arXiv subjects

Lucian Li

Publications and source records attributed to Lucian Li.

8 recordsLinked to original sources

arXiv2026

Macaron-V1: Towards Open Continual Learning with Self-Improvement and Mixture-of-LoRA

Macaron-V1 is an open agent-model family for experiential intelligence: learning from experience in real environments and continuing to learn after deployment. It is organized around two system goals. Adaptation is pursued through recursive improvement of versioned model-harness pairs, where experience from one configuration is evaluated under an external contract and used to construct its successor. Collaboration is pursued via the Mixture-of-LoRA (MoL) architecture that freezes a base model, composes specialist LoRA adapters, and selects one LoRA per user turn. The flagship Macaron-V1-Venti (748B) combines a 744B GLM-5.2 base with four LoRAs for chat, agent, coding, and GenUI; the Qwen3.6-35B-based Macaron-V1-Tall (50B) uses the same design for local deployment. This report presents Macaron-V1 as a co-designed system spanning architecture, algorithms, and infrastructure. The MoL architecture supports continual learning through extensible LoRA specialists. The algorithm combines Model-Harness Co-design and recursive self-improvement loop, including the UI4A component-native GenUI harness, a stateful action substrate, versioned Harness Context Protocol contract, and the agentic RL framework MindForge. The supporting infrastructure includes the post-training platform MinT, the long-context RL method LongStraw, and stability techniques for sparse MoE and DSA base models. We evaluate Macaron-V1 on Personal Intelligence, GenUI, and general capability benchmarks against frontier baselines. Our results validate the current system, while compounding gains from continual learning and collective intelligence remain open questions.

cs.LG↗

arXiv2026

LongStraw: Long-Context RL Beyond 2M Tokens under a Fixed GPU Budget

Long-context RL post-training is constrained by the lifetime of state and gradients, not attention cost alone. In GRPO, one multi-million-token prompt must serve old-policy and reference scoring plus multiple policy responses, while conventional autograd keeps the prompt graph and all response graphs live alongside model weights, caches, and distributed communication buffers. We present LongStraw, an objective-aware, architecture-aware system for resident-state virtualization, response replay, and distributed-gradient execution. Its transaction captures the shared prompt without autograd, retains only the architecture-required state on explicitly owned pages, restores that state for each group member, scores old/reference branches without a graph, replays one policy response at a time with autograd, and accumulates the resulting gradients before one distributed finalization and optimizer step. This schedule bounds the live training graph by the response suffix while reusing the expensive prompt computation across the complete GRPO group. We instantiate this design for two incompatible model structures. Qwen3.6-27B combines 48 recurrent GDN layers with 16 full-attention layers; LongStraw keeps the compact recurrent state and physically CP8-sharded KV pages, composes global attention through cross-rank LSE/output merging, and performs blockwise response replay. GLM-5.2 combines a 78-layer MLA/DSA attention stack with a 256-expert, top-8 MoE tail. Its implementation keeps CP-sharded MLA latent pages and DSA indexer-key pages in CPU memory, stages one layer at a time, reconstructs IndexShare-aware global sparse selection over CP32, and dispatches routed response tokens over EP32. The two paths share one transaction contract while specializing the retained state, replay operator, and collective communication to the architecture...

cs.LG↗

arXiv2026

On the Scaling of PEFT: Towards Million Personal Models of Trillion Parameters

Parameter-efficient fine-tuning (PEFT) is usually treated as a cheaper alternative to full fine-tuning. We study a broader role: small trainable adapters as persistent local state on top of strong shared foundation models. In this framing, the base model provides shared competence while adapters carry instance-specific behavior such as preferences, skills, tool habits, and memory-like updates. We organize the problem around three scaling axes: Scale Up, where stronger shared priors make small local updates more useful; Scale Down, where we study how small adapters can be while remaining reliable; and Scale Out, where many persistent adapted instances coexist. MinT provides one infrastructure example for managing adapter identity, revision, provenance, evaluation, and serving residency. Together, the results suggest that PEFT can be a compact substrate for persistent personal models rather than only a budget substitute for full fine-tuning.

