# CHERRY: Compressed Hierarchical Experts with Recurrent Representational Yield

arXiv: https://arxiv.org/abs/2606.31796 · HTML: https://mlsanity.com/paper/2606.31796/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 31 of its 48 references that are in the map.
- First posted on arXiv 30 Jun 2026.

## Builds on (most cited first, 30 of 31)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://mlsanity.com/paper/1910.10683/) (2019)
- [Distilling the Knowledge in a Neural Network](https://mlsanity.com/paper/1503.02531/) (2015)
- [Scaling Laws for Neural Language Models](https://mlsanity.com/paper/2001.08361/) (2020)
- [DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning](https://mlsanity.com/paper/2501.12948/) (2025)
- [Mamba: Linear-Time Sequence Modeling with Selective State Spaces](https://mlsanity.com/paper/2312.00752/) (2023)
- [Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layer](https://mlsanity.com/paper/1701.06538/) (2017)
- [Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity](https://mlsanity.com/paper/2101.03961/) (2021)
- [Neural Machine Translation of Rare Words with Subword Units](https://mlsanity.com/paper/1508.07909/) (2015)
- [ALBERT: A Lite BERT for Self-supervised Learning of Language Representations](https://mlsanity.com/paper/1909.11942/) (2019)
- [Neural Tangent Kernel: Convergence and Generalization in Neural Networks](https://mlsanity.com/paper/1806.07572/) (2018)
- [Fast Inference from Transformers via Speculative Decoding](https://mlsanity.com/paper/2211.17192/) (2022)
- [Model soups: averaging weights of multiple fine-tuned models improves accuracy without increasing inference time](https://mlsanity.com/paper/2203.05482/) (2022)
- [Gradient Surgery for Multi-Task Learning](https://mlsanity.com/paper/2001.06782/) (2020)
- [TinyBERT: Distilling BERT for Natural Language Understanding](https://mlsanity.com/paper/1909.10351/) (2019)
- [Accelerating Large Language Model Decoding with Speculative Sampling](https://mlsanity.com/paper/2302.01318/) (2023)
- [RWKV: Reinventing RNNs for the Transformer Era](https://mlsanity.com/paper/2305.13048/) (2023)
- [LLM-Pruner: On the Structural Pruning of Large Language Models](https://mlsanity.com/paper/2305.11627/) (2023)
- [Universal Transformers](https://mlsanity.com/paper/1807.03819/) (2018)
- [Scaling up Test-Time Compute with Latent Reasoning: A Recurrent Depth Approach](https://mlsanity.com/paper/2502.05171/) (2025)
- [ShortGPT: Layers in Large Language Models are More Redundant Than You Expect](https://mlsanity.com/paper/2403.03853/) (2024)
- [Better & Faster Large Language Models via Multi-token Prediction](https://mlsanity.com/paper/2404.19737/) (2024)
- [Not All Samples Are Created Equal: Deep Learning with Importance Sampling](https://mlsanity.com/paper/1803.00942/) (2018)
- [MiniLMv2: Multi-Head Self-Attention Relation Distillation for Compressing Pretrained Transformers](https://mlsanity.com/paper/2012.15828/) (2020)
- [Prioritized Training on Points that are Learnable, Worth Learning, and Not Yet Learnt](https://mlsanity.com/paper/2206.07137/) (2022)
- [The Unreasonable Ineffectiveness of the Deeper Layers](https://mlsanity.com/paper/2403.17887/) (2024)
- [MobileLLM: Optimizing Sub-billion Parameter Language Models for On-Device Use Cases](https://mlsanity.com/paper/2402.14905/) (2024)
- [Branch-Train-Merge: Embarrassingly Parallel Training of Expert Language Models](https://mlsanity.com/paper/2208.03306/) (2022)
- [SOLAR 10.7B: Scaling Large Language Models with Simple yet Effective Depth Up-Scaling](https://mlsanity.com/paper/2312.15166/) (2023)
- [Accelerating Deep Learning by Focusing on the Biggest Losers](https://mlsanity.com/paper/1910.00762/) (2019)

## Built on by (most cited first, 1 of 1)

- [Loop the Loopies!](https://mlsanity.com/paper/2607.16051/) (2026)

## Abstract (arXiv)

Frontier language capability is usually bought with frontier compute; CHERRY shows a different trade. It is a sovereign Korean model family built on one principle: supervise the tokens that decide the answer, and let shared weights carry the rest. Under matched compute this exposes a sharp, reproducible dissociation---selected-token supervision preserves held-out discrimination yet collapses free generation, and a full-sequence anchor recovers only part of the gap. The same signal drives a heal-after-merge recurrent-representational-yield loop that collapses 48 layers to 6 unique blocks at near-dense parity (227M at loss 2.934 vs a 566M dense model at 2.926) and composes them by MoEE fusion (2.789)---a recurrent-compression direction independently pursued by concurrent frontier looped-MoE work, which we project (not yet measure) to frontier scale. It also installs metacognition from two-token supervision (200 held-out KO prompts/type, kappa>0.82, +/-6.9pp): self-correction 12->47% and jailbreak 23->4% at 97.6% loss-retention on 1.2B, with a pre-registered 1B->13.7B ablation localising the operand-binding limit to capacity (1B lookup vs 13.7B H-PRESERVE); and it specializes a 1.8B model to reported human-expert level on CyberMetric (75.0% vs a 30-expert average 72.24%). The released 1.8B tokenizer is 9.2% more Korean-efficient than Gemma-4; the from-scratch 12B adds a sovereign Korean tokenizer (vocab 131,037). On the government-operated K-AI Korean-LLM leaderboard the released 1.8B leads the HLE(Ko) column (0.123 vs 0.077; overall 51/78). At the frontier we fine-tune and serve an adapted 122B model on one 120GB accelerator (83GB measured peak). Provenance differs by member and we state it exactly; the architecture and recipe are ours throughout, and every claim is bound to a released curated measurement.