| Atlas: Few-shot Learning with Retrieval Augmented Language Models | ✓ Link | 64.0 | Atlas (full, Wiki-dec-2018 index) | 2022-08-05 |
| Atlas: Few-shot Learning with Retrieval Augmented Language Models | ✓ Link | 60.4 | Atlas (full, Wiki-dec-2021+CC index) | 2022-08-05 |
| Understand What LLM Needs: Dual Preference Alignment for Retrieval-Augmented Generation | ✓ Link | 59.19 | DPA-RAG | 2024-06-26 |
| FiE: Building a Global Probability Space by Leveraging Early Fusion in Encoder for Open-Domain Question Answering | | 58.4 | FiE | 2022-11-18 |
| R2-D2: A Modular Baseline for Open-Domain Question Answering | ✓ Link | 55.9 | R2-D2 (full) | 2021-09-08 |
| WHALE: A Simple Recipe for Joint Harness-Weight Optimization | ✓ Link | 55.5 | Qwen3.5-2B + WHALE (E0.6,I6) | 2026-08-31 |
| Leveraging Passage Retrieval with Generative Models for Open Domain Question Answering | ✓ Link | 54.7 | FiD-KD (full) | 2020-07-02 |
| Retrieval as Attention: End-to-end Learning of Retrieval and Reading within a Single Transformer | ✓ Link | 54.7 | ReAtt | 2022-12-05 |
| RankRAG: Unifying Context Ranking with Retrieval-Augmented Generation in LLMs | | 54.2 | RankRAG-llama3-70b (Zero-Shot, KILT) | 2024-07-02 |
| DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model | ✓ Link | 53.4 | DeepSeek-V2 Chat (RL) (5-shot) | 2024-05-07 |
| AT^2PO: Agentic Turn-based Policy Optimization via Tree Search | ✓ Link | 52.91 | AT2PO (Qwen2.5-7B) | 2026-01-08 |
| End-to-End Training of Multi-Document Reader and Retriever for Open-Domain Question Answering | ✓ Link | 52.5 | EMDR^2 | 2021-06-09 |
| Leveraging Passage Retrieval with Generative Models for Open Domain Question Answering | ✓ Link | 51.4 | FID (full) | 2020-07-02 |
| RankRAG: Unifying Context Ranking with Retrieval-Augmented Generation in LLMs | | 50.6 | RankRAG-llama3-8b (Zero-Shot, KILT) | 2024-07-02 |
| Meta-Reinforcement Learning with Self-Reflection for Agentic Search | ✓ Link | 50.2 | MR-Search (Qwen2.5-7B-Base) | 2026-03-11 |
| RankRAG: Unifying Context Ranking with Retrieval-Augmented Generation in LLMs | | 50.0 | RankRAG-llama3-70b (Zero-Shot, DPR) | 2024-07-02 |
| DeepSeek LLM: Scaling Open-Source Language Models with Longtermism | ✓ Link | 47.0 | DeepSeek LLM 67B Chat | 2024-01-05 |
| ChatQA: Surpassing GPT-4 on Conversational QA and RAG | | 47.0 | ChatQA-1.5-llama3-70b (Zero-Shot, KILT) | 2024-01-18 |
| SEED: Self-Evolving On-Policy Distillation for Agentic Reinforcement Learning | ✓ Link | 47.0 | SEED (Qwen2.5-7B-Instruct) | 2026-07-16 |
| EVOKE: Eliciting World Knowledge in Agents for Transferable Decision-Making | ✓ Link | 46.8 | EVOKE (Qwen2.5-7B-Instruct) | 2026-09-29 |
| Shallow Prefill, Deep Decoding: Efficient Long-Context Inference via Layer-Asymmetric KV Visibility | | 46.4 | OffShelf-FT-SPEED+BoS-24 | 2026-05-07 |
| T^2PO: Uncertainty-Guided Exploration Control for Stable Multi-Turn Agentic Reinforcement Learning | ✓ Link | 46.13 | T2PO (Qwen3-4B-RFT) | 2026-05-04 |
| RankRAG: Unifying Context Ranking with Retrieval-Augmented Generation in LLMs | | 46.1 | RankRAG-llama3-8b (Zero-Shot, DPR) | 2024-07-02 |
| Jamba: A Hybrid Transformer-Mamba Language Model | ✓ Link | 45.9 | Jamba (5-shot) | 2024-03-28 |
| Improving language models by retrieving from trillions of tokens | ✓ Link | 45.5 | RETRO + DPR (full) | 2021-12-08 |
