Rho-1: Not All Tokens Are What You Need
Paper • 2404.07965 • Published • 94
LM Training
-
-
VASA-1: Lifelike Audio-Driven Talking Faces Generated in Real Time
Paper • 2404.10667 • Published • 24 -
Instruction-tuned Language Models are Better Knowledge Learners
Paper • 2402.12847 • Published • 26 -
DoRA: Weight-Decomposed Low-Rank Adaptation
Paper • 2402.09353 • Published • 32 -
QLoRA: Efficient Finetuning of Quantized LLMs
Paper • 2305.14314 • Published • 60 -
GaLore: Memory-Efficient LLM Training by Gradient Low-Rank Projection
Paper • 2403.03507 • Published • 189 -
Reverse Training to Nurse the Reversal Curse
Paper • 2403.13799 • Published • 13 -
SOLAR 10.7B: Scaling Large Language Models with Simple yet Effective Depth Up-Scaling
Paper • 2312.15166 • Published • 61 -
ReFT: Representation Finetuning for Language Models
Paper • 2404.03592 • Published • 101 -
Direct Nash Optimization: Teaching Language Models to Self-Improve with General Preferences
Paper • 2404.03715 • Published • 62 -
Learn Your Reference Model for Real Good Alignment
Paper • 2404.09656 • Published • 90 -
Megalodon: Efficient LLM Pretraining and Inference with Unlimited Context Length
Paper • 2404.08801 • Published • 66 -
Pre-training Small Base LMs with Fewer Tokens
Paper • 2404.08634 • Published • 36 -
JetMoE: Reaching Llama2 Performance with 0.1M Dollars
Paper • 2404.07413 • Published • 38 -
MiniCPM: Unveiling the Potential of Small Language Models with Scalable Training Strategies
Paper • 2404.06395 • Published • 24 -
SambaLingo: Teaching Large Language Models New Languages
Paper • 2404.05829 • Published • 13 -
Advancing LLM Reasoning Generalists with Preference Trees
Paper • 2404.02078 • Published • 46 -
Poro 34B and the Blessing of Multilinguality
Paper • 2404.01856 • Published • 15 -
Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone
Paper • 2404.14219 • Published • 259 -
The Instruction Hierarchy: Training LLMs to Prioritize Privileged Instructions
Paper • 2404.13208 • Published • 40 -
Eagle and Finch: RWKV with Matrix-Valued States and Dynamic Recurrence
Paper • 2404.05892 • Published • 40 -
Mamba: Linear-Time Sequence Modeling with Selective State Spaces
Paper • 2312.00752 • Published • 150 -
OpenELM: An Efficient Language Model Family with Open-source Training and Inference Framework
Paper • 2404.14619 • Published • 126 -
Jamba: A Hybrid Transformer-Mamba Language Model
Paper • 2403.19887 • Published • 112 -
Make Your LLM Fully Utilize the Context
Paper • 2404.16811 • Published • 55 -
Tele-FLM Technical Report
Paper • 2404.16645 • Published • 18 -
PLLaVA : Parameter-free LLaVA Extension from Images to Videos for Video Dense Captioning
Paper • 2404.16994 • Published • 37 -
LoRA Land: 310 Fine-tuned LLMs that Rival GPT-4, A Technical Report
Paper • 2405.00732 • Published • 122 -
Iterative Reasoning Preference Optimization
Paper • 2404.19733 • Published • 50 -
What matters when building vision-language models?
Paper • 2405.02246 • Published • 104 -
MoRA: High-Rank Updating for Parameter-Efficient Fine-Tuning
Paper • 2405.12130 • Published • 50 -
Your Transformer is Secretly Linear
Paper • 2405.12250 • Published • 157 -
Perplexed by Perplexity: Perplexity-Based Data Pruning With Small Reference Models
Paper • 2405.20541 • Published • 24 -
How Do Large Language Models Acquire Factual Knowledge During Pretraining?
