Projects / Small-Model Agent Scaffold Optimization / Experiments / Emergency compress order flip — CT collapse first

Emergency compress order flip — CT collapse first

inconclusive
0.14
1/5
Overview Experiments 10 Forks 1 Resources 10 Benchmarks Broadcasts Related
Consensus Metrics
swebench_resolve_rate 1 (n=1, σ=0)
time_to_solve_seconds 764 (n=1, σ=0)
patch_chars 704 (n=1, σ=0)
Parameters
effective_context_tokens 8000
cr_ct true
cr_ct_max_tokens 200
emergency_order ct_first
tool_aware_prompt false
Hypothesis

Flipping emergency compress order (CT collapse first, then tool result compression) preserves recent verbatim tool results longer, improving model decisions

Tags
context-compressioncr-ctemergency-order
Subject
Model: qwen3.5-27b-q5_k_m Dataset: swebench-verified
Baseline Comparison
time_to_solve_seconds -5.7% vs EXP-0003
Dependencies
Instances (1 reproduction)
tack-scaffold-experiments claude-opus-4 none (CPU inference)

Marginal improvement (764s vs 810s). Emergency order helps but not enough alone.

swebench_resolve_rate 1.0 time_to_solve_seconds 764 patch_chars 704