arxiv arXiv cs.AI · 1h ago · research

AgentZip reduces AI-agent sandbox memory by up to 8.7x

from English

Researchers present AgentZip, a memory compression system designed specifically for high-fanout AI-agent sandboxes to address memory bottlenecks caused by concurrent sessions.

Exploits template-relative and cross-sandbox redundancy to compress pages beyond identical matches.
Shifts overhead control from conservative page selection at compression time to restore-time prefetching.
Aligns expensive compression operations with LLM waiting periods to avoid interfering with foreground tool execution.
Achieves up to 8.7x reduction in sandbox-owned memory, compared to 2.1x for standard Linux configuration.
Reduces the performance slowdown of aggressive compression from 3.1x to 1.40x while retaining most memory savings.

AgentZip effectively mitigates memory pressure in agent workloads by optimizing when and how data is compressed without significantly impacting execution speed.

Importance 2/3 New harness with differentiators arXiv cs.AI AI agents Inference efficiency

Read original

arxiv arXiv cs.LG · 4d ago · 4 views

Speculative Uncertainty method cuts software engineering agent errors using draft-model gating

The authors present Speculative Uncertainty (SU), a technique that recovers a predictive failure signal for black-box LLM agents by analyzing their output tokens alone, without requiring access to logits, weights, or activations. By inverting speculative decoding, a small open-weight draft model scores the agent's trajectory in a single forward pass to extract phase-aware features and calibrate them against a verifiable objective.

lab Hugging Face Blog · 23d ago · 16 views

ALTK-Evolve calibrates agentic memory dosage to model capability for better performance

A study of eight models reveals that agentic memory is not a universal feature but a dose that must be calibrated to the specific model's capability. The ALTK-Evolve framework distills guidelines from an agent's past trajectories and injects them at inference time without weight updates, showing that optimal memory strategies vary significantly across model tiers.

lab Hugging Face Blog · 31d ago · 11 views

ALTK-Evolve achieves same-or-better accuracy than ACE at a fraction of inference cost

ALTK-Evolve, a system for agentic memory that converts agent trajectories into reusable guidelines without weight updates, matches or exceeds the accuracy of ACE while using significantly fewer tokens. Unlike ACE, which injects a comprehensive playbook on every step, ALTK-Evolve uses selective retrieval to deliver only relevant guidelines based on model capacity.

media AI News (smol.ai) · 31d ago · 21 views

Meta releases Muse Glimmer 30B open-weight model; Anthropic improves Riemann bound

Meta released Muse Glimmer, a 30B dense multimodal model optimized for local agent deployment under Apache 2.0, while also promising future weights for Muse Spark 1.2.

media Hugging Face Forums · 39d ago · 13 views

Quesma publishes curated map of 218 tools and papers for LLM token cost efficiency

Quesma has released an open, curated list containing 218 entries that track where tokens are spent in LLM and agent workflows. The resource covers monitoring, semantic caching, cheap local models, context engineering, and cost governance.