MME-RealWorld is the largest manually annotated high-resolution benchmark for MLLMs, where even the best models achieve less than 60% accuracy on challenging real-world tasks.
hub Canonical reference
PaLM 2 Technical Report
Canonical reference. 88% of citing Pith papers cite this work as background.
93
Pith papers citing it
Background
88% of classified citations
abstract
We introduce PaLM 2, a new state-of-the-art language model that has better multilingual and reasoning capabilities and is more compute-efficient than its predecessor PaLM. PaLM 2 is a Transformer-based model trained using a mixture of objectives. Through extensive evaluations on English and multilingual language, and reasoning tasks, we demonstrate that PaLM 2 has significantly improved quality on downstream tasks across different model sizes, while simultaneously exhibiting faster and more efficient inference compared to PaLM. This improved efficiency enables broader deployment while also allowing the model to respond faster, for a more natural pace of interaction. PaLM 2 demonstrates robust reasoning capabilities exemplified by large improvements over PaLM on BIG-Bench and other reasoning tasks. PaLM 2 exhibits stable performance on a suite of responsible AI evaluations, and enables inference-time control over toxicity without additional overhead or impact on other capabilities. Overall, PaLM 2 achieves state-of-the-art performance across a diverse set of tasks and capabilities. When discussing the PaLM 2 family, it is important to distinguish between pre-trained models (of various sizes), fine-tuned variants of these models, and the user-facing products that use these models. In particular, user-facing products typically include additional pre- and post-processing steps. Additionally, the underlying models may evolve over time. Therefore, one should not expect the performance of user-facing products to exactly match the results reported in this report.
hub tools
citation-role summary
background 23
baseline 2
method 1
citation-polarity summary
claims ledger
- abstract We introduce PaLM 2, a new state-of-the-art language model that has better multilingual and reasoning capabilities and is more compute-efficient than its predecessor PaLM. PaLM 2 is a Transformer-based model trained using a mixture of objectives. Through extensive evaluations on English and multilingual language, and reasoning tasks, we demonstrate that PaLM 2 has significantly improved quality on downstream tasks across different model sizes, while simultaneously exhibiting faster and more efficient inference compared to PaLM. This improved efficiency enables broader deployment while also all
- method backbone by representing actions as another language, and training action text tokens together with vision-language data. metadata) per embodiment (Fig. 2(b)), where Franka dom- inates. Fig. 2(c) shows the breakdown of trajectories per embodiment. To further analyze the diversity, we use the language annotations present in our data. We use the PaLM language model [3] to extract objects and behaviors from the instructions. Fig. 2(d,e) show the diversity of skills and objects. While most skills be
- background For instance, Flan- PaLM-540B, which is instruction-finetuned on 1.8K tasks, outperforms PaLM-540B by a large margin (+9.4% on av- erage). The finetuning data comprises 473 datasets, 146 task categories, and 1,836 total tasks, as illustrated in Fig 14. Fig. 14: Flan-PaLM finetuning consist of 473 datasets in above task categories. Courtesy of [74]. PaLM-2 [75] is a more compute-efficient LLM with bet- ter multilingual and reasoning capabilities, compared to its predecessor PaLM. PaLM-2 is traine
- baseline 0Prometheus-8x7b-v2 [353] Mixtral-8x7B-Instruct [319]93.0 47.1 80.5 77.4 74.5Critic-RM-Rank [991] Llama-3.1-70B-Instruct [168]97.0 58.0 84.0 92.0 82.8RM [689] Llama-3.1-70B-Instruct [168]98.3 74.5 83.8 88.0 86.4SynRM [968] Llama-3.1-70B-Instruct [168]97.5 76.8 86.3 88.5 87.3CLoud [17] Llama-3-70B-Instruct [168]98.0 75.6 87.6 89.0 87.6FLAMe-RM-24B [753] PaLM-2-24B [16] 92.2 75.7 89.6 93.8 87.8SteerLM-RM 70B [829] Llama-2-70B-chat [743]91.3 80.3 90.6 92.8 88.8Llama-3-OffsetBias-RM-8B [585]Llama-3-
- background standing of the survey authors by reading the papers, blog articles, interview reports and APIs released by OpenAI. 14. https://hackernoon.com/an-interview-with-ilya-sutskever-co- founder-of-openai models was already explored in the early days of Ope- nAI, while it was attempted with recurrent neural net- works (RNN) [121]. With the advent of Transformer, OpenAI developed two initial GPT models, namely GPT-1 [122] and GPT-2 [26], which can be considered as the foundation to more powerful models
co-cited works
representative citing papers
BLaIR is a new benchmark and 570M-review dataset showing that LLM performance rankings on recommendation tasks have little correlation with rankings on general embedding benchmarks like MTEB.
