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.
hub
The LAMBADA dataset: Word prediction requiring a broad discourse context
35 Pith papers cite this work. Polarity classification is still indexing.
35
Pith papers citing it
abstract
We introduce LAMBADA, a dataset to evaluate the capabilities of computational models for text understanding by means of a word prediction task. LAMBADA is a collection of narrative passages sharing the characteristic that human subjects are able to guess their last word if they are exposed to the whole passage, but not if they only see the last sentence preceding the target word. To succeed on LAMBADA, computational models cannot simply rely on local context, but must be able to keep track of information in the broader discourse. We show that LAMBADA exemplifies a wide range of linguistic phenomena, and that none of several state-of-the-art language models reaches accuracy above 1% on this novel benchmark. We thus propose LAMBADA as a challenging test set, meant to encourage the development of new models capable of genuine understanding of broad context in natural language text.
hub tools
citation-role summary
background 1
citation-polarity summary
roles
background 1polarities
background 1representative citing papers
Pythia releases 16 identically trained LLMs with full checkpoints and data tools to study training dynamics, scaling, memorization, and bias in language models.
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.
Uniform diffusion models rely on a leave-one-out denoiser rather than the usual denoising posterior, with exact conversions derived; an absorbing-state reformulation is introduced that matches or exceeds masked diffusion on language modeling while preserving the original joint distribution.
Chronicle is the first model jointly pretrained from scratch on text and time series in a unified transformer that matches a comparable language model on NLU tasks and sets new bars for time series classification and multimodal forecasting.
Presents a likelihood-based benchmark for equation-suffix prediction in technical papers with controls to detect shortcut vulnerabilities in model forecasts.
A survey of LLM copyright protection that unifies text watermarking, model watermarking, and model fingerprinting while presenting new coverage of fingerprint transfer and removal.
PRIMETIME generator reveals that LLM datetime parsing and arithmetic primitives are individually unreliable but fully learnable via fine-tuning, enabling frontier-level accuracy on event planning with small LoRA models.
MIXAR is the first autoregressive pixel-based language model for eight languages and scripts, with empirical gains on multilingual tasks, robustness to unseen languages, and further improvements when scaled to 0.5B parameters.
GPTQ quantizes 175B-parameter GPT models to 3-4 bits per weight in one shot using approximate second-order information, achieving negligible accuracy degradation and 3-4x inference speedups.
Intra-layer model parallelism in PyTorch enables training of 8.3B-parameter transformers, achieving SOTA perplexity of 10.8 on WikiText103 and 66.5% accuracy on LAMBADA.
Oryx hybridizes attention and linear recurrent mixers along the sequence axis with high parameter sharing, outperforming single-mixer baselines on language modeling and retrieval at up to 1.4B scale under mixed training.
Empirical study shows mixture pretraining tolerates higher target data repetition than single-source training, with a new repetition-aware scaling law enabling principled mixture selection based on data size, compute, and model scale.
KVM is a new block-recurrent compressed KV attention that turns transformers into O(N) chunked RNNs or growable sublinear-memory models while remaining implementable with standard operations.
A conditional scaling law fitted on over 200 models from 80M to 3B parameters identifies architectures that deliver up to 2.1% higher accuracy and 42% higher inference throughput than LLaMA-3.2 under the same training budget.
Attention-Shifting uses importance-aware suppression on unlearning data and retention enhancement on retained data via dual-loss optimization to achieve selective unlearning with better utility preservation than prior methods.
Gemini Ultra reaches human-expert performance on MMLU for the first time and sets new state-of-the-art results on 30 of 32 benchmarks, including all 20 multimodal ones tested.
Gated linear attention Transformers achieve competitive language modeling results with linear-time inference, superior length generalization, and higher training throughput than Mamba.
AGIEval shows GPT-4 exceeding average human scores on SAT Math at 95% and Chinese college entrance English at 92.5%, while revealing weaker results on complex reasoning tasks.
MoHGE achieves standard MoE performance with 20% fewer parameters and balanced GPU utilization via grouped heterogeneous experts, two-level routing, and specialized auxiliary losses.
