Noisy expert imitation learning requires exponential samples for offline methods but polynomial for a variant of on-policy distillation under a noise condition.
super hub Mixed citations
DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter
Mixed citation behavior. Most common role is background (62%).
216
Pith papers citing it
Background
62% of classified citations
abstract
As Transfer Learning from large-scale pre-trained models becomes more prevalent in Natural Language Processing (NLP), operating these large models in on-the-edge and/or under constrained computational training or inference budgets remains challenging. In this work, we propose a method to pre-train a smaller general-purpose language representation model, called DistilBERT, which can then be fine-tuned with good performances on a wide range of tasks like its larger counterparts. While most prior work investigated the use of distillation for building task-specific models, we leverage knowledge distillation during the pre-training phase and show that it is possible to reduce the size of a BERT model by 40%, while retaining 97% of its language understanding capabilities and being 60% faster. To leverage the inductive biases learned by larger models during pre-training, we introduce a triple loss combining language modeling, distillation and cosine-distance losses. Our smaller, faster and lighter model is cheaper to pre-train and we demonstrate its capabilities for on-device computations in a proof-of-concept experiment and a comparative on-device study.
hub tools
citation-role summary
background 18
method 11
citation-polarity summary
claims ledger
- abstract As Transfer Learning from large-scale pre-trained models becomes more prevalent in Natural Language Processing (NLP), operating these large models in on-the-edge and/or under constrained computational training or inference budgets remains challenging. In this work, we propose a method to pre-train a smaller general-purpose language representation model, called DistilBERT, which can then be fine-tuned with good performances on a wide range of tasks like its larger counterparts. While most prior work investigated the use of distillation for building task-specific models, we leverage knowledge di
authors
co-cited works
representative citing papers
Memory augmentation in LLMs amplifies sycophancy up to 25x compared to in-context baselines due to lossy memory extraction, with two lightweight mitigations that reduce the effect while preserving recall.
Derives geodesic ridge regularization and Riemannian Gibbs Process prior for feature-learning wide neural networks, generalizing kernel-regime results via function-space axiomatization.
Adaptive scheduling of interventions in discrete diffusion language models, timed to attribute-specific commitment schedules discovered with sparse autoencoders, delivers precise multi-attribute steering up to 93% strength while preserving generation quality.
A classifier trained only on transformer fine-tuning data detects an invariant memorization signature that transfers to Mamba, RWKV-4, and RecurrentGemma with AUCs of 0.963, 0.972, and 0.936.
Tiny language models under 10M parameters trained on a synthetic children's story dataset generate fluent, consistent, multi-paragraph English text with near-perfect grammar and reasoning.
GPT-3 shows that scaling an autoregressive language model to 175 billion parameters enables strong few-shot performance across diverse NLP tasks via in-context prompting without fine-tuning.
MetaCLIP-CMR applies CLIP-style contrastive learning to cardiac MRI by treating acquisition metadata as text labels, delivering 86.8% modality and 86.5% view accuracy plus top Dice scores on ACDC/M&Ms segmentation with far less pre-training data than recent large-scale CMR models.
ZPPO improves distillation to small vision-language models by using binary and negative candidate prompts plus a replay buffer for hard questions, outperforming standard distillation and GRPO on a 31-benchmark suite with largest gains at the 0.8B scale.
Otters++ realizes TTFS via measured device decay in optical synapses, uses hybrid QNN-equivalent training with noise awareness, and reports 84.17% average GLUE score with energy gains over prior spiking transformers.
AR-OPD disentangles privileged supervision via anchored residual guidance to reduce hindsight leakage in on-policy distillation, reporting gains of 2.3 points over full privileged OPD and 7.9 over SFT on reasoning tasks.
Introduces a pseudo-metric to quantify advice usefulness and shows reliable advice enables efficient approximate Stackelberg strategies while unreliable advice blocks simultaneous near-Stackelberg and no-regret guarantees but permits weak dominance in some correlated equilibria.
OPRD performs distillation in hidden-state space on on-policy data for deterministic gradients and better math benchmark performance, plus OPRD-Bridge for cross-architecture transfer via low-rank projectors.
Face-Feature Tuning is a label-free logit remapping method that reduces FPR/TPR gaps across groups in deepfake detection while preserving overall accuracy.
Introduces (ε,q,t,A)-behavioral indistinguishability and shows via Qwen/Llama experiments that LoRA distillation boosts semantic similarity but leaves detectable behavioral differences under adversarial evaluation.
MATCHA introduces a dual-view contrastive metric measuring proximity to gold text and distance from adversarial contradictions, outperforming ROUGE and BERTScore by up to 20% on TruthfulQA and other NLP benchmarks.