Paper • 2406.11813 • Published • 31 -
Magpie: Alignment Data Synthesis from Scratch by Prompting Aligned LLMs with Nothing
Paper • 2406.08464 • Published • 72 -
The Llama 3 Herd of Models
Paper • 2407.21783 • Published • 118 -
Gemma 2: Improving Open Language Models at a Practical Size
Paper • 2408.00118 • Published • 78 -
MoMa: Efficient Early-Fusion Pre-training with Mixture of Modality-Aware Experts
Paper • 2407.21770 • Published • 22 -
LongWriter: Unleashing 10,000+ Word Generation from Long Context LLMs
Paper • 2408.07055 • Published • 69 -
Data curation via joint example selection further accelerates multimodal learning
Paper • 2406.17711 • Published • 3 -
Jamba-1.5: Hybrid Transformer-Mamba Models at Scale
Paper • 2408.12570 • Published • 32 -
OLMoE: Open Mixture-of-Experts Language Models
Paper • 2409.02060 • Published • 80 -
Training Language Models to Self-Correct via Reinforcement Learning
Paper • 2409.12917 • Published • 140 -
GRIN: GRadient-INformed MoE
Paper • 2409.12136 • Published • 16 -
Preference Tuning with Human Feedback on Language, Speech, and Vision Tasks: A Survey
Paper • 2409.11564 • Published • 20 -
NVLM: Open Frontier-Class Multimodal LLMs
Paper • 2409.11402 • Published • 74 -
Instruction Following without Instruction Tuning
Paper • 2409.14254 • Published • 29 -
Molmo and PixMo: Open Weights and Open Data for State-of-the-Art Multimodal Models
Paper • 2409.17146 • Published • 121 -
Programming Every Example: Lifting Pre-training Data Quality like Experts at Scale
Paper • 2409.17115 • Published • 64 -
Connecting the Dots: LLMs can Infer and Verbalize Latent Structure from Disparate Training Data
Paper • 2406.14546 • Published • 3 -
Thinking LLMs: General Instruction Following with Thought Generation
Paper • 2410.10630 • Published • 20 -
Phi-4 Technical Report
Paper • 2412.08905 • Published • 122 -
Offline Reinforcement Learning for LLM Multi-Step Reasoning
Paper • 2412.16145 • Published • 38 -
RobustFT: Robust Supervised Fine-tuning for Large Language Models under Noisy Response
Paper • 2412.14922 • Published • 88 -
Diving into Self-Evolving Training for Multimodal Reasoning
Paper • 2412.17451 • Published • 42 -
OpenAI o1 System Card
Paper • 2412.16720 • Published • 37 -
TÜLU 3: Pushing Frontiers in Open Language Model Post-Training
Paper • 2411.15124 • Published • 67 -
Natural Language Reinforcement Learning
Paper • 2411.14251 • Published • 31 -
OpenCoder: The Open Cookbook for Top-Tier Code Large Language Models
Paper • 2411.04905 • Published • 127 -
Mixture-of-Transformers: A Sparse and Scalable Architecture for Multi-Modal Foundation Models
Paper • 2411.04996 • Published • 50 -
2.5 Years in Class: A Multimodal Textbook for Vision-Language Pretraining
Paper • 2501.00958 • Published • 110 -
Towards System 2 Reasoning in LLMs: Learning How to Think With Meta Chain-of-Though
Paper • 2501.04682 • Published • 99 -
Scaling Laws for Floating Point Quantization Training
Paper • 2501.02423 • Published • 26 -
Virgo: A Preliminary Exploration on Reproducing o1-like MLLM
Paper • 2501.01904 • Published • 33 -
O1 Replication Journey -- Part 3: Inference-time Scaling for Medical Reasoning
Paper • 2501.06458 • Published • 31 -
Enhancing Human-Like Responses in Large Language Models
Paper • 2501.05032 • Published • 61 -
Do generative video models learn physical principles from watching videos?
Paper • 2501.09038 • Published • 34 -
DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning
Paper • 2501.12948 • Published • 444 -
Kimi k1.5: Scaling Reinforcement Learning with LLMs
Paper • 2501.12599 • Published • 128 -
Demystifying Long Chain-of-Thought Reasoning in LLMs
Paper • 2502.03373 • Published • 58 -
SFT Memorizes, RL Generalizes: A Comparative Study of Foundation Model Post-training
Paper • 2501.17161 • Published • 125 -
Test-Time Preference Optimization: On-the-Fly Alignment via Iterative Textual Feedback
Paper • 2501.12895 • Published • 61 -
Step-KTO: Optimizing Mathematical Reasoning through Stepwise Binary Feedback
Paper • 2501.10799 • Published • 15 -
Qwen2.5-1M Technical Report
Paper • 2501.15383 • Published • 72 -
Baichuan-Omni-1.5 Technical Report
Paper • 2501.15368 • Published • 60 -
Optimizing Large Language Model Training Using FP4 Quantization
Paper • 2501.17116 • Published • 36 -
Over-Tokenized Transformer: Vocabulary is Generally Worth Scaling
Paper • 2501.16975 • Published • 32 -
Critique Fine-Tuning: Learning to Critique is More Effective than Learning to Imitate
Paper • 2501.17703 • Published • 59 -
SmolLM2: When Smol Goes Big -- Data-Centric Training of a Small Language Model
Paper • 2502.02737 • Published • 257 -
LIMO: Less is More for Reasoning
Paper • 2502.03387 • Published • 62 -
QuEST: Stable Training of LLMs with 1-Bit Weights and Activations
Paper • 2502.05003 • Published • 44 -
CMoE: Fast Carving of Mixture-of-Experts for Efficient LLM Inference
Paper • 2502.04416 • Published • 12 -
Scaling Pre-training to One Hundred Billion Data for Vision Language Models
Paper • 2502.07617 • Published • 29 -
Gemstones: A Model Suite for Multi-Faceted Scaling Laws
Paper • 2502.06857 • Published • 24 -
Typhoon T1: An Open Thai Reasoning Model
Paper • 2502.09042 • Published • 16 -
Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse Attention
Paper • 2502.11089 • Published • 169 -
How Much Knowledge Can You Pack into a LoRA Adapter without Harming LLM?
