{"id":28286597,"url":"https://github.com/wecoai/weco-cli","last_synced_at":"2026-04-16T12:01:05.314Z","repository":{"id":285959962,"uuid":"959804273","full_name":"WecoAI/weco-cli","owner":"WecoAI","description":"Production-Grade Autoresearch. Ideal for GPU kernels, ML model development, feature engineering, prompt engineering, and other optimizable code.","archived":false,"fork":false,"pushed_at":"2026-04-15T19:48:02.000Z","size":60691,"stargazers_count":41,"open_issues_count":4,"forks_count":9,"subscribers_count":2,"default_branch":"main","last_synced_at":"2026-04-15T21:26:19.003Z","etag":null,"topics":["code-generation","code-optimization","kernel-optimization","machine-learning","prompt-engineering"],"latest_commit_sha":null,"homepage":"https://weco.ai","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"apache-2.0","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/WecoAI.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":"contributing.md","funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2025-04-03T11:32:04.000Z","updated_at":"2026-04-15T19:46:09.000Z","dependencies_parsed_at":"2025-05-27T12:25:07.886Z","dependency_job_id":null,"html_url":"https://github.com/WecoAI/weco-cli","commit_stats":null,"previous_names":["wecoai/weco-cli"],"tags_count":54,"template":false,"template_full_name":null,"purl":"pkg:github/WecoAI/weco-cli","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/WecoAI%2Fweco-cli","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/WecoAI%2Fweco-cli/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/WecoAI%2Fweco-cli/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/WecoAI%2Fweco-cli/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/WecoAI","download_url":"https://codeload.github.com/WecoAI/weco-cli/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/WecoAI%2Fweco-cli/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":31884929,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-04-16T11:36:10.202Z","status":"ssl_error","status_checked_at":"2026-04-16T11:36:09.652Z","response_time":69,"last_error":"SSL_connect returned=1 errno=0 peeraddr=140.82.121.5:443 state=error: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["code-generation","code-optimization","kernel-optimization","machine-learning","prompt-engineering"],"created_at":"2025-05-21T20:22:43.073Z","updated_at":"2026-04-16T12:01:05.307Z","avatar_url":"https://github.com/WecoAI.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"\u003cdiv align=\"center\"\u003e\n\n\u003cdiv align=\"center\"\u003e\n  \u003cimg src=\"assets/weco.svg\" alt=\"Weco Logo\" width=\"120\" height=\"120\" style=\"margin-bottom: 20px;\"\u003e\n  \u003ch1\u003eWeco: Production-Grade Autoresearch\u003c/h1\u003e\n\u003c/div\u003e\n\n[![Python](https://img.shields.io/badge/Python-3.8.0+-blue)](https://www.python.org)\n[![PyPI version](https://img.shields.io/pypi/v/weco?label=PyPI%20version\u0026color=f05138\u0026labelColor=555555)](https://badge.fury.io/py/weco)\n[![docs](https://img.shields.io/website?url=https://docs.weco.ai/\u0026label=docs)](https://docs.weco.ai/)\n[![PyPI Downloads](https://static.pepy.tech/badge/weco?color=4c1)](https://pepy.tech/projects/weco)\n[![arXiv on AIDE](https://img.shields.io/badge/arXiv-AIDE-b31b1b?logo=arxiv\u0026logoColor=white)](https://arxiv.org/abs/2502.13138)\n[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg?labelColor=ffffff\u0026color=F17E01)](https://colab.research.google.com/github/WecoAI/weco-cli/blob/main/examples/hello-world/colab_notebook_walkthrough.ipynb)\n\n`pip install weco`\n\n[Docs](https://docs.weco.ai) \u0026nbsp;•\u0026nbsp; [Examples](https://weco.ai/examples) \u0026nbsp;•\u0026nbsp; [Dashboard](https://dashboard.weco.ai)\n\n\u003c/div\u003e\n\n---\n\nWeco systematically optimizes your code, guided directly by your evaluation metrics.\n\nExample applications include:\n\n- **GPU Kernel Optimization**: Reimplement PyTorch functions using [CUDA](/examples/cuda/README.md) or [Triton](/examples/triton/README.md), optimizing for `latency`, `throughput`, or `memory_bandwidth`.\n- **Model Development**: Tune feature transformations, architectures or [the whole training pipeline](/examples/spaceship-titanic/README.md), optimizing for `validation_accuracy`, `AUC`, or `Sharpe Ratio`.