sparsity after squared ReLU to cut memory and accelerate training with minimal performance loss.
-
Toward a Sustainable Software Architecture Community: Evaluating ICSA's Environmental Impact
The study provides exploratory estimates of carbon emissions from GenAI inference in ICSA papers and from the full operations of the ICSA 2025 conference.
-
Sustainability Is Not Linear: Quantifying Performance, Energy, and Privacy Trade-offs in On-Device Intelligence
Empirical case study on a flagship Android device profiles energy, latency, and quality trade-offs across eight LLMs, revealing a quantization energy paradox and identifying mid-sized models as practical sweet spots.
-
Neuro-Symbolic AI for Cybersecurity: State of the Art, Challenges, and Opportunities
A systematic review of neuro-symbolic AI in cybersecurity finds that deeper integration and causal reasoning improve performance across intrusion detection and vulnerability tasks, while identifying barriers and a research roadmap.
-
RAP: Runtime Adaptive Pruning for LLM Inference
RAP is a reinforcement learning framework for runtime-adaptive pruning of LLMs that jointly optimizes model weights and KV-cache usage under varying memory budgets.
-
DADA: Dual Averaging with Distance Adaptation
DADA is a parameter-free dual averaging method for convex optimization that adapts to local function growth and applies to nonsmooth, smooth, Holder-smooth, and other classes for both constrained and unbounded domains without prior knowledge of iteration count or accuracy.