PHAT-JeT combines geometric message-passing with hierarchical patch attention to reach state-of-the-art accuracy and background rejection among resource-constrained jet tagging models on four benchmarks.
CROC constructs finite-sample valid confidence sets for the root-cause index in multi-stream change detection using conformal p-values under independence and exchangeability assumptions.
AIGaitor is the first claimed end-to-end on-device monocular motion-capture and deep-learning gait analysis pipeline demonstrated on consumer smartphones.
Layer-wise Token Compression applies adaptive token pooling at middle transformer layers for cross-encoder rerankers, preserving MS MARCO ranking quality while raising QPS up to 25% on passages and 116% on documents, with added gains on listwise LLM rerankers and a regularizer effect for long inputs
TIDAL recovers temporal phase signals from LLM-derived semantics of provisioning metadata to enable complementary CVD placement, reducing overload frequency by 79.1% on production traces.
RISE is an inference-time semantic reranking framework that refines low-confidence predictions in rhetorical role labeling using contrastively learned label representations, delivering an average +9.15 macro-F1 gain on hard examples across eight datasets and seven models.
DMP-MH clips degrees to control triangle sensitivity, synthesizes an edge-DP graph with Noisy Mirror Descent, and distills it into dual-stream hash networks, beating private baselines by up to 11.4 mAP on MIRFlickr-25K and NUS-WIDE while keeping 92.5% of non-private performance.
Foundation model priors amplify worst-client disparity under extreme federated heterogeneity, creating a fairness paradox where larger models perform worse for disadvantaged clients.
citing papers explorer
-
Behavior Cloning is Not All You Need: The Optimality of On-Policy Distillation for Noisy Expert Feedback
Noisy expert imitation learning requires exponential samples for offline methods but polynomial for a variant of on-policy distillation under a noise condition.
-
Recalling Too Well: Sycophancy Evaluation and Mitigation in Memory-Augmented Models
Memory augmentation in LLMs amplifies sycophancy up to 25x compared to in-context baselines due to lossy memory extraction, with two lightweight mitigations that reduce the effect while preserving recall.
-
Canonical Regularisation of Wide Feature-Learning Neural Networks
Derives geodesic ridge regularization and Riemannian Gibbs Process prior for feature-learning wide neural networks, generalizing kernel-regime results via function-space axiomatization.
-
Steering Without Breaking: Mechanistically Informed Interventions for Discrete Diffusion Language Models
Adaptive scheduling of interventions in discrete diffusion language models, timed to attribute-specific commitment schedules discovered with sparse autoencoders, delivers precise multi-attribute steering up to 93% strength while preserving generation quality.
-
Learning the Signature of Memorization in Autoregressive Language Models
A classifier trained only on transformer fine-tuning data detects an invariant memorization signature that transfers to Mamba, RWKV-4, and RecurrentGemma with AUCs of 0.963, 0.972, and 0.936.
-
TinyStories: How Small Can Language Models Be and Still Speak Coherent English?
Tiny language models under 10M parameters trained on a synthetic children's story dataset generate fluent, consistent, multi-paragraph English text with near-perfect grammar and reasoning.
-
Language Models are Few-Shot Learners
GPT-3 shows that scaling an autoregressive language model to 175 billion parameters enables strong few-shot performance across diverse NLP tasks via in-context prompting without fine-tuning.
-
Learning from Acquisition: Metadata-driven Multimodal Pre-training for Cardiac MRI
MetaCLIP-CMR applies CLIP-style contrastive learning to cardiac MRI by treating acquisition metadata as text labels, delivering 86.8% modality and 86.5% view accuracy plus top Dice scores on ACDC/M&Ms segmentation with far less pre-training data than recent large-scale CMR models.
-
Zone of Proximal Policy Optimization: Teacher in Prompts, Not Gradients
ZPPO improves distillation to small vision-language models by using binary and negative candidate prompts plus a replay buffer for hard questions, outperforming standard distillation and GRPO on a 31-benchmark suite with largest gains at the 0.8B scale.
-
Otters++: A Time-to-first-spike Based Energy Efficient Optical Spiking Transformer
Otters++ realizes TTFS via measured device decay in optical synapses, uses hybrid QNN-equivalent training with noise awareness, and reports 84.17% average GLUE score with energy gains over prior spiking transformers.
-
Beyond Absolute Imitation: Anchored Residual Guidance for Privileged On-Policy Distillation
AR-OPD disentangles privileged supervision via anchored residual guidance to reduce hindsight leakage in on-policy distillation, reporting gains of 2.3 points over full privileged OPD and 7.9 over SFT on reasoning tasks.
-
Leveraging Machine-Learned Advice in Strategic Interactions with No-Regret Learners
Introduces a pseudo-metric to quantify advice usefulness and shows reliable advice enables efficient approximate Stackelberg strategies while unreliable advice blocks simultaneous near-Stackelberg and no-regret guarantees but permits weak dominance in some correlated equilibria.