\n- **Prompt Engineering**: Refine prompts for LLMs (e.g., for [math problems](/examples/prompt/README.md)), optimizing for `win_rate`, `relevance`, or `format_adherence`\n\n![image](assets/example-optimization.gif)\n\n---\n\n## Overview\n\nThe `weco` CLI leverages a tree search approach guided by LLMs to iteratively explore and refine your code. It automatically applies changes, runs your evaluation script, parses the results, and proposes further improvements based on the specified goal.\n\n## Skills (Claude Code \u0026 Cursor)\n\nWeco ships as a **skill** for AI coding assistants. A skill is a set of instructions that teaches your assistant how to use Weco end-to-end — from setting up optimizations to interpreting results. Once installed, just describe what you want to optimize in plain language and your assistant handles the rest.\n\n```bash\nweco setup claude-code   # installs skill into Claude Code\nweco setup cursor        # installs skill into Cursor\n```\n\nThen prompt naturally:\n\n```text\nUse Weco to make this function faster.\n```\n\nYour assistant will inspect your code, write the evaluation, configure `weco run`, monitor the iterations, and explain the results — no CLI flags needed.\n\nSee the full [Skills guide](https://docs.weco.ai/skills) for details.\n\n## Observe (Track External Experiments)\n\nRunning your own optimization loop with an LLM agent, a custom script, or a manual workflow? `weco observe` lets you track those experiments in the Weco dashboard with tree visualization, code diffs, and metric tracking — without handing off control of the optimization itself.\n\n```bash\n# Initialize a run\nWECO_RUN_ID=$(weco observe init --name \"my-experiment\" --metric val_bpb --goal min --source train.py)\n\n# Log experiments\nweco observe log --run-id \"$WECO_RUN_ID\" --step 0 --description \"baseline\" \\\n  --metrics '{\"val_bpb\": 2.36}' --source train.py\nweco observe log --run-id \"$WECO_RUN_ID\" --step 1 --description \"increase batch size\" \\\n  --metrics '{\"val_bpb\": 2.26}' --source train.py\n```\n\nAll observe commands are fire-and-forget (always exit 0), so they never crash an agent loop. There is also a [Python SDK](https://docs.weco.ai/observe#python-sdk) for scripts with a Python loop.\n\nSee the full [Observe guide](https://docs.weco.ai/observe) for branching, lifecycle, and more.\n\n\n## Install the Package\n\n**macOS / Linux** (recommended):\n\n```bash\ncurl -fsSL https://weco.ai/install.sh | sh\n```\n\n**Windows CMD:**\n\n```cmd\npowershell -ExecutionPolicy ByPass -c \"irm https://weco.ai/install.ps1 | iex\"\n```\n\n**Windows PowerShell:**\n\n```powershell\nirm https://weco.ai/install.ps1 | iex\n```\n\n**pip:**\n\n```bash\npip install weco\n```\n\n**From source:**\n\n```bash\ngit clone https://github.com/wecoai/weco-cli.git\ncd weco-cli\npip install -e .\n```\n\n## Getting Started\n\n### Quickstart with an example project\n\n**Configure optimization parameters yourself** - If you need precise control over the optimization parameters, you can use the direct `weco run` command:\n\n**Example: Optimizing Simple PyTorch Operations**\n\n```bash\ngit clone https://github.com/WecoAI/weco-cli.git\ncd weco-cli/examples/hello-world/\npip install -r requirements.txt\n\n# Run Weco with configuration\nweco run --source module.py \\\n     --eval-command \"python evaluate.py --path module.py\" \\\n     --metric speedup \\\n     --goal maximize \\\n     --steps 10 \\\n     --additional-instructions \"Fuse operations in the forward method while ensuring the max float deviation remains small. Maintain the same format of the code.\"\n```\n\n**Note:** If you have an NVIDIA GPU, change the device in the `--eval-command` to `cuda`. If you are running this on Apple Silicon, set it to `mps`.\n\n**Multi-file optimization:** If your code spans multiple files, use `--sources` to optimize them together:\n\n```bash\nweco run --sources model.py utils.py config.py \\\n     --eval-command \"python evaluate.py\" \\\n     --metric accuracy \\\n     --goal maximize \\\n     --steps 10\n```\n\nWeco will optimize all specified files simultaneously, allowing changes across file boundaries.\n\nFor more advanced examples, including [Triton](/examples/triton/README.md), [CUDA kernel optimization](/examples/cuda/README.md), [ML model optimization](/examples/spaceship-titanic/README.md), and [prompt engineering for math problems](examples/prompt/README.md), please see the `README.md` files within the corresponding subdirectories under the [`examples/`](examples/) folder.