LOVER creates an unsupervised logic-regularized verifier that reaches 95% of supervised verifier performance on reasoning tasks across 10 datasets.
AS-LoRA adaptively chooses which LoRA factor to update per layer and round using a curvature-aware second-order score, eliminating reconstruction error floors and improving performance in DP federated learning.
E-MIA converts document details into four types of exam questions and aggregates the RAG's answers into a membership score that separates member and non-member documents better than prior similarity-based or probe-based attacks.
Unembedding collapse in transformers prevents distinguishing unseen tokens in symbolic reasoning, but targeted interventions restore generalization.
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.
Drift-AR achieves 3.8-5.5x speedup in AR-diffusion image models by using entropy to enable entropy-informed speculative decoding and single-step (1-NFE) anti-symmetric drifting decoding.
PEEM is a multi-criteria LLM-based evaluator for prompts and responses that aligns with standard accuracy while enabling zero-shot prompt optimization via feedback.
M³KG-RAG improves multimodal reasoning in large language models by constructing multi-hop knowledge graphs and selectively pruning retrieved context with GRASP.
ActiveDPO is a theoretically grounded active data selection method for sample-efficient LLM alignment that parameterizes the reward model directly with the LLM being aligned.
WE-MATH benchmark reveals most LMMs rely on rote memorization for visual math while GPT-4o has shifted toward knowledge generalization.
Scaled vanilla autoregressive models based on Llama achieve 2.18 FID on ImageNet 256x256 image generation, beating popular diffusion models without visual inductive biases.
Absorbing discrete diffusion models the conditional distributions of clean data; reparameterizing yields a time-independent RADD that unifies with AO-ARMs and reaches SOTA perplexity among diffusion models on zero-shot language benchmarks.
Infini-attention combines compressive memory with masked local attention and long-term linear attention inside each Transformer block to support infinite context length with bounded resources.
A collaborative dataset spanning 22 robots and 527 skills enables RT-X models that transfer capabilities across different robot embodiments.
Varying decoding strategies such as temperature and sampling methods jailbreaks safety alignments in open-source LLMs, raising misalignment from 0% to over 95% at 30x lower cost than prior attacks.
UniSim learns a universal real-world simulator from orchestrated diverse datasets, enabling zero-shot deployment of policies trained purely in simulation.
A new shared video-image tokenizer enables large language models to surpass diffusion models on standard visual generation benchmarks.
Large language models can optimize by being prompted with histories of past solutions and scores to propose better ones, producing prompts that raise accuracy up to 8% on GSM8K and 50% on Big-Bench Hard over human-designed baselines.
Introduces BanglaMedVQA dataset of clinically validated image-question-answer pairs and benchmarks foundation models, finding substantially lower performance than on English MedVQA especially on diagnostic questions.
A dataset-agnostic framework converts text tool-calling benchmarks to paired audio evaluations via TTS, speaker variation and noise, then evaluates seven omni-modal models showing model- and task-dependent performance with small text-to-voice gaps.
LoKA enables practical FP8 use in numerically sensitive large recommendation models via online profiling of activations, reusable model modifications for stability, and dynamic kernel dispatching.
XPERT extracts and reuses cross-domain expert knowledge from pre-trained MoE LLMs via inference analysis and tensor decomposition to improve performance and convergence in downstream language model training.
citing papers explorer
-
MME-RealWorld: Could Your Multimodal LLM Challenge High-Resolution Real-World Scenarios that are Difficult for Humans?
