Tag
The article analyzes why external agents on an open network rarely return after initial interactions, emphasizing that retention challenges stem from issues in subsequent calls and a measurement bug in tracking completions.
A developer shares that after optimizing for an Electron application, the performance is nearing its maximum potential following a cold start with cleared page files.
Presents EGRL, a graph neural network framework for RNA-protein interaction prediction that improves cold-start generalization via edge generation and multi-relational attention, achieving competitive results on benchmark datasets.
This paper introduces domain-conditional position offsets, a learned vector added to initial token embeddings, to reduce the cold-start penalty in language models. The method trains in minutes on few documents, requires no model weight changes, and achieves up to 27% perplexity reduction across various model sizes.
This paper presents a dependency-aware autoscaling framework for serverless environments, integrating graph-based bottleneck identification, multi-model forecasting (MLP, LSTM, CNN) via a probabilistic ensemble, and cost-aware scaling control. Experiments show 99.88% prediction accuracy and reduced infrastructure costs.
This paper introduces Proximity Features, a privacy-compliant system that uses aggregated geo-IP data to personalize recommendations for cold-start users at Airbnb, achieving significant booking lifts in production experiments.
GPUHedge is an open-source tool that uses speculative execution to hedge between serverless GPU providers, reducing cold start p95 latency from 117s to 30s.
A detailed breakdown of Netflix's hybrid weighted recommendation system design, covering scale estimation, cold start strategies for new users, behavioral signal capture, and the balance between recall and precision.
The paper proposes DIF, a model-agnostic method for denoising implicit feedback in cold-start recommendation by using pseudo-labels from content-similar warm items and uncertainty estimation, achieving significant improvements in a billion-user video app.
This article provides a detailed review of Qi Junyuan's complete entrepreneurial journey from college to the success of Teambition, covering five stages: groundwork, first startup failure, self-built internal tool, cold start growth, and capital-driven breakout. It showcases typical early-stage strategies for grassroots 2B SaaS.
This paper proposes Representation Curriculum (RC), a training-time intervention that stages feature utilization to reduce over-reliance on exposure-confounded historical signals and improve cold-start generalization in ranking systems. The method is theoretically analyzed and validated on public benchmarks and large-scale eBay search experiments.
A discussion on how AI agents should handle user context: upfront disclosure or gradual learning, with various existing approaches like project memory and chat summaries found lacking.
Introduces CausalPOI, a spatio-temporal graph-based causal representation learning framework for cold-start POI check-in forecasting, which outperforms state-of-the-art baselines on real-world SafeGraph datasets.
This paper identifies a 'cold-start safety gap' in tool-calling LLM agents, where they are most vulnerable at the beginning of a session and become safer after completing regular agentic tasks. The authors introduce the SODA benchmark to evaluate this phenomenon and recommend a simple deployment strategy of warming up agents with regular tasks before safety-critical requests.
PaperFlow is a framework for scientific paper recommendation that processes user profiles, daily paper streams, and interest drift through three stages: profiling, recommending, and adapting, evaluated on a longitudinal benchmark with 24 users and 50 daily streams.
The paper introduces Human-in-the-Loop Gated Bandit (HITL-GB) for short-term rental dynamic pricing, showing that historical pricing data under a prior policy is structurally equivalent to on-policy warm-up data, reducing cold-start from ~150 to ~30 episodes.
Browser Use launches a new browser infrastructure service featuring subsecond cold starts, lower cost at $0.02/h, and unlimited scaling, now live for developers.
Discusses the cold start problem in AI personalization, where new products lack user data, and proposes a unified user data API as a potential solution that is consented and user-owned.
Modal explains how it reduces AI inference cold starts by 40x using cloud buffers, a custom filesystem, checkpoint/restore, and CUDA checkpoint/restore, framing cloud buffer management as a linear optimization problem solved with GLOP.
A software engineer asks for strategies to bootstrap personalization for new users with no behavioral data, discussing the cold-start problem in content recommendation.