\n\n\u003e Note: When recommend removing any backticks from your code if any are present. We currently don't support backticks but will support this in the future.\n\n---\n\n### Arguments for `weco run`\n\n**Required:**\n\n| Argument            | Description                                                                                                                                                                                  | Example               |\n| :------------------ | :------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | :-------------------- |\n| `-s, --source`      | Path to a single source code file to be optimized. Mutually exclusive with `--sources`.                                                                                | `-s model.py`      |\n| `--sources`          | Paths to multiple source code files to be optimized together. Mutually exclusive with `-s, --source`.                                                                  | `--sources model.py utils.py config.py` |\n| `-c, --eval-command`| Command to run for evaluating the code in `--source`. This command should print the target `--metric` and its value to the terminal (stdout/stderr). See note below.                        | `-c \"python eval.py\"` |\n| `-m, --metric`      | The name of the metric you want to optimize (e.g., 'accuracy', 'speedup', 'loss'). This metric name does not need to match what's printed by your `--eval-command` exactly (e.g., its okay to use \"speedup\" instead of \"Speedup:\").                                    | `-m speedup`          |\n| `-g, --goal`        | `maximize`/`max` to maximize the `--metric` or `minimize`/`min` to minimize it.                                                                                                              | `-g maximize`         |\n\n\u003cbr\u003e\n\n**Optional:**\n\n| Argument                       | Description                                                                                                                                                                                                                | Default                                                                                                                                                | Example             |\n| :----------------------------- | :------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | :------------------------------------------------------------------------------------------------------------------------------------------------------ | :------------------ |\n| `-n, --steps`                  | Number of optimization steps (LLM iterations) to run.                                                                                                                                                                      | 100                                                                                                                                                     | `-n 50`             |\n| `-M, --model`                  | Model identifier for the LLM to use (e.g., `o4-mini`, `claude-sonnet-4-5`, `gpt-5`).                                                                                                        | `o4-mini` | `-M o4-mini`         |\n| `-i, --additional-instructions`| Natural language description of specific instructions **or** path to a file containing detailed instructions to guide the LLM. Supported file formats include - `.txt`, `.md`, and `.rst`.                                                                                             | `None`                                                                                                                                                  | `-i instructions.md` or `-i \"Optimize the model for faster inference\"`|\n| `-l, --log-dir`                | Path to the directory to log intermediate steps and final optimization result.                                                                                                                                             | `.runs/`                                                                                                                                               | `-l ./logs/`        |\n| `--eval-timeout`       | Timeout in seconds for each step in evaluation.                                                                                                                                                                             | No timeout (unlimited)                                                                                                                                                  | `--eval-timeout 3600`             |\n| `--save-logs`          | Save execution output from each optimization step to disk. Creates timestamped directories with raw output files and a JSONL index for tracking execution history.                                                        | `False`                                                                                                                                                 | `--save-logs`       |\n| `--apply-change`       | Automatically apply the best solution to the source file without prompting.                                                                                                                                                | `False`                                                                                                                                                 | `--apply-change`       |\n| `--api-key`            | API keys for LLM providers (BYOK). Format: `provider=key`. Can specify multiple providers.                                                                                                                                  | `None`                                                                                                                                                  | `--api-key openai=sk-xxx` |\n\n---\n\n## Command Reference\n\n### Basic Usage Patterns\n\n| Command | Description | When to Use |\n|---------|-------------|-------------|\n| `weco run [options]` | Start a new optimization | When you know what to optimize and how |\n| `weco resume \u003crun-id\u003e` | Resume an interrupted run | Continue from the last completed step |\n| `weco login` | Authenticate with Weco | First-time setup or switching accounts |\n| `weco logout` | Clear authentication credentials | Switch accounts or troubleshoot auth |\n| `weco credits balance` | Check your current credit balance | Monitor usage |\n| `weco credits topup [amount]` | Purchase additional credits | When you need more credits (default: 10) |\n| `weco credits autotopup` | Configure automatic top-up | Set up automatic credit replenishment |\n\n### Run Subcommands\n\nInspect and manage optimization runs. All output is JSON, designed for programmatic access (AI coding agents, scripts).\n\n| Command | Description |\n|---------|-------------|\n| `weco run status \u003crun-id\u003e` | Run progress, pending nodes, review mode flag |\n| `weco run results \u003crun-id\u003e` | Results sorted by metric |\n| `weco run show \u003crun-id\u003e --step \u003cN\\|best\u003e` | Single node detail with code |\n| `weco run diff \u003crun-id\u003e --step \u003cN\\|best\u003e` | Unified code diff between steps |\n| `weco run stop \u003crun-id\u003e` | Graceful termination (tree preserved) |\n| `weco run instruct \u003crun-id\u003e \"\u003ctext\u003e\"` | Update instructions mid-run |\n| `weco run review \u003crun-id\u003e` | List pending approval nodes (review mode) |\n| `weco run revise \u003crun-id\u003e --node \u003cid\u003e --source \u003cfile\u003e` | Replace a node's code |\n| `weco run submit \u003crun-id\u003e --node \u003cid\u003e` | Evaluate and submit a node |\n\n```bash\n# Check progress\nweco run status 0002e071-1b67-411f-a514-36947f0c4b31\n\n# Top 5 results as JSON\nweco run results 0002e071-1b67-411f-a514-36947f0c4b31 --top 5\n\n# Diff best solution against baseline\nweco run diff 0002e071-1b67-411f-a514-36947f0c4b31 --step best\n\n# Review mode: inspect, optionally edit, and submit\nweco run review 0002e071-1b67-411f-a514-36947f0c4b31\nweco run submit 0002e071-1b67-411f-a514-36947f0c4b31 --node \u003cnode-id\u003e\n\n# Submit with your own code (explicit path mapping)\nweco run submit \u003crun-id\u003e --node \u003cid\u003e --source module.py=./my_version.py\n```\n\n**Source path mapping:** When using `--source` with `revise` or `submit`, you can map local files to the run's source paths using `target_path=local_path` syntax (e.g., `--source module.py=./optimized.py`). Without an explicit mapping, files are matched positionally to the run's original source paths.