MME-RealWorld is the largest manually annotated high-resolution benchmark for MLLMs, where even the best models achieve less than 60% accuracy on challenging real-world tasks.
-
Bridging Language and Items for Retrieval and Recommendation: Benchmarking LLMs as Semantic Encoders
BLaIR is a new benchmark and 570M-review dataset showing that LLM performance rankings on recommendation tasks have little correlation with rankings on general embedding benchmarks like MTEB.
-
Logic-Regularized Verifier Elicits Reasoning from LLMs
LOVER creates an unsupervised logic-regularized verifier that reaches 95% of supervised verifier performance on reasoning tasks across 10 datasets.
-
Adaptive Selection of LoRA Components in Privacy-Preserving Federated Learning
AS-LoRA adaptively chooses which LoRA factor to update per layer and round using a curvature-aware second-order score, eliminating reconstruction error floors and improving performance in DP federated learning.
-
E-MIA: Exam-Style Black-Box Membership Inference Attacks against RAG Systems
E-MIA converts document details into four types of exam questions and aggregates the RAG's answers into a membership score that separates member and non-member documents better than prior similarity-based or probe-based attacks.
-
To See the Unseen: on the Generalization Ability of Transformers in Symbolic Reasoning
Unembedding collapse in transformers prevents distinguishing unseen tokens in symbolic reasoning, but targeted interventions restore generalization.
-
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.
-
Drift-AR: Single-Step Visual Autoregressive Generation via Anti-Symmetric Drifting
Drift-AR achieves 3.8-5.5x speedup in AR-diffusion image models by using entropy to enable entropy-informed speculative decoding and single-step (1-NFE) anti-symmetric drifting decoding.
-
PEEM: Prompt Engineering Evaluation Metrics for Interpretable Joint Evaluation of Prompts and Responses
PEEM is a multi-criteria LLM-based evaluator for prompts and responses that aligns with standard accuracy while enabling zero-shot prompt optimization via feedback.
-
M$^3$KG-RAG: Multi-hop Multimodal Knowledge Graph-enhanced Retrieval-Augmented Generation
M³KG-RAG improves multimodal reasoning in large language models by constructing multi-hop knowledge graphs and selectively pruning retrieved context with GRASP.
-
ActiveDPO: Active Direct Preference Optimization for Sample-Efficient Alignment
ActiveDPO is a theoretically grounded active data selection method for sample-efficient LLM alignment that parameterizes the reward model directly with the LLM being aligned.
-
We-Math: Does Your Large Multimodal Model Achieve Human-like Mathematical Reasoning?
WE-MATH benchmark reveals most LMMs rely on rote memorization for visual math while GPT-4o has shifted toward knowledge generalization.
-
Autoregressive Model Beats Diffusion: Llama for Scalable Image Generation
Scaled vanilla autoregressive models based on Llama achieve 2.18 FID on ImageNet 256x256 image generation, beating popular diffusion models without visual inductive biases.
-
Your Absorbing Discrete Diffusion Secretly Models the Conditional Distributions of Clean Data
Absorbing discrete diffusion models the conditional distributions of clean data; reparameterizing yields a time-independent RADD that unifies with AO-ARMs and reaches SOTA perplexity among diffusion models on zero-shot language benchmarks.
-
Leave No Context Behind: Efficient Infinite Context Transformers with Infini-attention
Infini-attention combines compressive memory with masked local attention and long-term linear attention inside each Transformer block to support infinite context length with bounded resources.
-
Open X-Embodiment: Robotic Learning Datasets and RT-X Models
A collaborative dataset spanning 22 robots and 527 skills enables RT-X models that transfer capabilities across different robot embodiments.
-
Catastrophic Jailbreak of Open-source LLMs via Exploiting Generation
Varying decoding strategies such as temperature and sampling methods jailbreaks safety alignments in open-source LLMs, raising misalignment from 0% to over 95% at 30x lower cost than prior attacks.
-
Learning Interactive Real-World Simulators
UniSim learns a universal real-world simulator from orchestrated diverse datasets, enabling zero-shot deployment of policies trained purely in simulation.