| REPLUG: Retrieval-Augmented Black-Box Language Models | ✓ Link | 45.5 | code-davinci-002 175B + REPLUG LSR (few-shot) | 2023-01-30 |
| Atlas: Orchestrating Heterogeneous Models and Tools for Multi-Domain Complex Reasoning | | 45.3 | Atlas (RL) (pool extension) | 2026-01-07 |
| Atlas: Few-shot Learning with Retrieval Augmented Language Models | ✓ Link | 45.1 | Atlas (few-shot, k=64, Wiki-Dec-2018 index) | 2022-08-05 |
| REPLUG: Retrieval-Augmented Black-Box Language Models | ✓ Link | 44.7 | code-davinci-002 175B + REPLUG (few-shot) | 2023-01-30 |
| Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks | ✓ Link | 44.5 | RAG | 2020-05-22 |
| Encoder-Decoder Gemma: Improving the Quality-Efficiency Trade-Off via Adaptation | ✓ Link | 43.1 | T5Gemma 9B-9B PrefixLM | 2025-04-08 |
| ChatQA: Surpassing GPT-4 on Conversational QA and RAG | | 42.7 | ChatQA-1.5-llama3-8b (Zero-Shot, KILT) | 2024-01-18 |
| Blended RAG: Improving RAG (Retriever-Augmented Generation) Accuracy with Semantic Search and Hybrid Query-Based Retrievers | ✓ Link | 42.63 | Blended RAG | 2024-03-22 |
| Agentic-R: Learning to Retrieve for Agentic Search | ✓ Link | 42.43 | Agentic-R | 2026-01-17 |
| Atlas: Few-shot Learning with Retrieval Augmented Language Models | ✓ Link | 42.4 | Atlas (few-shot, k=64, Wiki-dec-2021+CC index) | 2022-08-05 |
| Demystifying Reinforcement Learning for Long-Horizon Tool-Using Agents: A Comprehensive Recipe | ✓ Link | 42.2 | Qwen2.5-Instruct (7B) + STAR (Macro reward) | 2026-03-23 |
| Dr. Zero: Self-Evolving Search Agents without Training Data | ✓ Link | 41.6 | Dr. Zero (Qwen2.5-7B-Instruct) | 2026-01-11 |
| Dense Passage Retrieval for Open-Domain Question Answering | ✓ Link | 41.5 | DPR | 2020-04-10 |
| REALM: Retrieval-Augmented Language Model Pre-Training | ✓ Link | 40.4 | REALM | 2020-02-10 |
| LLaMA: Open and Efficient Foundation Language Models | ✓ Link | 39.9 | LLaMA 65B (few-shot, k=64) | 2023-02-27 |
| PaLM: Scaling Language Modeling with Pathways | ✓ Link | 39.6 | PaLM-540B (Few-Shot, k=64) | 2022-04-05 |
| InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Models | ✓ Link | 39.0 | InternVL3-78B | 2025-04-14 |
| PaLM 2 Technical Report | ✓ Link | 37.5 | PaLM 2-L (one-shot) | 2023-05-17 |
| AdaTutoRank: Learning to Rerank Document Sets via Adaptive Tutoring Optimization for RAG and Deep Research | ✓ Link | 36.87 | AdaTutoRank (8B) | 2026-09-26 |
| Training Compute-Optimal Large Language Models | ✓ Link | 35.5 | Chinchilla (few-shot, k=64) | 2022-03-29 |
| LLaMA: Open and Efficient Foundation Language Models | ✓ Link | 35.0 | LLaMA 65B (few-shot, k=5) | 2023-02-27 |
| Gemma 2: Improving Open Language Models at a Practical Size | ✓ Link | 34.5 | Gemma 2 27B (5-shot) | 2024-07-31 |
| Search-o1: Agentic Search-Enhanced Large Reasoning Models | ✓ Link | 34 | Search-o1 | 2025-01-09 |
| Llama 2: Open Foundation and Fine-Tuned Chat Models | ✓ Link | 33.0 | LLaMA 2 70B (one-shot) | 2023-07-18 |
| GLaM: Efficient Scaling of Language Models with Mixture-of-Experts | | 32.5 | GLaM 62B/64E (Few-Shot) | 2021-12-13 |
| PaLM 2 Technical Report | ✓ Link | 32.0 | PaLM 2-M (one-shot) | 2023-05-17 |
| LLaMA: Open and Efficient Foundation Language Models | ✓ Link | 31.0 | LLaMA 65B (one-shot) | 2023-02-27 |
