Acceptance Cards is a new four-diagnostic standard for safe fine-tuning defense claims that requires statistical reliability, fresh semantic generalization, mechanism alignment, and cross-task transfer; under this protocol SafeLoRA fails the full-card pass on Gemma-2-2B-it.
hub Canonical reference
Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone
Canonical reference. 71% of citing Pith papers cite this work as background.
75
Pith papers citing it
Background
71% of classified citations
abstract
We introduce phi-3-mini, a 3.8 billion parameter language model trained on 3.3 trillion tokens, whose overall performance, as measured by both academic benchmarks and internal testing, rivals that of models such as Mixtral 8x7B and GPT-3.5 (e.g., phi-3-mini achieves 69% on MMLU and 8.38 on MT-bench), despite being small enough to be deployed on a phone. Our training dataset is a scaled-up version of the one used for phi-2, composed of heavily filtered publicly available web data and synthetic data. The model is also further aligned for robustness, safety, and chat format. We also provide parameter-scaling results with a 7B, 14B models trained for 4.8T tokens, called phi-3-small, phi-3-medium, both significantly more capable than phi-3-mini (e.g., respectively 75%, 78% on MMLU, and 8.7, 8.9 on MT-bench). To enhance multilingual, multimodal, and long-context capabilities, we introduce three models in the phi-3.5 series: phi-3.5-mini, phi-3.5-MoE, and phi-3.5-Vision. The phi-3.5-MoE, a 16 x 3.8B MoE model with 6.6 billion active parameters, achieves superior performance in language reasoning, math, and code tasks compared to other open-source models of similar scale, such as Llama 3.1 and the Mixtral series, and on par with Gemini-1.5-Flash and GPT-4o-mini. Meanwhile, phi-3.5-Vision, a 4.2 billion parameter model derived from phi-3.5-mini, excels in reasoning tasks and is adept at handling both single-image and text prompts, as well as multi-image and text prompts.
hub tools
citation-role summary
background 5
baseline 1
method 1
citation-polarity summary
claims ledger
- abstract We introduce phi-3-mini, a 3.8 billion parameter language model trained on 3.3 trillion tokens, whose overall performance, as measured by both academic benchmarks and internal testing, rivals that of models such as Mixtral 8x7B and GPT-3.5 (e.g., phi-3-mini achieves 69% on MMLU and 8.38 on MT-bench), despite being small enough to be deployed on a phone. Our training dataset is a scaled-up version of the one used for phi-2, composed of heavily filtered publicly available web data and synthetic data. The model is also further aligned for robustness, safety, and chat format. We also provide param
co-cited works
representative citing papers
ArgBench unifies 33 existing datasets into a standardized benchmark for testing LLMs across 46 argumentation tasks and analyzes the impact of prompting techniques and model factors on performance.
Molmo VLMs trained on newly collected PixMo open datasets achieve state-of-the-art performance among open-weight models and surpass multiple proprietary VLMs including Claude 3.5 Sonnet and Gemini 1.5 Pro.
MMMU-Pro is a stricter multimodal benchmark that removes text-only solvable questions, augments options, and requires reading text from images, yielding substantially lower model scores of 16.8-26.9%.
LiveBench is a contamination-limited LLM benchmark with auto-scored challenging tasks from recent sources across math, coding, reasoning and more, where top models score below 70%.
RULER shows most long-context LMs drop sharply in performance on complex tasks as length and difficulty increase, with only half maintaining results at 32K tokens.
Temperature adjustment on the reference model generalizes inference-time alignment to SLOP ensembles of reward models, with a calibration algorithm that improves robustness to reward hacking while preserving alignment performance.
DisaBench supplies a participatory taxonomy of twelve disability harm types, paired benign-adversarial prompts across seven life domains, and human-annotated data showing that standard safety tests miss context-dependent harms.
Presents CUActSpot benchmark and renderer-LLM data synthesis that lets a 4B model outperform larger open-source models on complex computer interactions.
FinSTaR reaches 78.9% accuracy on a new financial time series reasoning benchmark by applying Compute-in-CoT for deterministic assessments and Scenario-Aware CoT for stochastic predictions.
MemFlow routes queries by intent to tiered memory operations, nearly doubling accuracy of a 1.7B SLM on long-horizon benchmarks compared to full-context baselines.
RouteHijack is a routing-aware jailbreak that identifies safety-critical experts via activation contrast and optimizes suffixes to suppress them, reaching 69.3% average attack success rate on seven MoE LLMs with strong transfer to variants and VLMs.
MASCing uses an LSTM surrogate and optimized steering masks to enable flexible, inference-time control over MoE expert routing for safety objectives, improving jailbreak defense and content generation success rates substantially across multiple models.
Language models frequently violate temporal scope stability in multi-turn dialogues by drifting toward present-day assumptions even when they possess the correct facts.