\n\n### Observe Commands\n\nTrack experiments from your own optimization loop (LLM agents, custom scripts, manual experiments) in the Weco dashboard:\n\n| Command | Description |\n|---------|-------------|\n| `weco observe init` | Create a run and print the run ID |\n| `weco observe log` | Log a step with metrics and code |\n\n```bash\n# Initialize a run\nWECO_RUN_ID=$(weco observe init --name \"my-experiment\" --metric val_bpb --goal min --source train.py)\n\n# Log baseline (step 0) and experiments (step 1, 2, ...)\nweco observe log --run-id \"$WECO_RUN_ID\" --step 0 --description \"baseline\" --metrics '{\"val_bpb\": 2.36}' --source train.py\nweco observe log --run-id \"$WECO_RUN_ID\" --step 1 --description \"increase batch size\" --metrics '{\"val_bpb\": 2.26}' --source train.py\nweco observe log --run-id \"$WECO_RUN_ID\" --step 2 --status failed --description \"OOM\" --metrics '{\"val_bpb\": 0.0}'\n```\n\nAll observe commands are fire-and-forget — they always exit 0, so they never crash an agent's loop. For branching, pass `--parent-step` explicitly. See `weco observe init --help` and `weco observe log --help` for all options.\n\n### Setup Commands (Experimental)\n\n| Command | Description |\n|---------|-------------|\n| `weco setup claude-code` | Set up Weco skill for Claude Code |\n| `weco setup cursor` | Set up Weco skill for Cursor |\n| `weco setup codex` | Set up Weco skill for Codex |\n| `weco setup openclaw` | Set up Weco skill for OpenClaw |\n| `weco setup all` | Set up Weco for all supported AI tools |\n\nThe `setup` command installs Weco skills for AI coding assistants:\n\n```bash\nweco setup              # Interactive picker, defaults to \"All of the above\"\nweco setup claude-code  # For Claude Code\nweco setup cursor       # For Cursor\nweco setup codex        # For Codex\nweco setup openclaw     # For OpenClaw\nweco setup all          # For all supported tools\n```\n\n- **Claude Code**: Downloads the Weco skill to `~/.claude/skills/weco/` and writes `CLAUDE.md` inside the installed skill\n- **Cursor**: Downloads the Weco skill to `~/.cursor/skills/weco/`\n- **Codex**: Downloads the Weco skill to `$CODEX_HOME/skills/weco/` (defaults to `~/.codex/skills/weco/`)\n- **OpenClaw**: Downloads the Weco skill to `~/.openclaw/skills/weco/`\n\n### Model Selection\n\nYou can specify which LLM model to use with the `-M` or `--model` flag:\n\n```bash\nweco run --model gpt-5 --source optimize.py [other options...]\n```\n\n**Available models (37 total):**\n\n**OpenAI Models:**\n- GPT-5 Series: `gpt-5.2`, `gpt-5.2-pro`, `gpt-5.2-codex`, `gpt-5.1`, `gpt-5.1-codex`, `gpt-5.1-codex-mini`, `gpt-5.1-codex-max`, `gpt-5-codex`, `gpt-5-pro`, `gpt-5`, `gpt-5-mini`, `gpt-5-nano`\n- O-Series Reasoning: `o3-pro`, `o3`, `o3-mini`, `o4-mini`, `o1-pro`, `o1`\n- GPT-4 Series: `gpt-4.1`, `gpt-4.1-mini`, `gpt-4.1-nano`, `gpt-4o`, `gpt-4o-mini`\n\n**Anthropic Claude (via Vertex AI):**\n- `claude-opus-4-6`, `claude-sonnet-4-6`, `claude-opus-4-5`, `claude-opus-4-1`, `claude-opus-4`, `claude-sonnet-4-5`, `claude-sonnet-4`, `claude-haiku-4-5`\n\n**Google Gemini:**\n- `gemini-3.1-pro-preview`, `gemini-3-flash-preview`, `gemini-2.5-pro`, `gemini-2.5-flash`, `gemini-2.5-flash-lite`\n\nAll models are available through Weco. If no model is specified, Weco automatically selects the best model for your optimization task.\n\n---\n\n### Resuming Interrupted Runs\n\nIf your optimization run is interrupted (network issues, restart, etc.), resume from the most recent node:\n\n```bash\n# Resume an interrupted run\nweco resume 0002e071-1b67-411f-a514-36947f0c4b31\n\n```\n\nArguments for `weco resume`:\n\n| Argument | Description | Example |\n|----------|-------------|---------|\n| `run-id` | The UUID of the run to resume (shown at the start of each run) | `0002e071-1b67-411f-a514-36947f0c4b31` |\n| `--apply-change` | Automatically apply the best solution to the source file without prompting | `--apply-change` |\n| `--api-key` | (Optional) API keys for LLM providers (BYOK). Format: `provider=key` | `--api-key openai=sk-xxx` |\n\nNotes:\n- Works only for interrupted runs (status: `error`, `terminated`, etc.).\n- You’ll be prompted to confirm that your evaluation environment (source file + evaluation command) hasn’t changed.\n- The source file is restored to the most recent solution before continuing.\n- All progress and metrics from the original run are preserved.\n- Log directory, save-logs behavior, and evaluation timeout are reused from the original run.