optim-agent

Optim-Agent/optim-agent is ranked 4,113 among these 8 peers; facebookresearch/videoseal is ranked 4,111.

optim-agent Agentic system optimization with coding agents. Automate the iterative parameter-tuning work of an algorithm engineer. Use it when your system exposes configurable parameters and a measurable objective. It combines what each parameter *means* with what the trial history *shows*, then proposes the next configuration to evaluate. Objective evaluations remain authoritative: optim-agent proposes values, validates them against the declared space, records outcomes, and falls back to safe sampling when an agent reply is invalid. **Small-budget leverage

** - useful when evaluations are expensive and classical surrogates are still data-starved. **Agent CLI upside** - proposal quality can improve as the underlying coding agents improve, such as moving from GPT-5. 6, without changing your optimization code. **Auditable decisions** - JSON/SQLite studies retain configurations, outcomes, states, context, and optional agent rationale. **Bounded execution** - the agent only proposes values; optim-agent validates them against the declared space, and invalid output falls back to safe sampling. Install Install the Codex skill: Install

the Claude Code plugin: Install the Python package: Requires one authenticated agent CLI on : claude, codex, or opencode. Quickstart Optional gives domain meaning to the study and parameters. Provide it study-wide on , per parameter on , or both. Optimization Trajectory This seed-0 Branin trace compares TPE and GPT-5. 5 under the same 10-trial budget, with incumbent objective values after each trial. It is a trajectory illustration; aggregate benchmark results and reproduction commands follow. Optimizing Math Functions without

Context: Branin-2D and Ackley-5D Hard-function agents receive **no supplied task context**: only generic parameter names, numeric bounds, and trial history. Runs use 10 trials over five seeds; Random and TPE are unchanged baselines. 023 | The pinned models are , , , , , and . 8 reaches the Branin optimum on average and has the strongest five-seed Ackley mean. 597 | OpenCode-hosted models require no paid model API. The free pool rotates; this refresh pins , , , and . DeepSeek V4 Flash has the strongest

free-model Ackley mean, while MiMo-v2. 5 has the strongest free-model Branin mean. Tuning ResNet-based Image Classifier: MNIST and CIFAR-10 The classification benchmark compares **Random**, Optuna **TPE**, **GPT-5. 5 w/o context** over five seeds ( ) and 10 trials. The context condition receives natural-language study and parameter descriptions; the no-context condition receives only bounds and trial history. For classification, the primary metric emphasizes fast improvement: Lower is better. 5 w/ context reduces cumulative best

-so-far error by **20. 9%** relative to TPE on MNIST and by **20. The gap includes both semantic parameter information and earlier access to agent-guided proposals. Both and tune learning rate, batch size, weight decay, label smoothing, three stage widths, three stage depths, and four dropout controls. MNIST adds translation and rotation; CIFAR-10 uses crop padding and flip probability. Tuning Q-learning Controllers: Acrobot-v1 and LunarLander-v3 This CPU-only Gymnasium benchmark tunes a discretized Q

-learning controller for Acrobot-v1 and LunarLander-v3. Each method runs 20 trials over five seeds ( ); the objective is mean