LAT-Audio introduces a global-to-local reasoning approach with TWA-CoT that outperforms prior models on temporal tasks for audio up to 30 minutes.
Clinical narrative format beats raw JSON for LLMs up to 8B parameters on medication reconciliation but raw JSON wins at 70B scale, with omissions as the main error type.
Single-agent systems with tools provide the optimal performance-efficiency trade-off for small language models, outperforming base models and multi-agent setups.
Adaptive trie-guided decoding with document context and tunable penalties improves in-document query auto-completion, outperforming baselines and larger models like LLaMA-3 on seen queries.
Introduces the U-HOI task and shows MLLMs plus a language-to-graph pipeline can handle human-object interactions without any predefined vocabulary at training or inference time.
Alignment of vision-language models with human V1-V3 early visual cortex negatively predicts resistance to sycophantic gaslighting attacks.
GeoMMBench reveals deficiencies in current multimodal LLMs for geoscience tasks while GeoMMAgent demonstrates that tool-integrated agents achieve significantly higher performance.
LongMemEval benchmarks long-term memory in chat assistants, revealing 30% accuracy drops across sustained interactions and proposing indexing-retrieval-reading optimizations that boost performance.
LLMs display high variance and major accuracy drops on GSM-Symbolic variants of grade-school math problems, indicating they replicate training patterns rather than execute logical reasoning.
MSIFR stops faulty LLM generations early via staged rule-based checks, reducing token consumption 11-78% with no accuracy loss.
citing papers explorer
-
Acceptance Cards:A Four-Diagnostic Standard for Safe Fine-Tuning Defense Claims
Acceptance Cards is a new four-diagnostic standard for safe fine-tuning defense claims that requires statistical reliability, fresh semantic generalization, mechanism alignment, and cross-task transfer; under this protocol SafeLoRA fails the full-card pass on Gemma-2-2B-it.
-
ArgBench: Benchmarking LLMs on Computational Argumentation Tasks
ArgBench unifies 33 existing datasets into a standardized benchmark for testing LLMs across 46 argumentation tasks and analyzes the impact of prompting techniques and model factors on performance.
-
Molmo and PixMo: Open Weights and Open Data for State-of-the-Art Vision-Language Models
Molmo VLMs trained on newly collected PixMo open datasets achieve state-of-the-art performance among open-weight models and surpass multiple proprietary VLMs including Claude 3.5 Sonnet and Gemini 1.5 Pro.
-
MMMU-Pro: A More Robust Multi-discipline Multimodal Understanding Benchmark
MMMU-Pro is a stricter multimodal benchmark that removes text-only solvable questions, augments options, and requires reading text from images, yielding substantially lower model scores of 16.8-26.9%.
-
LiveBench: A Challenging, Contamination-Limited LLM Benchmark
LiveBench is a contamination-limited LLM benchmark with auto-scored challenging tasks from recent sources across math, coding, reasoning and more, where top models score below 70%.
-
RULER: What's the Real Context Size of Your Long-Context Language Models?
RULER shows most long-context LMs drop sharply in performance on complex tasks as length and difficulty increase, with only half maintaining results at 32K tokens.
-
Temper and Tilt Lead to SLOP: Reward Hacking Mitigation with Inference-Time Alignment
Temperature adjustment on the reference model generalizes inference-time alignment to SLOP ensembles of reward models, with a calibration algorithm that improves robustness to reward hacking while preserving alignment performance.
-
DisaBench: A Participatory Evaluation Framework for Disability Harms in Language Models
DisaBench supplies a participatory taxonomy of twelve disability harm types, paired benign-adversarial prompts across seven life domains, and human-annotated data showing that standard safety tests miss context-dependent harms.
-
Covering Human Action Space for Computer Use: Data Synthesis and Benchmark
Presents CUActSpot benchmark and renderer-LLM data synthesis that lets a 4B model outperform larger open-source models on complex computer interactions.
-
FinSTaR: Towards Financial Reasoning with Time Series Reasoning Models
FinSTaR reaches 78.9% accuracy on a new financial time series reasoning benchmark by applying Compute-in-CoT for deterministic assessments and Scenario-Aware CoT for stochastic predictions.
-
MemFlow: Intent-Driven Memory Orchestration for Small Language Model Agents
MemFlow routes queries by intent to tiered memory operations, nearly doubling accuracy of a 1.7B SLM on long-horizon benchmarks compared to full-context baselines.
-
RouteHijack: Routing-Aware Attack on Mixture-of-Experts LLMs
RouteHijack is a routing-aware jailbreak that identifies safety-critical experts via activation contrast and optimizes suffixes to suppress them, reaching 69.3% average attack success rate on seven MoE LLMs with strong transfer to variants and VLMs.
