Personalized Memory Rubrics: Learning What to Remember per User via Rubric-Guided Credit Assignment
Authors: Yupeng Han, Kai Zhang, Shuochen Liu, Ze Liu, Xianquan Wang, Zhihong Pan, Xiangyu Zhao, Qi Liu, Enhong Chen
Organizations: University of Science and Technology of China & City University of Hong Kong & State Key Laboratory of Cognitive Intelligence · University of Science and Technology of China & State Key Laboratory of Cognitive Intelligence · City University of Hong Kong
Long-horizon personalized agents maintain an evolving memory by continuously deciding what to add, modify, or discard. However, learning an optimal update policy remains challenging. Conventional training relies solely on terminal question-answering accuracy, yielding a coarse feedback signal that treats all intermediate updates identically. Furthermore, update utility is inherently user-specific, as identical memory actions may yield opposite effects across different user personas. Existing methods fail to address this heterogeneity, relying on globally shared evaluation standards with at most a single scalar intermediate reward. To address these challenges, we propose Personalized Memory Rubrics (PMR), a framework that translates terminal feedback into fine-grained, per-step, and per-user credit assignment. PMR decomposes evaluation into a universal layer capturing cross-persona failure patterns and a per-user adaptive layer modeling individual traits as a specialized delta over the universal baseline. Through Rubric-Guided Credit Assignment, PMR redistributes credit at the token level, selectively amplifying validated decisions on successful trajectories while falling back to standard policy optimization on failed ones to avoid unreliable attribution. The rubric and memory policy co-evolve in a closed loop with anti-circularity guarantees. Crucially, PMR incurs zero inference overhead, operating without any rubric or external judge during deployment. Extensive experiments across four benchmarks demonstrate consistent gains over strong memory-augmented baselines.
Figures & tables
Figure 1. Long-horizon memory learning involves multiple updates but only a terminal QA reward, making fine-grained credit assignment difficult. Hand-crafted methods use fixed, non-personalized rules, while RL-based methods rely on sparse scalar feedback with ambiguous step/token attribution. PMR decouples direction from magnitude : the QA advantage determines the learning direction, while personalized rubric evidence modulates credit magnitude through direction-safe gating.
Figure 2. Overview of PMR. PMR first induces a personalized rubric Ru from shared cross-user patterns and residual per-user failures, and then uses it to shape direction-safe, per-step credit while preserving the trajectory-level update direction. A co-evolution loop iteratively updates the student, rubric, and credit assignment, with standard GRPO used as a fallback for failures. At inference, only the student model is required.
Benchmark
Type
#Users
#Queries
Settings
Metric
PersonaMem
ID
20
589 / 2727
32K, 128K
Accuracy
PrefEval
OOD
–
1000 / 1000
Explicit, Implicit
Accuracy
PersonaBench
OOD
6
263
4 noise levels
Macro-F1
PERMA
OOD
10
705 / 389
C/N × S/M
Accuracy
Table 1. Summary of evaluation benchmarks. PersonaMem is in-domain; the rest are out-of-domain.
Method
PersonaMem
PrefEval
PersonaBench
PERMA
32K
128K
Expl.
Impl.
0
0.3
0.5
0.7
C-S
C-M
N-S
N-M
Long Context
45.18
35.57
34.70
27.30
27.77
22.38
19.45
12.59
24.11
22.88
24.82
22.81
RAG
54.22
38.50
47.40
32.40
32.21
29.13
25.28
23.36
51.63
36.84
51.77
35.09
Mem0
48.53
39.67
59.10
48.80
17.43
18.95
19.31
16.49
52.76
51.67
51.49
51.15
A-Mem
59.63
40.34
64.70
54.30
30.09
28.52
25.81
24.10
53.05
52.70
52.91
49.87
LightMem
54.22
37.40
67.20
58.90
19.14
17.83
19.61
17.51
59.43
53.73
59.43
55.53
Table 2. Overall comparison across twelve evaluation settings. PersonaMem (32K/128K) is in-domain; PrefEval (Explicit/Implicit), PersonaBench (4 noise levels), and PERMA (C-S/C-M/N-S/N-M denote Clean/Noise × Single/Multi-Domain) are out-of-domain. Higher is better. Best results are in bold .
PersonaMem
PERMA
Configuration
32K
128K
C-S
C-M
N-S
N-M
PMR (Full)
76.76
69.45
63.82
53.98
61.70
56.56
w/o Positive-only
57.23
51.34
46.24
40.36
47.52
39.07
w/o Rubric amplitude ( wt )
62.16
52.81
47.52
35.99
46.24
41.90
w/o Diagnostic (5-dim)
63.85
58.05
51.06
46.27
50.35
49.36
w/o Entropy gate
65.66
60.24
59.57
52.18
57.16
51.41
Table 3. Ablation on credit assignment components. Cell color indicates drop from Full ( < 3 to ≥ 15 pp). Rows ordered by average drop.
Figure 3. Co-evolution dynamics across 8 rounds on PersonaMem 32K. Left axis: accuracy; PMR (full, rising) and w/o co-evolution (frozen rubric, plateauing). Right axis: ∣Runi∣ (growing) and active Rspec (decreasing as student learns).
PersonaMem
PERMA
Configuration
32K
128K
C-S
C-M
N-S
N-M
PMR (Full)
76.76
69.45
63.82
53.98
61.70
56.56
w/o Cross-persona sharing
54.05
45.76
42.69
34.70
42.98
34.45
w/o Coverage test
59.63
52.99
46.10
38.30
42.98
35.99
w/o Runi (per-user only)
66.49
62.96
51.06
41.90
49.22
44.21
w/o Quality gate
67.03
62.56
56.45
46.27
54.33
46.79
Table 4. Ablation on rubric structure components. Cell color indicates drop from Full ( < 3 to ≥ 15 pp).
Figure 4. Case study of learned personalized rubrics. PMR selectively adapts rubric dimensions to user behavior: Malia emphasizes timeliness and conciseness, Maria largely retains universal criteria, and Alex strengthens completeness and conciseness. Consistency remains universal across all three personas.
Figure 5. Per-user gains from Rspec across four persona groups defined by median splits of volatility and interaction frequency. Δ=Acc(PMR)−Acc(PMR-Global) , with larger gains for stable users.
Figure 6. Sensitivity to credit modulation strength λ and co-evolution rounds. (a) Held-out accuracy across the λ× rounds grid. (b) Accuracy versus λ at rounds 4 and 8, with shading indicating the robust plateau ( λ∈[0.25,1.0] ).
University of Science and Technology of China & State Key Laboratory of Cognitive Intelligence · City University of Hong Kong · Dalian University of Technology +1