Searcharxiv⌕ Search

arXiv subjects

Yishen Chen

Publications and source records attributed to Yishen Chen.

7 recordsLinked to original sources

Hierarchical Alignment: Block-Level LoRA-DPO for Preference Tuning in Large Language Models

Preference tuning typically distributes adapter updates across every Transformer layer. We study how the placement of a fixed LoRA budget affects Direct Preference Optimization (DPO). Hierarchical Alignment partitions decoder depth into Local, Intermediate, and Global regions and restricts LoRA updates to a chosen contiguous block. Across Qwen and Llama models, three-seed sliding-window sweeps reveal broad low-loss basins around intermediate depth. Under exact trainableparameter matching, Middle achieves lower held-out DPO loss than an all-depth allocation on Qwen2.5-7B and Llama-3.1-8B. Generationlevel evidence is more variable. LLM judges and a blinded human study leave Selected versus All unresolved on Qwen; on Llama, two pairwise judges favor All while the human interval includes zero. Trained-adapter interventions show that learned depth components affect the optimization objective; removing the Llama Selected window also lowers generation quality. Across Qwen2.5 at 3B, 7B, and 14B, Middle versus All remains unresolved under Qwen3-Max; at 3B and 14B, both improve over Base. In our implementation, Middle also uses less peak memory and takes shorter steps than a parameter-matched all-depth allocation. Together, the results support intermediate depth as a practical starting point for fixed-budget preference tuning, though the best window and its behavioral value still need per-setting validation.

cs.CL↗

How code helps different tasks? A decompositional lens on LLM post-training

Evaluating code data as a single corpus can obscure which types of code data benefit which models and downstream tasks. Effective data selection requires understanding both the benefits of individual categories and whether these benefits persist when categories are combined. We introduce a decompositional lens for studying these effects in LLM post-training. We first decompose an execution-verified code corpus into interpretable categories based on the computational patterns of its solutions. Through controlled fine-tuning experiments, we compare individual categories with a balanced mixture across instruction-tuned models on question answering, mathematics, and code generation. The resulting response maps reveal recurring gains in average question-answering performance, while the same category can improve one model or task and degrade another. The best-performing category also varies with the starting model and target task. We then compose compact mixtures guided by these results and examine whether benefits observed in individual categories persist under joint training. On selected model--task pairs, mixtures whose constituents each improve the target task outperform both their best constituent and full-corpus training while using roughly 10--15\% of the full corpus. These exploratory findings illustrate a \emph{less is more} pattern and highlight how the value of code data in post training depends on which categories are combined for which model and task.

cs.AI↗

Welfare-Opaque Income: Taxation under AI-Agent Delegation

We study income taxation when an AI agent implements economically relevant choices through a rule hidden from the government. Alongside unobserved productive ability, this hidden preference-to-execution mapping creates \emph{double unobservability}: the same observable tax-base response can carry different welfare consequences. We call the resulting income \emph{welfare-opaque}. Our constructions show that tax-base statistics can coincide while reform welfare effects differ, even when mechanical welfare weights are identical. We derive an optimal-tax condition that adds a response-weighted execution wedge to the familiar sufficient statistics. A higher marginal rate gains a corrective benefit under local over-execution and an additional cost under local under-execution. Observing the wedge identifies the welfare effect of a marginal reform at the prevailing schedule; bounds on it deliver bounds on that effect. A controlled laboratory compares 4,500 model runs across five AI engines. Faithful delegation selects the score maximizer in essentially all runs. Conflicted objectives produce heterogeneous responses: Claude largely preserves the score maximizer, GLM moves predominantly downward, and GPT-mini and Qwen show concentrated lower-tail increases. Qwen also makes substantial downward adjustments. Different engines locate their departures at different points and in different directions of the designed distribution. Explicit scores align model rankings; formula-based objective instructions yield more uneven agreement. Qwen shows a clear positive tax-by-objective interaction, but its direction does not generalize across engines and the pooled sign depends on its inclusion. The analysis identifies execution information as a complement to conventional tax-base statistics.

cs.CY↗

The Organization of Inference: Information, Resource Constraints, and AI Production