\n\n### Performance \u0026 Expectations\n\nWeco, powered by the AIDE algorithm, optimizes code iteratively based on your evaluation results. Achieving significant improvements, especially on complex research-level tasks, often requires substantial exploration time.\n\nThe following plot from the independent [Research Engineering Benchmark (RE-Bench)](https://metr.org/AI_R_D_Evaluation_Report.pdf) report shows the performance of AIDE (the algorithm behind Weco) on challenging ML research engineering tasks over different time budgets.\n\n\u003cp align=\"center\"\u003e\n\u003cimg src=\"https://github.com/user-attachments/assets/ff0e471d-2f50-4e2d-b718-874862f533df\" alt=\"RE-Bench Performance Across Time\" width=\"60%\"/\u003e\n\u003c/p\u003e\n\nAs shown, AIDE demonstrates strong performance gains over time, surpassing lower human expert percentiles within hours and continuing to improve. This highlights the potential of evaluation-driven optimization but also indicates that reaching high levels of performance comparable to human experts on difficult benchmarks can take considerable time (tens of hours in this specific benchmark, corresponding to many `--steps` in the Weco CLI). Factor this into your planning when setting the number of `--steps` for your optimization runs.\n\n---\n\n### Saving Execution Logs\n\nWhen using the `--save-logs` flag, Weco saves the execution output from each optimization step to help with debugging and analysis. The logs are organized as follows:\n\n```\n.runs/\n└── \u003csource-file-name\u003e/\n    └── \u003crun-uuid\u003e/\n        ├── exec_output.jsonl      # Index file with metadata for each step\n        ├── outputs/\n        │   ├── step_0.out.txt      # Raw output from initial evaluation\n        │   ├── step_1.out.txt      # Raw output from step 1\n        │   ├── step_2.out.txt      # Raw output from step 2\n        │   └── ...\n        ├── step_0.py               # Code snapshot from initial evaluation\n        ├── step_1.py               # Code snapshot from step 1\n        ├── step_2.py               # Code snapshot from step 2\n        └── ...\n```\n\nEach run is organized under the source file name (e.g., `spaceship-titanic` for `spaceship-titanic.py`) and a unique UUID. The `outputs/` directory and `exec_output.jsonl` file are only created when the `--save-logs` flag is used.\n\nThe `exec_output.jsonl` file contains one JSON object per line with:\n- `step`: The optimization step number\n- `timestamp`: When the execution occurred\n- `output_file`: Relative path to the full output file\n- `output_length`: Total length of the output\n\nThis is particularly useful for:\n- Debugging why certain optimizations fail\n- Analyzing patterns in evaluation results\n- Keeping records of long-running optimization sessions\n- Troubleshooting evaluation script issues\n\n---\n\n### Important Note on Evaluation\n\nThe command specified by `--eval-command` is crucial. It's responsible for executing the potentially modified code from `--source` (or `--sources`) and assessing its performance. **This command MUST print the metric you specified with `--metric` along with its numerical value to the terminal (standard output or standard error).** Weco reads this output to understand how well each code version performs and guide the optimization process.\n\nFor example, if you set `--metric speedup`, your evaluation script (`eval.py` in the examples) should output a line like:\n\n```\nspeedup: 1.5\n```\n\nor\n\n```\nFinal speedup value = 1.5\n```\n\nWeco will parse this output to extract the numerical value (1.5 in this case) associated with the metric name ('speedup').\n\n**Note on Output Truncation:** When evaluation output exceeds 51,000 characters, Weco truncates it to show the first 25,000 and last 25,000 characters. For best results, ensure your evaluation script prints the metric value near the end of its output.\n\n## Supported Models\n\nA list of models we support can be found in our documentation [here](https://docs.weco.ai/cli/supported-models).\n\n---\n\n## Contributing\n\nWe welcome contributions! Please see [contributing.md](contributing.md) for detailed guidelines on how to contribute to this project.\n\n---\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fwecoai%2Fweco-cli","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fwecoai%2Fweco-cli","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fwecoai%2Fweco-cli/lists"}