SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2602.06025 · ICML · 2026

Learning Query-Aware Budget-Tier Routing for Runtime Agent Memory

Jiaxuan You, Xiao Li, Chengwei Qin, Tao Feng, Bowen Jin, Wenya Wang, Weizhi Zhang, Haozhen Zhang, Jianzhu Bao, Quanyu Long, Haodong Yue

arXiv · PDF · Open in the Atlas

Code that ran

1 of 4 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (1 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 1 has no contract verdict; 0 failures of our instrument; 3 no recorded run), measured 2026-11-22; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2602.06025.json · badge: /api/ran/2602.06025.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
ViktorAxelsen/BudgetMem community
Papers with Code archive
1 ran · 3 no recorded run
FunctionStatusWhere it lives
Experience Ran
ran on a synthesized input · constructed an object; no result was computed
ViktorAxelsen/BudgetMem/src/trainer/ppo_trainer.py
code served (permissive licence) · get_code("22603ec8294c0b15")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, Experience: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# Create minimal tensors with distinct dimensions
query_emb = torch.randn(4)  # 1D embedding
memory_embs = torch.randn(2, 4)  # 2 memory items, 4-dim embeddings
initial_memory_emb = torch.randn(2, 4)  # same shape as memory_embs
filtered_memory_emb = torch.randn(3, 4)  # different number of items
entity_emb = torch.randn(5, 4)  # another distinct size
temporal_emb = torch.randn(6, 4)  # yet another size
topic_emb = torch.randn(7, 4)  # and another
aggregated_memory_emb = torch.randn(2, 4)  # back to 2 items

ARGS = {
    "query_emb": query_emb,
    "memory_embs": memory_embs,
    "actions": [0, 1, 2],
    "log_probs": [0.5, -0.3, -0.8],
    "reward": 1.5,
    "cost": 0.2,
    "value": 0.8,
    "done": True,
    "initial_memory_emb": initial_memory_emb,
    "filtered_memory_emb": filtered_memory_emb,
    "entity_emb": entity_emb,
    "temporal_emb": temporal_emb,
    "topic_emb": topic_emb,
    "aggregated_memory_emb": aggregated_memory_emb,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Experience — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a Experience output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
ActorCriticNetwork Not yet run
unverified (harvested, no recorded run)
ViktorAxelsen/BudgetMem/src/trainer/ppo_trainer.py
code served (permissive licence) · get_code("f7c912ccfce8d065")
How it ran

no invocation recorded.

ExperienceBuffer Not yet run
unverified (harvested, no recorded run)
ViktorAxelsen/BudgetMem/src/trainer/ppo_trainer.py
code served (permissive licence) · get_code("2d6976ec2792b33c")
How it ran

no invocation recorded.

PPOTrainer Not yet run
unverified (harvested, no recorded run)
ViktorAxelsen/BudgetMem/src/trainer/ppo_trainer.py
code served (permissive licence) · get_code("f2f164a70fdde915")
How it ran

no invocation recorded.

Also present in other papers' code

Of 4 samples lifted from this paper's linked repositories, 4 carry a function identity and 0 do not. Of the 4 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 4 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Memory is increasingly central to Large Language Model (LLM) agents operating beyond a single context window, yet most existing systems rely on offline, query-agnostic memory construction that can be inefficient and may discard querycritical information. Although runtime memory utilization is a natural alternative, prior work often incurs substantial overhead and offers limited explicit control over the performance-cost tradeoff. In this work, we present BudgetMem, a runtime agent memory framework for explicit, queryaware performance-cost control. BudgetMem structures memory processing as a set of memory modules, each offered in three budget tiers (i.e., LOW/MID/HIGH). A lightweight router performs budget-tier routing across modules to balance task performance and memory construction cost, which is implemented as a compact neural policy trained with reinforcement learning. Using Bud-getMem as a unified testbed, we study three complementary strategies for realizing budget tiers: implementation (method complexity), reasoning (inference behavior), and capacity (module model size). Across LoCoMo, LongMemEval, and Hot-potQA, BudgetMem surpasses strong baselines when performance is prioritized (i.e., high-budget setting), and delivers better accuracy-cost frontiers under tighter budgets. Moreover, our analysis disentangles the strengths and weaknesses of different tiering strategies, clarifying when each axis delivers the most favorable trade-offs under varying budget regimes. Code is available at https://github.com/ViktorAxelsen/BudgetMem

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2602.06025")
get_code_for_paper("2602.06025")
have("2602.06025")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2602.06025.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2602.06025.svg)](https://syntology.ai/paper/2602.06025)

Connect an agent — have() is free.