The economic value of inference depends on how capacity and task information are distributed across stages of AI production. We study these organizational margins using controlled workflow experiments on externally verified software-engineering tasks. In two matched resource panels, direct execution records the same success rate of 59.6 percent at logical-token ceilings of 12,000 and 24,000, while success under information-constrained planning rises from 36.2 to 51.2 percent. The planning disadvantage narrows by 15.0 percentage points (95 percent task-cluster bootstrap interval: 4.2 to 25.8). A strict read-only planning campaign varies whether the planner sees the task issue. At 12,000 tokens, issue access raises success by about 16 percentage points over issue-hidden planning. Compared with direct execution, task-informed planning is about 10 points lower at 12,000 tokens; at 24,000 tokens, it shows a 29.6-point advantage. In the resource panels, direct execution uses substantially less than either ceiling, while the planning workflow's binding rate falls from 46.2 to 0.8 percent and downstream execution accounts for 89.9 percent of the increase in total use. Scale determines the capacity available to a system; workflow and information structure shape the productive value

cs.AI↗

How Do Agent Harnesses Create Value? Planning Information and Release Control in Stateful LLM Agents

Agent harnesses supply planning guidance, organize execution, and check completion. We study how these components affect success, erroneous acceptance, and cost in two Retail experiments and an Airline pilot in $τ^2$-bench. The primary comparison pairs prewritten task-specific plans (Fixed) with shuffled policy text matched in word count (Sham), isolating the contribution of guidance content. Across 265 matched cells, Fixed improves oracle-verified success by 7.17 percentage points (90\% task-clustered bootstrap interval, 1.15--13.36 points), with gains concentrated in higher-complexity tasks. A read-only terminal verifier rejects 61\% of Retail oracle-invalid episodes while withholding 17\% of correct ones, at less than one cent of additional cost per episode. Which component matters more depends on the loss assigned to erroneous acceptance: at low liability the planning gain dominates; at high liability the verifier's avoided false passes dominate---and a standalone verifier captures nearly all the false-pass benefit of the full planning-plus-verification stack at a fraction of its cost.

cs.AI↗

Bridging Semantics and Utilities: A Controllable and Scalable Market Simulation

Large language models(LLMs) have recently been used to construct generative agents for financial market simulation. Existing approaches typically prompt each agent separately with public market information and a private investor profile, and ask the model to directly generate trading actions. In this agent-wise design, the model repeatedly interprets information shared across the population, while individual behavior is controlled through textual descriptions of investor preferences and traits. This creates two structural difficulties: continuous traits such as risk sensitivity do not reliably induce ordered behavioral changes, and repeatedly processing the same public information causes inference costs to grow with the number of agents. We propose \textbf{SUM}, a Semantic-Utility Model that separates cohort-level semantic perception from agent-level utility-based decision-making. A language model interprets public market information for each investor type and produces semantic signals shared within the corresponding cohort. Individual agents then combine these signals with their private states, preferences, and constraints to generate orders through utility-based decisions. Experiments in an A-share limit-order-book market show that \textbf{SUM} provides more stable and monotonic behavioral control than prompt-based agents while substantially reducing repeated language-model inference as the population grows. Closed-loop simulations further reproduce directional market responses, volatility persistence, differentiated investment styles, and persistent wealth heterogeneity. These results establish \textbf{SUM} as a controllable, scalable, and semantically informed framework for financial market simulation.

cs.CL↗

DiffTune: Optimizing CPU Simulator Parameters with Learned Differentiable Surrogates

CPU simulators are useful tools for modeling CPU execution behavior. However, they suffer from inaccuracies due to the cost and complexity of setting their fine-grained parameters, such as the latencies of individual instructions. This complexity arises from the expertise required to design benchmarks and measurement frameworks that can precisely measure the values of parameters at such fine granularity. In some cases, these parameters do not necessarily have a physical realization and are therefore fundamentally approximate, or even unmeasurable. In this paper we present DiffTune, a system for learning the parameters of x86 basic block CPU simulators from coarse-grained end-to-end measurements. Given a simulator, DiffTune learns its parameters by first replacing the original simulator with a differentiable surrogate, another function that approximates the original function; by making the surrogate differentiable, DiffTune is then able to apply gradient-based optimization techniques even when the original function is non-differentiable, such as is the case with CPU simulators. With this differentiable surrogate, DiffTune then applies gradient-based optimization to produce values of the simulator's parameters that minimize the simulator's error on a dataset of ground truth end-to-end performance measurements. Finally, the learned parameters are plugged back into the original simulator. DiffTune is able to automatically learn the entire set of microarchitecture-specific parameters within the Intel x86 simulation model of llvm-mca, a basic block CPU simulator based on LLVM's instruction scheduling model. DiffTune's learned parameters lead llvm-mca to an average error that not only matches but lowers that of its original, expert-provided parameter values.

cs.LG↗