Paper • 2502.14502 • Published • 92 -
Continuous Diffusion Model for Language Modeling
Paper • 2502.11564 • Published • 53 -
Train Small, Infer Large: Memory-Efficient LoRA Training for Large Language Models
Paper • 2502.13533 • Published • 13 -
LongRoPE2: Near-Lossless LLM Context Window Scaling
Paper • 2502.20082 • Published • 36 -
Stable-SPAM: How to Train in 4-Bit More Stably than 16-Bit Adam
Paper • 2502.17055 • Published • 20 -
Visual-RFT: Visual Reinforcement Fine-Tuning
Paper • 2503.01785 • Published • 86 -
Predictive Data Selection: The Data That Predicts Is the Data That Teaches
Paper • 2503.00808 • Published • 57 -
Gemini Robotics: Bringing AI into the Physical World
Paper • 2503.20020 • Published • 31 -
Large-Scale Data Selection for Instruction Tuning
Paper • 2503.01807 • Published • 14 -
Unified Reward Model for Multimodal Understanding and Generation
Paper • 2503.05236 • Published • 124 -
Light-R1: Curriculum SFT, DPO and RL for Long COT from Scratch and Beyond
Paper • 2503.10460 • Published • 30 -
TTRL: Test-Time Reinforcement Learning
Paper • 2504.16084 • Published • 122 -
Learning from Failures in Multi-Attempt Reinforcement Learning
Paper • 2503.04808 • Published • 18 -
TinyR1-32B-Preview: Boosting Accuracy with Branch-Merge Distillation
Paper • 2503.04872 • Published • 15 -
Self-Taught Self-Correction for Small Language Models
Paper • 2503.08681 • Published • 15 -
Cosmos-Reason1: From Physical Common Sense To Embodied Reasoning
Paper • 2503.15558 • Published • 50 -
SkyLadder: Better and Faster Pretraining via Context Window Scheduling
Paper • 2503.15450 • Published • 12 -
Gemma 3 Technical Report
Paper • 2503.19786 • Published • 55 -
Modifying Large Language Model Post-Training for Diverse Creative Writing
Paper • 2503.17126 • Published • 36 -
FastCuRL: Curriculum Reinforcement Learning with Progressive Context Extension for Efficient Training R1-like Reasoning Models
Paper • 2503.17287 • Published • 11 -
ZClip: Adaptive Spike Mitigation for LLM Pre-Training
Paper • 2504.02507 • Published • 88 -
RL Tango: Reinforcing Generator and Verifier Together for Language Reasoning
Paper • 2505.15034 • Published • 5 -
Improved Visual-Spatial Reasoning via R1-Zero-Like Training
Paper • 2504.00883 • Published • 67 -
Open-Reasoner-Zero: An Open Source Approach to Scaling Up Reinforcement Learning on the Base Model
Paper • 2503.24290 • Published • 62 -
JudgeLRM: Large Reasoning Models as a Judge
Paper • 2504.00050 • Published • 62 -
Inference-Time Scaling for Generalist Reward Modeling
Paper • 2504.02495 • Published • 58 -
Understanding R1-Zero-Like Training: A Critical Perspective
Paper • 2503.20783 • Published • 59 -
Exploring Data Scaling Trends and Effects in Reinforcement Learning from Human Feedback
Paper • 2503.22230 • Published • 45 -
Unicorn: Text-Only Data Synthesis for Vision Language Model Training
Paper • 2503.22655 • Published • 38 -
Open-Qwen2VL: Compute-Efficient Pre-Training of Fully-Open Multimodal LLMs on Academic Resources
Paper • 2504.00595 • Published • 37 -
Rethinking RL Scaling for Vision Language Models: A Transparent, From-Scratch Framework and Comprehensive Evaluation Scheme
Paper • 2504.02587 • Published • 32 -
Scaling Analysis of Interleaved Speech-Text Language Models
Paper • 2504.02398 • Published • 31 -
Scaling Language-Free Visual Representation Learning
Paper • 2504.01017 • Published • 33 -
RIG: Synergizing Reasoning and Imagination in End-to-End Generalist Policy
Paper • 2503.24388 • Published • 29 -
Z1: Efficient Test-time Scaling with Code
Paper • 2504.00810 • Published • 27 -
Command A: An Enterprise-Ready Large Language Model
Paper • 2504.00698 • Published • 29 -
Expanding RL with Verifiable Rewards Across Diverse Domains
Paper • 2503.23829 • Published • 24 -
ActionStudio: A Lightweight Framework for Data and Training of Large Action Models