-
OPRD: On-Policy Representation Distillation
OPRD performs distillation in hidden-state space on on-policy data for deterministic gradients and better math benchmark performance, plus OPRD-Bridge for cross-architecture transfer via low-rank projectors.
-
Toward Calibrated, Fair, and accurate Deepfake Detection
Face-Feature Tuning is a label-free logit remapping method that reduces FPR/TPR gaps across groups in deepfake detection while preserving overall accuracy.
-
Bounded Behavioral Indistinguishability for Black-Box LLM Distillation
Introduces (ε,q,t,A)-behavioral indistinguishability and shows via Qwen/Llama experiments that LoRA distillation boosts semantic similarity but leaves detectable behavioral differences under adversarial evaluation.
-
MATCHA: Matching Text via Contrastive Semantic Alignment
MATCHA introduces a dual-view contrastive metric measuring proximity to gold text and distance from adversarial contradictions, outperforming ROUGE and BERTScore by up to 20% on TruthfulQA and other NLP benchmarks.
-
Patch Hierarchical Attention Transformer for Efficient Particle Jet Tagging
PHAT-JeT combines geometric message-passing with hierarchical patch attention to reach state-of-the-art accuracy and background rejection among resource-constrained jet tagging models on four benchmarks.
-
Distribution-free root cause analysis
CROC constructs finite-sample valid confidence sets for the root-cause index in multi-stream change detection using conformal p-values under independence and exchangeability assumptions.
-
AIGaitor: Privacy-preserving and cloud-free motion analysis for everyone, using edge computing
AIGaitor is the first claimed end-to-end on-device monocular motion-capture and deep-learning gait analysis pipeline demonstrated on consumer smartphones.
-
Layer-wise Token Compression for Efficient Document Reranking
Layer-wise Token Compression applies adaptive token pooling at middle transformer layers for cross-encoder rerankers, preserving MS MARCO ranking quality while raising QPS up to 25% on passages and 116% on documents, with added gains on listwise LLM rerankers and a regularizer effect for long inputs
-
TIDAL: Recovering Temporal Phase for Cloud Block Storage Placement from LLM-Derived Semantics
TIDAL recovers temporal phase signals from LLM-derived semantics of provisioning metadata to enable complementary CVD placement, reducing overload frequency by 79.1% on production traces.
-
Semantic Reranking at Inference Time for Hard Examples in Rhetorical Role Labeling
RISE is an inference-time semantic reranking framework that refines low-confidence predictions in rhetorical role labeling using contrastively learned label representations, delivering an average +9.15 macro-F1 gain on hard examples across eight datasets and seven models.
-
Differentially Private Motif-Preserving Multi-modal Hashing
DMP-MH clips degrees to control triangle sensitivity, synthesizes an edge-DP graph with Noisy Mirror Descent, and distills it into dual-stream hash networks, beating private baselines by up to 11.4 mAP on MIRFlickr-25K and NUS-WIDE while keeping 92.5% of non-private performance.
-
When More Parameters Hurt: Foundation Model Priors Amplify Worst-Client Disparity Under Extreme Federated Heterogeneity
Foundation model priors amplify worst-client disparity under extreme federated heterogeneity, creating a fairness paradox where larger models perform worse for disadvantaged clients.
-
Switchcraft: AI Model Router for Agentic Tool Calling
Switchcraft routes agentic tool-calling queries to the lowest-cost model that preserves correctness, reaching 82.9% accuracy and 84% cost reduction on five benchmarks.
-
TRACE: Transport Alignment Conformal Prediction via Diffusion and Flow Matching Models
TRACE creates valid conformal prediction sets for complex generative models by scoring outputs via averaged denoising or velocity errors along stochastic transport paths instead of likelihoods.
-
A Multi-View Media Profiling Suite: Resources, Evaluation, and Analysis
Presents MBFC-2025 dataset and multi-view embeddings with fusion methods for media bias and factuality, reporting SOTA results on ACL-2020 and new benchmarks on MBFC-2025.
-
VOW: Verifiable and Oblivious Watermark Detection for Large Language Models
VOW formulates LLM watermark detection as a secure two-party computation using a Verifiable Oblivious Pseudorandom Function to achieve private and cryptographically verifiable detection.
-
Homogeneous Stellar Parameters from Heterogeneous Spectra with Deep Learning
A single end-to-end Transformer model unifies stellar labels from heterogeneous spectroscopic surveys into a self-consistent scale without post-hoc recalibration.
-
AgentPulse: A Continuous Multi-Signal Framework for Evaluating AI Agents in Deployment
AgentPulse is a continuous multi-signal framework that scores AI agents on benchmark performance, adoption, sentiment and ecosystem health, showing these factors are complementary and that benchmark-plus-sentiment predicts external adoption metrics.