-
Language Model Beats Diffusion -- Tokenizer is Key to Visual Generation
A new shared video-image tokenizer enables large language models to surpass diffusion models on standard visual generation benchmarks.
-
Large Language Models as Optimizers
Large language models can optimize by being prompted with histories of past solutions and scores to propose better ones, producing prompts that raise accuracy up to 8% on GSM8K and 50% on Big-Bench Hard over human-designed baselines.
-
How Good LLMs Are at Answering Bangla Medical Visual Questions? Dataset and Benchmarking
Introduces BanglaMedVQA dataset of clinically validated image-question-answer pairs and benchmarks foundation models, finding substantially lower performance than on English MedVQA especially on diagnostic questions.
-
From Text to Voice: A Reproducible and Verifiable Framework for Evaluating Tool Calling LLM Agents
A dataset-agnostic framework converts text tool-calling benchmarks to paired audio evaluations via TTS, speaker variation and noise, then evaluates seven omni-modal models showing model- and task-dependent performance with small text-to-voice gaps.
-
LoKA: Low-precision Kernel Applications for Recommendation Models At Scale
LoKA enables practical FP8 use in numerically sensitive large recommendation models via online profiling of activations, reusable model modifications for stability, and dynamic kernel dispatching.
-
XPERT: Expert Knowledge Transfer for Effective Training of Language Models
XPERT extracts and reuses cross-domain expert knowledge from pre-trained MoE LLMs via inference analysis and tensor decomposition to improve performance and convergence in downstream language model training.
-
HyperEyes: Dual-Grained Efficiency-Aware Reinforcement Learning for Parallel Multimodal Search Agents
HyperEyes presents a parallel multimodal search agent using dual-grained efficiency-aware RL with a new TRACE reward and IMEB benchmark, claiming 9.9% higher accuracy and 5.3x fewer tool calls than prior open-source agents.
-
Towards Reliable LLM Evaluation: Correcting the Winner's Curse in Adaptive Benchmarking
SIREN corrects winner's curse bias in adaptive LLM benchmarking via selection-aware repeated splits and bootstrap for valid procedure-level confidence intervals.
-
Breaking Lock-In: Preserving Steerability under Low-Data VLA Post-Training
DeLock mitigates lock-in in low-data VLA post-training via visual grounding preservation and test-time contrastive prompt guidance, outperforming baselines across eight evaluations while matching data-heavy generalist policies.
-
BarrierSteer: LLM Safety via Learning Barrier Steering
BarrierSteer applies control barrier functions to LLM latent states for constraint-guided steering that reduces unsafe generations while preserving utility.
-
The Effect of Mini-Batch Noise on the Implicit Bias of Adam
Mini-batch noise reverses how Adam's β2 controls anti-regularization, making default momentum values suitable for small batches but requiring β1 closer to β2 for large batches to favor flatter minima.
-
ALL-FEM: Agentic Large Language models Fine-tuned for Finite Element Methods
ALL-FEM fine-tunes LLMs on a corpus of verified FEniCS scripts and uses multi-agent workflows to automate finite element code generation, achieving 71.79% success on 39 benchmarks across elasticity, flow, and coupled problems.
-
Scaling Sequence-to-Sequence Generative Neural Rendering
Kaleido is a masked autoregressive generative model that unifies 3D view synthesis and video modeling by pre-training a single transformer on video data, achieving SOTA zero-shot and many-view performance on view synthesis benchmarks.
-
Step-Audio 2 Technical Report
Step-Audio 2 integrates a latent audio encoder, reasoning-centric reinforcement learning, and discrete audio token generation into language modeling to deliver state-of-the-art performance on audio understanding and conversational benchmarks.
-
Autoregressive Video Generation without Vector Quantization
NOVA reformulates video generation as non-quantized autoregressive frame-by-frame temporal prediction combined with set-by-set spatial prediction, outperforming prior AR video models and some diffusion models in efficiency and quality.
-
EMMA: End-to-End Multimodal Model for Autonomous Driving
EMMA is an end-to-end multimodal LLM that converts camera data into trajectories, objects, and road graphs via text prompts and reports state-of-the-art motion planning on nuScenes plus competitive detection results on Waymo.