cs.LG↗

arXiv2026

MinT: Managed Infrastructure for Training and Serving Millions of LLMs

We present MindLab Toolkit (MinT), a managed infrastructure system for Low-Rank Adaptation (LoRA) post-training and online serving. MinT targets a setting where many trained policies are produced over a small number of expensive base-model deployments. Instead of materializing each policy as a merged full checkpoint, MinT keeps the base model resident and moves exported LoRA adapter revisions through rollout, update, export, evaluation, serving, and rollback, hiding distributed training, serving, scheduling, and data movement behind a service interface. MinT scales this path along three axes. Scale Up extends LoRA RL to frontier-scale dense and MoE architectures, including MLA and DSA attention paths, with training and serving validated beyond 1T total parameters. Scale Down moves only the exported LoRA adapter, which can be under 1% of base-model size in rank-1 settings; adapter-only handoff reduces the measured step by 18.3x on a 4B dense model and 2.85x on a 30B MoE, while concurrent multi-policy GRPO shortens wall time by 1.77x and 1.45x without raising peak memory. Scale Out separates durable policy addressability from CPU/GPU working sets: a tensor-parallel deployment supports 10^6-scale addressable catalogs (measured single-engine sweeps through 100K) and thousand-adapter active waves at cluster scale, with cold loading treated as scheduled service work and packed MoE LoRA tensors improving live engine loading by 8.5-8.7x. MinT thus manages million-scale LoRA policy catalogs while training and serving selected adapter revisions over shared 1T-class base models.

cs.LG↗

arXiv2024

Locating the Leading Edge of Cultural Change

Measures of textual similarity and divergence are increasingly used to study cultural change. But which measures align, in practice, with social evidence about change? We apply three different representations of text (topic models, document embeddings, and word-level perplexity) to three different corpora (literary studies, economics, and fiction). In every case, works by highly-cited authors and younger authors are textually ahead of the curve. We don't find clear evidence that one representation of text is to be preferred over the others. But alignment with social evidence is strongest when texts are represented through the top quartile of passages, suggesting that a text's impact may depend more on its most forward-looking moments than on sustaining a high level of innovation throughout.

cs.CL↗

arXiv2024

Detecting text level intellectual influence with knowledge graph embeddings

Introduction: Tracing the spread of ideas and the presence of influence is a question of special importance across a wide range of disciplines, ranging from intellectual history to cultural analytics, computational social science, and the science of science. Method: We collect a corpus of open source journal articles, generate Knowledge Graph representations using the Gemini LLM, and attempt to predict the existence of citations between sampled pairs of articles using previously published methods and a novel Graph Neural Network based embedding model. Results: We demonstrate that our knowledge graph embedding method is superior at distinguishing pairs of articles with and without citation. Once trained, it runs efficiently and can be fine-tuned on specific corpora to suit individual researcher needs. Conclusion(s): This experiment demonstrates that the relationships encoded in a knowledge graph, especially the types of concepts brought together by specific relations can encode information capable of revealing intellectual influence. This suggests that further work in analyzing document level knowledge graphs to understand latent structures could provide valuable insights.

cs.CL↗

arXiv2024

Handwriting Recognition in Historical Documents with Multimodal LLM

There is an immense quantity of historical and cultural documentation that exists only as handwritten manuscripts. At the same time, performing OCR across scripts and different handwriting styles has proven to be an enormously difficult problem relative to the process of digitizing print. While recent Transformer based models have achieved relatively strong performance, they rely heavily on manually transcribed training data and have difficulty generalizing across writers. Multimodal LLM, such as GPT-4v and Gemini, have demonstrated effectiveness in performing OCR and computer vision tasks with few shot prompting. In this paper, I evaluate the accuracy of handwritten document transcriptions generated by Gemini against the current state of the art Transformer based methods. Keywords: Optical Character Recognition, Multimodal Language Models, Cultural Preservation, Mass digitization, Handwriting Recognitio

cs.CV↗

arXiv2024

Tracing the Genealogies of Ideas with Large Language Model Embeddings

In this paper, I present a novel method to detect intellectual influence across a large corpus. Taking advantage of the unique affordances of large language models in encoding semantic and structural meaning while remaining robust to paraphrasing, we can search for substantively similar ideas and hints of intellectual influence in a computationally efficient manner. Such a method allows us to operationalize different levels of confidence: we can allow for direct quotation, paraphrase, or speculative similarity while remaining open about the limitations of each threshold. I apply an ensemble method combining General Text Embeddings, a state-of-the-art sentence embedding method optimized to capture semantic content and an Abstract Meaning Representation graph representation designed to capture structural similarities in argumentation style and the use of metaphor. I apply this method to vectorize sentences from a corpus of roughly 400,000 nonfiction books and academic publications from the 19th century for instances of ideas and arguments appearing in Darwin's publications. This functions as an initial evaluation and proof of concept; the method is not limited to detecting Darwinian ideas but is capable of detecting similarities on a large scale in a wide range of corpora and contexts.

cs.CL↗