-
RoLegalGEC: Legal Domain Grammatical Error Detection and Correction Dataset for Romanian
RoLegalGEC is the first Romanian legal-domain dataset for grammatical error detection and correction, consisting of 350,000 examples, with evaluations of several neural models.
-
GuardPhish: Securing Open-Source LLMs from Phishing Abuse
Open-source LLMs detect phishing intent at high rates but still generate actionable phishing content, and GuardPhish supplies a dataset plus modular classifiers to close the gap.
-
Depth Adaptive Efficient Visual Autoregressive Modeling
DepthVAR adaptively allocates per-token computational depth in VAR models using a cyclic rotated scheduler and dynamic layer masking to achieve 2.3-3.1x inference speedup with minimal quality loss.
-
SecureRouter: Encrypted Routing for Efficient Secure Inference
SecureRouter accelerates secure transformer inference by 1.95x via an encrypted router that selects input-adaptive models from an MPC-optimized pool with negligible accuracy loss.
-
Multilingual Multi-Label Emotion Classification at Scale with Synthetic Data
Synthetic data of 1M+ multi-label samples across 23 languages trains models that match or exceed English-only specialists on zero-shot benchmarks for emotion classification.
-
Demystifying OPD: Length Inflation and Stabilization Strategies for Large Language Models
OPD for LLMs suffers length inflation and repetition collapse; StableOPD uses reference divergence and rollout mixing to prevent it and improve math reasoning performance by 7.2% on average.
-
Kathleen: Oscillator-Based Byte-Level Text Classification Without Tokenization or Attention
Kathleen performs byte-level text classification via recurrent oscillator banks, FFT wavetable encoding, and phase harmonics, matching pretrained baselines on standard benchmarks with 36% fewer parameters.
-
A Paradigm Shift: Fully End-to-End Training for Temporal Sentence Grounding in Videos
Fully end-to-end training with a sentence-conditioned adapter outperforms frozen-backbone baselines for localizing video segments that match sentence queries.
-
Explainable Semantic Textual Similarity via Dissimilar Span Detection
Introduces the Dissimilar Span Detection task and Span Similarity Dataset to explain semantic textual similarity by identifying differing spans between text pairs.
-
Self-Distilled Reasoner: On-Policy Self-Distillation for Large Language Models
A single LLM improves its own reasoning by self-distilling from privileged verified traces as teacher to its question-only student policy, outperforming off-policy distillation and RL on math benchmarks with better token efficiency.
-
DualGuard: Dual-stream Large Language Model Watermarking Defense against Paraphrase and Spoofing Attack
DualGuard uses adaptive dual-stream watermark signals to detect and trace both paraphrase and spoofing attacks in LLM outputs while preserving text quality.
-
Language-Conditioned Safe Trajectory Generation for Spacecraft Rendezvous
SAGES translates natural-language commands into constraint-respecting spacecraft trajectories, achieving over 90% semantic-behavioral consistency in proximity operations and robotic tests.
-
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.
-
Task complexity shapes internal representations and robustness in neural networks
Harder classification tasks produce neural representations whose accuracy collapses under binarization and shuffling while easier tasks remain robust, defining task complexity via the performance gap between full-precision and perturbed networks.
-
A Woman with a Knife or A Knife with a Woman? Measuring Directional Bias Amplification in Image Captions
DBAC is a new directional metric for bias amplification in image captions that is less sensitive to sentence encoders and more accurate than LIC, validated on COCO gender and race attributes.
-
Post-detection inference for sequential changepoint localization
Develops a general nonparametric framework for constructing non-asymptotically valid confidence sets for changepoint location using data up to an arbitrary detection stopping time.
-
Neurons Speak in Ranges: Breaking Free from Discrete Neuronal Attribution
Neurons exhibit concept-conditioned activation ranges forming Gaussian-like distributions with minimal overlap, and range-based interventions via NeuronLens outperform neuron-level masking in targeted manipulation with reduced collateral effects.
-
Eliciting Latent Predictions from Transformers with the Tuned Lens
Training per-layer affine probes on frozen transformers yields more reliable latent predictions than the logit lens and enables detection of malicious inputs from prediction trajectories.
-
Accelerating Large Language Model Decoding with Speculative Sampling
Speculative sampling accelerates LLM decoding 2-2.5x by letting a draft model propose short sequences that the target model scores in parallel, then applies modified rejection sampling to keep the exact target distribution.
-
Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer
T5 casts all NLP tasks as text-to-text generation, systematically explores pre-training choices, and reaches strong performance on summarization, QA, classification and other tasks via large-scale training on the Colossal Clean Crawled Corpus.