-
MASCing: Configurable Mixture-of-Experts Behavior via Activation Steering Masks
MASCing uses an LSTM surrogate and optimized steering masks to enable flexible, inference-time control over MoE expert routing for safety objectives, improving jailbreak defense and content generation success rates substantially across multiple models.
-
Evaluating Temporal Consistency in Multi-Turn Language Models
Language models frequently violate temporal scope stability in multi-turn dialogues by drifting toward present-day assumptions even when they possess the correct facts.
-
Listening with Time: Precise Temporal Awareness for Long-Form Audio Understanding
LAT-Audio introduces a global-to-local reasoning approach with TWA-CoT that outperforms prior models on temporal tasks for audio up to 30 minutes.
-
Serialisation Strategy Matters: How FHIR Data Format Affects LLM Medication Reconciliation
Clinical narrative format beats raw JSON for LLMs up to 8B parameters on medication reconciliation but raw JSON wins at 70B scale, with omissions as the main error type.
-
Rethinking Scale: Deployment Trade-offs of Small Language Models under Agent Paradigms
Single-agent systems with tools provide the optimal performance-efficiency trade-off for small language models, outperforming base models and multi-agent setups.
-
DocQAC: Adaptive Trie-Guided Decoding for Effective In-Document Query Auto-Completion
Adaptive trie-guided decoding with document context and tunable penalties improves in-document query auto-completion, outperforming baselines and larger models like LLaMA-3 on seen queries.
-
Towards Unconstrained Human-Object Interaction
Introduces the U-HOI task and shows MLLMs plus a language-to-graph pipeline can handle human-object interactions without any predefined vocabulary at training or inference time.
-
Gaslight, Gatekeep, V1-V3: Early Visual Cortex Alignment Shields Vision-Language Models from Sycophantic Manipulation
Alignment of vision-language models with human V1-V3 early visual cortex negatively predicts resistance to sycophantic gaslighting attacks.
-
GeoMMBench and GeoMMAgent: Toward Expert-Level Multimodal Intelligence in Geoscience and Remote Sensing
GeoMMBench reveals deficiencies in current multimodal LLMs for geoscience tasks while GeoMMAgent demonstrates that tool-integrated agents achieve significantly higher performance.
-
LongMemEval: Benchmarking Chat Assistants on Long-Term Interactive Memory
LongMemEval benchmarks long-term memory in chat assistants, revealing 30% accuracy drops across sustained interactions and proposing indexing-retrieval-reading optimizations that boost performance.
-
GSM-Symbolic: Understanding the Limitations of Mathematical Reasoning in Large Language Models
LLMs display high variance and major accuracy drops on GSM-Symbolic variants of grade-school math problems, indicating they replicate training patterns rather than execute logical reasoning.
-
Know When To Fold 'Em: Token-Efficient LLM Synthetic Data Generation via Multi-Stage In-Flight Rejection
MSIFR stops faulty LLM generations early via staged rule-based checks, reducing token consumption 11-78% with no accuracy loss.
-
Not Just RLHF: Why Alignment Alone Won't Fix Multi-Agent Sycophancy
Pretrained base models exhibit higher yield to peer disagreement than RLHF instruct variants, with the effect localized to mid-layer attention and mitigated by structured dissent rather than prompt defenses.
-
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.
-
MoE-Hub: Taming Software Complexity for Seamless MoE Overlap with Hardware-Accelerated Communication on Multi-GPU Systems
MoE-Hub enables seamless MoE communication overlap via hardware-accelerated destination-agnostic data transmission, delivering 1.40x-3.08x per-layer and 1.21x-1.98x end-to-end speedups over prior systems.
-
ZAYA1-8B Technical Report
ZAYA1-8B is a reasoning MoE model with 700M active parameters that matches larger models on math and coding benchmarks and reaches 91.9% on AIME'25 via Markovian RSA test-time compute.
-
Power Distribution Bridges Sampling, Self-Reward RL, and Self-Distillation
The power distribution is the target of power sampling, the closed-form solution to self-reward KL-regularized RL, and the basis for power self-distillation that matches sampling performance at lower cost.
-
Edge-Efficient Image Restoration: Transformer Distillation into State-Space Models
Hybrid transformer-SSM networks found by multi-objective search run 1.17x to 3.4x faster on edge CPUs for image restoration tasks with competitive quality.
-
DocSync: Agentic Documentation Maintenance via Critic-Guided Reflexion
DocSync fuses AST-aware retrieval with an iterative critic loop to update documentation, outperforming CodeT5-base on semantic alignment and automated judge scores in a proxy code-to-text task.
-
Test-Time Safety Alignment
Optimizing input embeddings sub-lexically via black-box zeroth-order gradients neutralizes all safety-flagged responses from aligned models on standard benchmarks.
