| Large Language Diffusion Models | ✓ Link | 88.5 | LLaDA 8B Instruct | 2025-02-14 |
| Introducing LFM-7B: Setting New Standards for Efficient Language Models | | 70.56 | LFM-7B (25-shot) | 2025-01-20 |
| Liquid Foundation Models: Our First Series of Generative AI Models | | 67.24 | LFM-40B Preview (25-shot) | 2024-09-30 |
| PaLM: Scaling Language Modeling with Pathways | ✓ Link | 65.9 | PaLM 540B | 2022-04-05 |
| Improved Large Language Diffusion Models | ✓ Link | 60.8 | iLLaDA 8B Base | 2026-06-24 |
| Dream 7B: Diffusion Large Language Models | | 59.8 | Dream 7B | 2025-08-21 |
| Scheduling Recursive Reasoning in Looped Transformers | | 58.87 | Ouro-2.6B + TAPS (Momentum) | 2026-09-29 |
| PaLM: Scaling Language Modeling with Pathways | ✓ Link | 53.7 | PaLM 62B | 2022-04-05 |
| LatentLoop: Scaling Looped MoE Models with Asymmetric Compute Allocation | ✓ Link | 51.71 | LatentLoop (7B / 1B active) | 2026-10-14 |
| Language Models are Few-Shot Learners | ✓ Link | 51.5 | GPT-3 175B (few-shot) | 2020-05-28 |
| Language Models are Few-Shot Learners | ✓ Link | 51.4 | GPT-3 175B (zero-shot) | 2020-05-28 |
| Universal YOCO for Efficient Depth Scaling | | 48.72 | YOCO-U (Equal Steps) | 2026-04-01 |
| Scaling Laws for Looped Mixture of Experts | | 47.5 | A0.3B-1.3B LoopMoE (R=5) | 2026-09-30 |
| Large Language Diffusion Models | ✓ Link | 45.9 | LLaDA 8B Base | 2025-02-14 |
| Language Models are Few-Shot Learners | ✓ Link | 44.8 | GPT-3 13B (few-shot) | 2020-05-28 |
| Mamba-3: Improved Sequence Modeling using State Space Principles | ✓ Link | 44.5 | Mamba-3-MIMO-1.5B | 2026-03-16 |
| Language Models are Few-Shot Learners | ✓ Link | 43.7 | GPT-3 6.7B (few-shot) | 2020-05-28 |
| Language Models are Few-Shot Learners | ✓ Link | 43.7 | GPT-3 13B (zero-shot) | 2020-05-28 |
| Mamba-3: Improved Sequence Modeling using State Space Principles | ✓ Link | 42.7 | Mamba-3-SISO-1.5B | 2026-03-16 |
| Copy-Paste Language Models | ✓ Link | 41.55 | Copy-Paste Language Model (1.7B, Hybrid 3:1) | 2026-10-13 |
| Language Models are Few-Shot Learners | ✓ Link | 41.4 | GPT-3 6.7B (zero-shot) | 2020-05-28 |
| Mamba-3: Improved Sequence Modeling using State Space Principles | ✓ Link | 41.2 | Mamba-3-MIMO-880M | 2026-03-16 |
| Modular TTT: Rethinking Test-Time Training as Composable Modules | ✓ Link | 40.78 | Modular TTT 1.45B (inner-product M) | 2026-08-07 |
| Towards Looped Models Done Right, Part II: Rethinking at Fixed Points | ✓ Link | 40.44 | LoopedLM-P2 learned-depth lambda_H=0.01 (1.6B) | 2026-10-05 |
| PaLM: Scaling Language Modeling with Pathways | ✓ Link | 40.2 | PaLM 8B | 2022-04-05 |
| Mamba-3: Improved Sequence Modeling using State Space Principles | ✓ Link | 40.2 | Mamba-3-SISO-880M | 2026-03-16 |
| Distributionally Robust Mixture-of-Experts Training | ✓ Link | 38.05 | DRMoET (1.7B active, 10.3B total, eta=0.01) | 2026-10-05 |
| Language Models are Few-Shot Learners | ✓ Link | 36.7 | GPT-3 XL (few-shot) | 2020-05-28 |
| Language Models are Few-Shot Learners | ✓ Link | 35.5 | GPT-3 XL (zero-shot) | 2020-05-28 |
| Mamba-3: Improved Sequence Modeling using State Space Principles | ✓ Link | 34.4 | Mamba-3-SISO-440M | 2026-03-16 |
| Grouped Value Attention: Efficient KV Caching via On-Demand Key Reconstruction | | 28.32 | GVA + DRoPE d_r=16 | 2026-09-08 |
| Mamba-3: Improved Sequence Modeling using State Space Principles | ✓ Link | 27.9 | Mamba-3-SISO-180M | 2026-03-16 |
| Mamba-3: Improved Sequence Modeling using State Space Principles | ✓ Link | 27.7 | Mamba-3-MIMO-180M | 2026-03-16 |
| Language Models are Few-Shot Learners | ✓ Link | 26.6 | GPT-3 Small (zero-shot) | 2020-05-28 |
| Language Models are Few-Shot Learners | ✓ Link | 25.5 | GPT-3 Small (few-shot) | 2020-05-28 |
| Gated Recurrent Transformers: Expressive Depth through Recurrent Modulation | ✓ Link | 25.43 | GRT Large (isoFLOPs, zero-shot) | 2026-08-15 |
| WhiteMatter: All-to-All Cross-Layer Connections via KV Source Mixing | ✓ Link | 25.43 | WhiteMatter k=16 | 2026-09-27 |