Paper • 2503.22673 • Published • 12 -
Reasoning-SQL: Reinforcement Learning with SQL Tailored Partial Rewards for Reasoning-Enhanced Text-to-SQL
Paper • 2503.23157 • Published • 10 -
Kimi-VL Technical Report
Paper • 2504.07491 • Published • 138 -
Skywork R1V: Pioneering Multimodal Reasoning with Chain-of-Thought
Paper • 2504.05599 • Published • 86 -
Rethinking Reflection in Pre-Training
Paper • 2504.04022 • Published • 80 -
OLMoTrace: Tracing Language Model Outputs Back to Trillions of Training Tokens
Paper • 2504.07096 • Published • 77 -
Scaling Laws for Native Multimodal Models Scaling Laws for Native Multimodal Models
Paper • 2504.07951 • Published • 30 -
VAPO: Efficient and Reliable Reinforcement Learning for Advanced Reasoning Tasks
Paper • 2504.05118 • Published • 26 -
A Sober Look at Progress in Language Model Reasoning: Pitfalls and Paths to Reproducibility
Paper • 2504.07086 • Published • 21 -
SoTA with Less: MCTS-Guided Sample Selection for Data-Efficient Visual Reasoning Self-Improvement
Paper • 2504.07934 • Published • 21 -
VideoChat-R1: Enhancing Spatio-Temporal Perception via Reinforcement Fine-Tuning
Paper • 2504.06958 • Published • 13 -
Efficient Reinforcement Finetuning via Adaptive Curriculum Learning
Paper • 2504.05520 • Published • 11 -
InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Models
Paper • 2504.10479 • Published • 308 -
CLIMB: CLustering-based Iterative Data Mixture Bootstrapping for Language Model Pre-training
Paper • 2504.13161 • Published • 97 -
xVerify: Efficient Answer Verifier for Reasoning Model Evaluations
Paper • 2504.10481 • Published • 85 -
BitNet b1.58 2B4T Technical Report
Paper • 2504.12285 • Published • 83 -
Genius: A Generalizable and Purely Unsupervised Self-Training Framework For Advanced Reasoning
Paper • 2504.08672 • Published • 55 -
VL-Rethinker: Incentivizing Self-Reflection of Vision-Language Models with Reinforcement Learning
Paper • 2504.08837 • Published • 44 -
Heimdall: test-time scaling on the generative verification
Paper • 2504.10337 • Published • 33 -
SFT or RL? An Early Investigation into Training R1-Like Reasoning Large Vision-Language Models
Paper • 2504.11468 • Published • 30 -
NoisyRollout: Reinforcing Visual Reasoning with Data Augmentation
Paper • 2504.13055 • Published • 19 -
A Minimalist Approach to LLM Reasoning: from Rejection Sampling to Reinforce
Paper • 2504.11343 • Published • 20 -
DUMP: Automated Distribution-Level Curriculum Learning for RL-based LLM Post-training
Paper • 2504.09710 • Published • 19 -
DataDecide: How to Predict Best Pretraining Data with Small Experiments
Paper • 2504.11393 • Published • 18 -
Breaking the Data Barrier -- Building GUI Agents Through Task Generalization
Paper • 2504.10127 • Published • 17 -
M1: Towards Scalable Test-Time Compute with Mamba Reasoning Models
Paper • 2504.10449 • Published • 15 -
SimpleAR: Pushing the Frontier of Autoregressive Visual Generation through Pretraining, SFT, and RL
Paper • 2504.11455 • Published • 14 -
Efficient Process Reward Model Training via Active Learning
Paper • 2504.10559 • Published • 13 -
Exploring Expert Failures Improves LLM Agent Tuning
Paper • 2504.13145 • Published • 12 -
Eagle 2.5: Boosting Long-Context Post-Training for Frontier Vision-Language Models
Paper • 2504.15271 • Published • 68 -
ToolRL: Reward is All Tool Learning Needs
Paper • 2504.13958 • Published • 49 -
OTC: Optimal Tool Calls via Reinforcement Learning
Paper • 2504.14870 • Published • 35 -
QuaDMix: Quality-Diversity Balanced Data Selection for Efficient LLM Pretraining
Paper • 2504.16511 • Published • 22 -
LLMs are Greedy Agents: Effects of RL Fine-tuning on Decision-Making Abilities
Paper • 2504.16078 • Published • 21 -
Efficient Pretraining Length Scaling
Paper • 2504.14992 • Published • 20