Strix delivers sub-microsecond fault localisation, detection, and correction on NPUs with 1.04x slowdown and minimal hardware cost by system-level re-partitioning and targeted safeguards.
A two-stage distillation recipe converts a Pythia-1B Transformer into a Mamba model that preserves performance with perplexity 14.11 versus the teacher's 13.86.
DeepSpeed-Ulysses keeps communication volume constant for sequence-parallel attention when sequence length and device count scale together, delivering 2.5x faster training on 4x longer sequences than prior SOTA.
PaLM 540B demonstrates continued scaling benefits by setting new few-shot SOTA results on hundreds of benchmarks and outperforming humans on BIG-bench.
citing papers explorer
-
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.
-
Pythia: A Suite for Analyzing Large Language Models Across Training and Scaling
Pythia releases 16 identically trained LLMs with full checkpoints and data tools to study training dynamics, scaling, memorization, and bias in language models.
-
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.
-
Uniform Diffusion Models Revisited: Leave-One-Out Denoiser and Absorbing State Reformulation
Uniform diffusion models rely on a leave-one-out denoiser rather than the usual denoising posterior, with exact conversions derived; an absorbing-state reformulation is introduced that matches or exceeds masked diffusion on language modeling while preserving the original joint distribution.
-
Chronicle: A Multimodal Foundation Model for Joint Language and Time Series Understanding
Chronicle is the first model jointly pretrained from scratch on text and time series in a unified transformer that matches a comparable language model on NLU tasks and sets new bars for time series classification and multimodal forecasting.
-
Likelihood scoring for continuations of mathematical text: a self-supervised benchmark with tests for shortcut vulnerabilities
Presents a likelihood-based benchmark for equation-suffix prediction in technical papers with controls to detect shortcut vulnerabilities in model forecasts.
-
Copyright Protection for Large Language Models: A Survey of Methods, Challenges, and Trends
A survey of LLM copyright protection that unifies text watermarking, model watermarking, and model fingerprinting while presenting new coverage of fingerprint transfer and removal.
-
PRIMETIME : Limits of LLMs in Temporal Primitives
PRIMETIME generator reveals that LLM datetime parsing and arithmetic primitives are individually unreliable but fully learnable via fine-tuning, enabling frontier-level accuracy on event planning with small LoRA models.
-
MIXAR: Scaling Autoregressive Pixel-based Language Models to Multiple Languages and Scripts
MIXAR is the first autoregressive pixel-based language model for eight languages and scripts, with empirical gains on multilingual tasks, robustness to unseen languages, and further improvements when scaled to 0.5B parameters.
-
GPTQ: Accurate Post-Training Quantization for Generative Pre-trained Transformers
GPTQ quantizes 175B-parameter GPT models to 3-4 bits per weight in one shot using approximate second-order information, achieving negligible accuracy degradation and 3-4x inference speedups.
-
Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism
Intra-layer model parallelism in PyTorch enables training of 8.3B-parameter transformers, achieving SOTA perplexity of 10.8 on WikiText103 and 66.5% accuracy on LAMBADA.
-
Multi-Mixer Models: Flexible Sequence Modeling with Shared Representations
Oryx hybridizes attention and linear recurrent mixers along the sequence axis with high parameter sharing, outperforming single-mixer baselines on language modeling and retrieval at up to 1.4B scale under mixed training.
-
Scaling Laws for Mixture Pretraining Under Data Constraints
Empirical study shows mixture pretraining tolerates higher target data repetition than single-source training, with a new repetition-aware scaling law enabling principled mixture selection based on data size, compute, and model scale.
-
Key-Value Means: Transformers with Expandable Block-Recurrent Compressed Memory
KVM is a new block-recurrent compressed KV attention that turns transformers into O(N) chunked RNNs or growable sublinear-memory models while remaining implementable with standard operations.
-
Scaling Laws Meet Model Architecture: Toward Inference-Efficient LLMs
A conditional scaling law fitted on over 200 models from 80M to 3B parameters identifies architectures that deliver up to 2.1% higher accuracy and 42% higher inference throughput than LLaMA-3.2 under the same training budget.
-
Wisdom is Knowing What not to Say: Hallucination-Free LLMs Unlearning via Attention Shifting
Attention-Shifting uses importance-aware suppression on unlearning data and retention enhancement on retained data via dual-loss optimization to achieve selective unlearning with better utility preservation than prior methods.
