Article Per Year (5 Year)
p-Index From 2021 - 2026
0.444
P-Index
This Author published in this journals
All Journal Journal of Technology Informatics and Engineering
Shilu He
Mathematics, UW-Madison, WI, USA
Computer Science & IT

Published : 2 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 2 Documents
Search

 1 
Few-Shot Cold-Start Workload Forecasting for New AI Inference Tenants with Time-Series Foundation Models Shilu He; Chengliang Li; Hengning Rao
Journal of Technology Informatics and Engineering Vol. 4 No. 1 (2025): APRIL | JTIE : Journal of Technology Informatics and Engineering
Publisher : University of Science and Computer Technology

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.51903/jtie.v4i1.546

Abstract

This paper presents a reproducible empirical study of few-shot cold-start workload forecasting for new AI inference tenants using the Alibaba GPU-disaggregated DLRM serving trace. Instance lifecycles are transformed into hourly active-demand series, and resource reservations are normalized into capacity units to evaluate 24-hour forecasting under zero-shot, 5-shot, 10-shot, and full-history settings. Seven forecasting methods are compared: archetype mean prior, persistence, moving average, linear trend, seasonal naive, global residual ridge, and CT-TSFM, a compact cross-tenant time-series foundation model. The cold-start evaluation uses 46 held-out tenants, with 110 source tenants for pretraining and calibration. Results show that hourly demand is strongly persistence-dominated. Zero-shot forecasting yields a mean absolute error (MAE) of 326.26 normalized capacity units, whereas only five observations reduce MAE to 4.00 for persistence, global residual ridge, and CT-TSFM. Validation consistently selects a residual gate of 0.0 for CT-TSFM, indicating that retaining the persistence prior and rejecting cross-tenant residual transfer is the most reliable strategy. Calibration intervals achieve approximately 85–87% coverage against a 90% target. The findings demonstrate that a few recent observations substantially improve cold-start forecasting, while source-tenant metadata alone provides limited zero-shot planning capability.
Power-Aware Inventory Planning for AI Infrastructure Using Job-Level Forecasting and LLM Workload Explanations Shilu He; Jiayi Nie; Chengliang Li
Journal of Technology Informatics and Engineering Vol. 5 No. 1 (2026): APRIL | JTIE : Journal of Technology Informatics and Engineering
Publisher : University of Science and Computer Technology

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.51903/jtie.v5i1.548

Abstract

AI infrastructure planning is commonly expressed as a GPU-count problem, yet operational risk is created by the electric and thermal envelope that accompanies each accelerator. This paper evaluates a power-aware planning method on Dataset A, using the B200 eight-GPU Llama-8B training trace with 45,000 raw 20 ms telemetry rows and 8,940 reproducible supervised decision records after a 100 ms decision stride. The forecasting task predicts total eight-GPU power one second ahead from job-level counters, autoregressive lags, and rolling statistics. The planning task converts forecasts into a peak-aware admission rule and a circuit-inventory simulation for 32 concurrent jobs. XGBoost produced the strongest mean forecast, with MAE 273.26 W, RMSE 636.74 W, and R2 0.923. A calibrated high-quantile forecast produced lower peak-error behavior, reducing the scheduling violation rate from 5.31% under GPU-count-only admission to 0.18% while admitting 61.63% of decision points. In the inventory simulation, XGBoost mean forecasting used 21.00 mean circuits with 1.80% violation risk, whereas the calibrated p95 plan used 22.70 circuits and eliminated observed violations in 1,000 trials. The results show that capacity plans based only on GPU count hide measurable electrical risk. A combined GPU-capacity, power-envelope, and workload-explanation view produces a reproducible basis for AI data center purchasing, placement, and sustainability decisions.
Co-Authors Chengliang Li Hengning Rao Jiayi Nie