| Mixtral of Experts | ✓ Link | 30.6 | Mixtral 8x7B (5-shot) | 2024-01-08 |
| Language Models are Few-Shot Learners | ✓ Link | 29.9 | GPT-3 175B (Few-Shot, k=64) | 2020-05-28 |
| PaLM: Scaling Language Modeling with Pathways | ✓ Link | 29.3 | PaLM-540B (One-Shot) | 2022-04-05 |
| olmOCR: Unlocking Trillions of Tokens in PDFs with Vision Language Models | ✓ Link | 29.1 | OLMo-2-7B-1124 (olmOCR-peS2o) | 2025-02-25 |
| Mistral 7B | ✓ Link | 28.8 | Mistral 7B (5-shot) | 2023-10-10 |
| Scaling Language Models: Methods, Analysis & Insights from Training Gopher | ✓ Link | 28.2 | Gopher (few-shot, k=64) | 2021-12-08 |
| Gated DeltaNet-2: Decoupling Erase and Write in Linear Attention | ✓ Link | 26.31 | Gated DeltaNet-2 (1.3B) | 2026-05-21 |
| GLaM: Efficient Scaling of Language Models with Mixture-of-Experts | | 26.3 | GLaM 62B/64E (One-Shot) | 2021-12-13 |
| | 26.07 | LLaMA 7B (Contriever) | |
| PaLM 2 Technical Report | ✓ Link | 25.3 | PaLM 2-S (one-shot) | 2023-05-17 |
| Knowledge Distillation During Mid-Training Favors Reasoning over Factual Recall | ✓ Link | 25.3 | Switch Distillation (1B; 7B teacher, q30% mid-training) | 2026-09-01 |
| LLaMA: Open and Efficient Foundation Language Models | ✓ Link | 24.9 | LLaMA 33B (zero-shot) | 2023-02-27 |
| Massive Activations in Hybrid Linear Attention Large Language Models: Pre-Attention Spikes and Inter-Spike Plateaus | ✓ Link | 24.9 | GDN 1.3B (3:1 hybrid, 50B tokens) | 2026-08-12 |
| Gated Delta Networks: Improving Mamba2 with Delta Rule | ✓ Link | 24.8 | Gated DeltaNet-H2 | 2024-12-09 |
| GLaM: Efficient Scaling of Language Models with Mixture-of-Experts | | 24.7 | GLaM 62B/64E (Zero-Shot) | 2021-12-13 |
| Kalman Delta Networks: Uncertainty-aware Associative Memory | ✓ Link | 23.63 | Diagonal KDN (1.3B) | 2026-09-07 |
| Moshi: a speech-text foundation model for real-time dialogue | ✓ Link | 23.3 | Helium | 2024-09-17 |
| Gemma: Open Models Based on Gemini Research and Technology | ✓ Link | 23.0 | Gemma 7B (5-shot) | 2024-03-13 |
| RecurrentGemma: Moving Past Transformers for Efficient Open Language Models | ✓ Link | 21.7 | RecurrentGemma-9B (5-shot) | 2024-04-11 |
| PaLM: Scaling Language Modeling with Pathways | ✓ Link | 21.2 | PaLM-540B (Zero-Shot) | 2022-04-05 |
| Ask Me Anything: A simple strategy for prompting language models | ✓ Link | 19.7 | Neo-6B (QA) | 2022-10-05 |
| Ask Me Anything: A simple strategy for prompting language models | ✓ Link | 19.6 | Neo-6B (QA + WS) | 2022-10-05 |
| T5Gemma 2: Seeing, Reading, and Understanding Longer | | 17.2 | T5Gemma 2 4B-4B | 2025-12-16 |
| Ask Me Anything: A simple strategy for prompting language models | ✓ Link | 13.7 | Neo-6B (Few-Shot) | 2022-10-05 |
| Dynamic Linear Attention | | 12.1 | Gated DeltaNet-1.3B w/ DLA | 2026-06-09 |
| RecurrentGemma: Moving Past Transformers for Efficient Open Language Models | ✓ Link | 11.5 | RecurrentGemma-2B (5-shot) | 2024-04-11 |
| Scaling Laws for Looped Mixture of Experts | | 9.9 | A0.3B-1.3B LoopMoE (R=5) | 2026-09-30 |
| SmolLM2: When Smol Goes Big -- Data-Centric Training of a Small Language Model | ✓ Link | 8.7 | SmolLM2 1.7B | 2025-02-04 |
| Language Models are Unsupervised Multitask Learners | ✓ Link | 4.1 | GPT-2 | 2019-02-14 |