-
Gemini: A Family of Highly Capable Multimodal Models
Gemini Ultra reaches human-expert performance on MMLU for the first time and sets new state-of-the-art results on 30 of 32 benchmarks, including all 20 multimodal ones tested.
-
Gated Linear Attention Transformers with Hardware-Efficient Training
Gated linear attention Transformers achieve competitive language modeling results with linear-time inference, superior length generalization, and higher training throughput than Mamba.
-
AGIEval: A Human-Centric Benchmark for Evaluating Foundation Models
AGIEval shows GPT-4 exceeding average human scores on SAT Math at 95% and Chinese college entrance English at 92.5%, while revealing weaker results on complex reasoning tasks.
-
Mixture of Heterogeneous Grouped Experts for Language Modeling
MoHGE achieves standard MoE performance with 20% fewer parameters and balanced GPU utilization via grouped heterogeneous experts, two-level routing, and specialized auxiliary losses.
-
Strix: Re-thinking NPU Reliability from a System Perspective
Strix delivers sub-microsecond fault localisation, detection, and correction on NPUs with 1.04x slowdown and minimal hardware cost by system-level re-partitioning and targeted safeguards.
-
Attention to Mamba: A Recipe for Cross-Architecture Distillation
A two-stage distillation recipe converts a Pythia-1B Transformer into a Mamba model that preserves performance with perplexity 14.11 versus the teacher's 13.86.
-
DeepSpeed Ulysses: System Optimizations for Enabling Training of Extreme Long Sequence Transformer Models
DeepSpeed-Ulysses keeps communication volume constant for sequence-parallel attention when sequence length and device count scale together, delivering 2.5x faster training on 4x longer sequences than prior SOTA.
-
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.
-
A General Language Assistant as a Laboratory for Alignment
Ranked preference modeling outperforms imitation learning for language model alignment and scales more favorably with model size.
-
Rethinking Output Alignment For 1-bit Post-Training Quantization of Large Language Models
A post-training quantization technique for 1-bit LLMs that corrects layer-wise error accumulation and anisotropic representation distortion to preserve output behavior more effectively than existing methods.
-
Motivating Next-Gen Accelerators with Flexible (N:M) Activation Sparsity via Benchmarking Lightweight Post-Training Sparsification Approaches
Post-training N:M activation pruning preserves generative performance in LLMs better than equivalent weight pruning, with the 8:16 pattern emerging as a practical hardware-friendly choice.
-
Combating the Memory Walls: Optimization Pathways for Long-Context Agentic LLM Inference
PLENA introduces a co-designed system with three optimization pathways for long-context agentic LLM inference, claiming up to 2.23x throughput over A100 and 4.04x energy efficiency.
-
Toeplitz MLP Mixers are Low Complexity, Information-Rich Sequence Models
Toeplitz MLP Mixers replace attention with masked Toeplitz multiplications for sub-quadratic complexity while retaining more sequence information and outperforming on copying and in-context tasks.
-
Rethinking Data Curation in LLM Training: Online Reweighting Offers Better Generalization than Offline Methods
ADAPT is an online reweighting framework for LLM training that outperforms offline data selection and mixing methods in cross-benchmark generalization under equal compute.
-
Aging Aware Adaptive Voltage Scaling for Reliable and Efficient AI Accelerators
An aging-aware adaptive voltage scaling framework for AI accelerators reduces predicted threshold voltage shifts by ~19% and aging degradation by up to 46% while saving 14% lifetime power by leveraging neural network resilience.
-
Gemma: Open Models Based on Gemini Research and Technology
Gemma introduces open 2B and 7B LLMs derived from Gemini technology that beat comparable open models on 11 of 18 text tasks and come with safety assessments.
-
Gemma 2: Improving Open Language Models at a Practical Size
Gemma 2 models achieve leading performance at their sizes by combining established Transformer modifications with knowledge distillation for the 2B and 9B variants.
- Structured Recurrent Mixers for Massively Parallelized Sequence Generation
- Lessons from the Trenches on Reproducible Evaluation of Language Models