-
Architecture Determines Observability of Transformers
Architecture and training determine whether transformers retain a readable internal signal that lets activation monitors catch errors missed by output confidence.
-
Mind the Unseen Mass: Unmasking LLM Hallucinations via Soft-Hybrid Alphabet Estimation
SHADE adaptively combines coverage and spectral signals to estimate semantic alphabet size from few LLM samples, yielding better performance than baselines in low-sample regimes for alphabet estimation and QA error detection.
-
Co-Located Tests, Better AI Code: How Test Syntax Structure Affects Foundation Model Code Generation
Co-locating tests with implementation code yields substantially higher preservation and correctness in foundation-model-generated programs than separated test syntax.
-
Accuracy Is Speed: Towards Long-Context-Aware Routing for Distributed LLM Serving
In long-context LLM serving, accuracy becomes speed via retry dynamics, and accuracy-aware routing reduces time-to-correct-answer.
-
SIMMER: Cross-Modal Food Image--Recipe Retrieval via MLLM-Based Embedding
SIMMER uses a single multimodal LLM (VLM2Vec) with custom prompts and partial-recipe augmentation to embed food images and recipes, achieving new state-of-the-art retrieval accuracy on Recipe1M.
-
Robust Ultra Low-Bit Post-Training Quantization via Stable Diagonal Curvature Estimate
DASH-Q uses a stable diagonal curvature estimate and weighted least squares to achieve robust ultra-low-bit post-training quantization of LLMs, improving zero-shot accuracy by 7% on average over baselines.
-
From Anchors to Supervision: Memory-Graph Guided Corpus-Free Unlearning for Large Language Models
MAGE builds a memory graph from a user anchor to generate its own supervision signals for corpus-free unlearning, matching the effectiveness of methods that use external reference data on TOFU and RWKU benchmarks.
-
MedRCube: A Multidimensional Framework for Fine-Grained and In-Depth Evaluation of MLLMs in Medical Imaging
MedRCube is a new fine-grained evaluation framework that benchmarks 33 MLLMs on medical imaging, ranks Lingshu-32B highest, and finds a significant positive link between shortcut behaviors and diagnostic performance.
-
Local-Splitter: A Measurement Study of Seven Tactics for Reducing Cloud LLM Token Usage on Coding-Agent Workloads
Combining local routing with prompt compression saves 45-79% cloud tokens on edit and explanation workloads, while a fuller set including draft-review saves 51% on RAG-heavy tasks.
-
MLLM-as-a-Judge Exhibits Model Preference Bias
MLLMs show self-preference bias and family-level mutual bias when judging captions; Philautia-Eval quantifies it and Pomms ensemble reduces it.
-
EdgeCIM: A Hardware-Software Co-Design for CIM-Based Acceleration of Small Language Models
A CIM-based hardware-software co-design in 65nm achieves up to 7.3x higher throughput and 49.59x better energy efficiency than NVIDIA Orin Nano for LLaMA3.2-1B, averaging 336 tokens/s and 173 tokens/J under INT4 across multiple SLMs.
-
Adapting 2D Multi-Modal Large Language Model for 3D CT Image Analysis
Transferring a 2D MLLM to 3D CT inputs via parameter reuse, a Text-Guided Hierarchical MoE framework, and two-stage training yields better performance than prior 3D medical MLLMs on medical report generation and visual question answering.
-
Dead Weights, Live Signals: Feedforward Graphs of Frozen Language Models
A feedforward graph of heterogeneous frozen LLMs linked by linear projections in a shared latent space outperforms single models on ARC-Challenge, OpenBookQA, and MMLU using just 17.6M trainable parameters.
-
ReAlign: Optimizing the Visual Document Retriever with Reasoning-Guided Fine-Grained Alignment
ReAlign improves visual document retrieval by training retrievers to match query-induced rankings with rankings derived from VLM-generated, region-focused descriptions of relevant page content.
-
In-Place Test-Time Training
In-Place TTT adapts LLM MLP projection matrices at test time with a next-token-aligned objective and chunk-wise updates, enabling better long-context performance as a drop-in enhancement.
-
Do Hallucination Neurons Generalize? Evidence from Cross-Domain Transfer in LLMs
Hallucination neurons in LLMs are domain-specific, with cross-domain classifiers dropping from AUROC 0.783 within-domain to 0.563 across domains.
-
SAM 3D: 3Dfy Anything in Images
SAM 3D reconstructs 3D objects from single images with geometry, texture, and pose using human-model annotated data at scale and synthetic-to-real training, achieving 5:1 human preference wins.
-
Expanding Performance Boundaries of Open-Source Multimodal Models with Model, Data, and Test-Time Scaling
InternVL 2.5 is the first open-source MLLM to surpass 70% on the MMMU benchmark via model, data, and test-time scaling, with a 3.7-point gain from chain-of-thought reasoning.