Reinforced Attention Learning
Paper
• 2602.04884 • Published
• 28
Papers reimplemented
List of research papers, architectures, and techniques I re implemented in LLM-quest or Hugging Face's TRL. Missing papers: Qwen3-Next, GPT-2
-
-
Learning to Reason in 13 Parameters
Paper • 2602.04118 • Published • 6 -
LoRA-XS: Low-Rank Adaptation with Extremely Small Number of Parameters
Paper • 2405.17604 • Published • 3 -
mHC-lite: You Don't Need 20 Sinkhorn-Knopp Iterations
Paper • 2601.05732 • Published • 1 -
mHC: Manifold-Constrained Hyper-Connections
Paper • 2512.24880 • Published • 312 -
Hyper-Connections
Paper • 2409.19606 • Published • 26 -
MiMo-V2-Flash Technical Report
Paper • 2601.02780 • Published • 35 -
NVIDIA Nemotron 3: Efficient and Open Intelligence
Paper • 2512.20856 • Published • 37 -
DeepSeek-V3.2: Pushing the Frontier of Open Large Language Models
Paper • 2512.02556 • Published • 258 -
Stabilizing MoE Reinforcement Learning by Aligning Training and Inference Routers
Paper • 2510.11370 • Published • 4 -
Soft Adaptive Policy Optimization
Paper • 2511.20347 • Published • 42 -
Kimi K2: Open Agentic Intelligence
Paper • 2507.20534 • Published • 13 -
Reinforcement Pre-Training
Paper • 2506.08007 • Published • 263 -
Min P Sampling: Balancing Creativity and Coherence at High Temperature
Paper • 2407.01082 • Published • 1 -
Gated Attention for Large Language Models: Non-linearity, Sparsity, and Attention-Sink-Free
Paper • 2505.06708 • Published • 11 -
Gated Delta Networks: Improving Mamba2 with Delta Rule
Paper • 2412.06464 • Published • 15 -
Approximating Two-Layer Feedforward Networks for Efficient Transformers
Paper • 2310.10837 • Published • 11 -
Transformers are RNNs: Fast Autoregressive Transformers with Linear Attention
Paper • 2006.16236 • Published • 4 -
Qwen3 Technical Report
Paper • 2505.09388 • Published • 336 -
Fast Inference from Transformers via Speculative Decoding
Paper • 2211.17192 • Published • 11 -
Regress, Don't Guess -- A Regression-like Loss on Number Tokens for Language Models
Paper • 2411.02083 • Published • 2 -
Group Sequence Policy Optimization
Paper • 2507.18071 • Published • 318 -
DAPO: An Open-Source LLM Reinforcement Learning System at Scale
Paper • 2503.14476 • Published • 144 -
Understanding R1-Zero-Like Training: A Critical Perspective
Paper • 2503.20783 • Published • 59 -
LoRA: Low-Rank Adaptation of Large Language Models
Paper • 2106.09685 • Published • 58 -
The Curious Case of Neural Text Degeneration
Paper • 1904.09751 • Published • 3 -
Hierarchical Neural Story Generation
Paper • 1805.04833 • Published -
Transformers without Normalization
Paper • 2503.10622 • Published • 170 -
An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
Paper • 2010.11929 • Published • 15 -
DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
Paper • 2402.03300 • Published • 141 -
Math-Shepherd: Verify and Reinforce LLMs Step-by-step without Human Annotations
Paper • 2312.08935 • Published • 4 -
DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning
Paper • 2501.12948 • Published • 440 -
Direct Preference Optimization: Your Language Model is Secretly a Reward Model
Paper • 2305.18290 • Published • 64 -
DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model
Paper • 2405.04434 • Published • 25 -
DeepSeek-V3 Technical Report
Paper • 2412.19437 • Published • 76 -
DeepSeekMoE: Towards Ultimate Expert Specialization in Mixture-of-Experts Language Models
Paper • 2401.06066 • Published • 59 -
Auxiliary-Loss-Free Load Balancing Strategy for Mixture-of-Experts
Paper • 2408.15664 • Published • 15 -
Instruction Tuning With Loss Over Instructions
Paper • 2405.14394 • Published • 1 -
Mixtral of Experts
Paper • 2401.04088 • Published • 160 -
ST-MoE: Designing Stable and Transferable Sparse Expert Models
Paper • 2202.08906 • Published • 3 -
Gemma 3 Technical Report
Paper • 2503.19786 • Published • 55 -
Scaling Vision Transformers to 22 Billion Parameters
Paper • 2302.05442 • Published • 2 -
GLU Variants Improve Transformer
Paper • 2002.05202 • Published • 4 -
Gemma 2: Improving Open Language Models at a Practical Size
Paper • 2408.00118 • Published • 78 -
Effective Approaches to Attention-based Neural Machine Translation
Paper • 1508.04025 • Published • 3 -
Longformer: The Long-Document Transformer
Paper • 2004.05150 • Published • 4 -
The Llama 3 Herd of Models
Paper • 2407.21783 • Published • 117 -
Root Mean Square Layer Normalization
Paper • 1910.07467 • Published • 2 -
Sigmoid-Weighted Linear Units for Neural Network Function Approximation in Reinforcement Learning
Paper • 1702.03118 • Published -
Language Modeling with Gated Convolutional Networks
Paper • 1612.08083 • Published -
Searching for Activation Functions
Paper • 1710.05941 • Published • 1 -
GQA: Training Generalized Multi-Query Transformer Models from Multi-Head Checkpoints
Paper • 2305.13245 • Published • 6 -
RoFormer: Enhanced Transformer with Rotary Position Embedding
Paper • 2104.09864 • Published • 17 -
YaRN: Efficient Context Window Extension of Large Language Models
Paper • 2309.00071 • Published • 80 -
Layer Normalization
Paper • 1607.06450 • Published • 4 -
Gaussian Error Linear Units (GELUs)
Paper • 1606.08415 • Published -
Attention Is All You Need
Paper • 1706.03762 • Published • 115 -
Language Models are Few-Shot Learners
Paper • 2005.14165 • Published • 19