-
Senna: Bridging Large Vision-Language Models and End-to-End Autonomous Driving
Senna decouples language-based high-level planning from an LVLM with low-level trajectory prediction from an E2E model, reporting 27% lower planning error and 33% lower collisions after pre-training on DriveX and fine-tuning on nuScenes.
-
A Survey on Vision-Language-Action Models for Embodied AI
This is the first survey on vision-language-action models, providing a taxonomy across three lines, plus summaries of datasets, simulators, benchmarks, challenges, and future directions in embodied AI.
-
Capabilities of Gemini Models in Medicine
Med-Gemini sets new records on 10 of 14 medical benchmarks including 91.1% on MedQA-USMLE, beats GPT-4V by 44.5% on multimodal tasks, and surpasses humans on medical text summarization.
-
Laissez-Faire Harms: Algorithmic Biases in Generative Language Models
Generative LMs in laissez-faire open-ended prompting settings disproportionately generate subordinated portrayals of minoritized race, gender, and sexual orientation identities at rates hundreds to thousands of times higher than empowering ones.
-
LiveCodeBench: Holistic and Contamination Free Evaluation of Large Language Models for Code
LiveCodeBench collects 400 recent contest problems to create a contamination-free benchmark evaluating LLMs on code generation and related capabilities like self-repair and execution.
-
Corrective Retrieval Augmented Generation
CRAG improves RAG robustness via a retrieval quality evaluator that triggers web augmentation and a decompose-recompose filter to focus on relevant information, yielding better results on short- and long-form generation tasks.
-
VideoPoet: A Large Language Model for Zero-Shot Video Generation
VideoPoet is a large language model that performs zero-shot video generation with audio from diverse multimodal conditioning signals.
-
Math-Shepherd: Verify and Reinforce LLMs Step-by-step without Human Annotations
Math-Shepherd is an automatically trained process reward model that scores solution steps to verify and reinforce LLMs, lifting Mistral-7B from 77.9% to 89.1% on GSM8K and 28.6% to 43.5% on MATH.
-
The Falcon Series of Open Language Models
Falcon-180B is a 180B-parameter open decoder-only model trained on 3.5 trillion tokens that approaches PaLM-2-Large performance at lower cost and is released with dataset extracts.
-
Video-LLaVA: Learning United Visual Representation by Alignment Before Projection
Video-LLaVA creates a unified visual representation for images and videos via pre-projection alignment, enabling mutual enhancement from joint training and strong results on image and video benchmarks.
-
Qwen-Audio: Advancing Universal Audio Understanding via Unified Large-Scale Audio-Language Models
Qwen-Audio trains a unified model on diverse audio and tasks with hierarchical tags to enable strong zero-shot performance on audio understanding benchmarks and multi-turn audio chat.
-
SALMONN: Towards Generic Hearing Abilities for Large Language Models
SALMONN integrates speech and audio encoders with a text-based LLM to process general audio inputs, achieve competitive results on trained tasks, and exhibit emergent cross-modal abilities.
-
BitNet: Scaling 1-bit Transformers for Large Language Models
BitNet shows that 1-bit Transformers can match the performance of 8-bit and FP16 models on language modeling with much smaller memory footprint and energy use, while following a similar scaling law.
-
Large Language Models Cannot Self-Correct Reasoning Yet
LLMs cannot reliably self-correct reasoning mistakes using only their internal capabilities and often degrade in performance without external feedback.
-
ToRA: A Tool-Integrated Reasoning Agent for Mathematical Problem Solving
ToRA trains language models on interactive tool-use trajectories with imitation learning and output shaping to integrate reasoning and external tools, yielding 13-19% gains on math datasets and new highs like 44.6% on MATH for a 7B model.
-
Aligning Large Multimodal Models with Factually Augmented RLHF
Factually Augmented RLHF aligns large multimodal models to reduce hallucinations, reaching 94% of GPT-4 on LLaVA-Bench and 60% improvement on the new MMHAL-BENCH.