# xingjian-bai/opal

> A PyTorch optimizer wrapper that lets a coding agent rewrite the optimizer program during training. It is for ML engineers who want the update rule, learning rates, and schedules to adapt automatically while a model trains.

- Magnitude: 4.4 out of 10 — Early signal
- Stars: 125 total · +8 stars measured 2026-10-22, 09:11–11:27 UTC, ≈ 37 by evening
- Star trust: star growth looks organic
- Category: AI agents · Language: Python · License: MIT · Created: 2026-10-21 · Last push: 2026-10-22
- GitHub: https://github.com/xingjian-bai/opal · Homepage: https://xingjianbai.com/reactive-optimizers/ · Page: https://gitnova.dev/en/r/xingjian-bai/opal

## Useful for

- Wrap an existing PyTorch training loop to let an agent adjust learning rates and momentum during a run.
- Run a dry-run with backend="keep" to verify the wrapped optimizer matches the plain run step for step.
- Estimate agent waiting time and cost before launching a real run with a cost cap.

## Why it’s here

- Star-counter measurements on 2026-10-22 (UTC), 09:11–11:27: 117 → 125 stars (+8). This is the change over that interval.
- Estimated end-of-day forecast: about +37 stars, using observed gains and the previous day.
- The repository is 1 day old and already has 125 stars. With less than two weeks of history, there's no usual pace to compare the spike against yet.
- Top new repositories this week: #186.

## Star trust

Star growth looks organic.

Star-trust labels are heuristics based on the repository’s behavior, not a check of every stargazer.

## Numbers

- Forks: 4
- Issues and pull requests: 6
- Watchers: 1
- Average over the last week: 66 per day
- Usual pace: too little history (under two weeks)
- Stars in the last hour (measured): 1
- Latest release: v0.3.5 (2026-10-22)

## Stars per day, last 2 days (oldest → newest, today is partial)

2026-10-21 … 2026-10-22: 95, 8

## Spotted in now

- Top new repositories this week: #186

## More in this category

1. **google-research/harness2** — 7.9 · Peaking · AI agents · Python · +429 stars measured 2026-10-22, 00:37–11:25 UTC, ≈ 1,017 by evening
   A framework for recursive optimization of AI agent harnesses (prompts, tools, memory, skills) without fine-tuning model weights, with JobBench, LAB and WorkBuddy-Bench benchmarks.
   Full card: https://gitnova.dev/en/r/google-research/harness2.md
2. **Graphify-Labs/graphify** — 7.0 · Breakout · AI agents · Python · +616 stars measured 2026-10-22, 00:37–11:25 UTC, ≈ 1,306 by evening
   Turns any codebase, docs, SQL schemas, configs, and PDFs into a queryable knowledge graph. Works as a /graphify skill in Claude Code, Cursor, Codex, and Gemini CLI with local deterministic AST parsing.
   Full card: https://gitnova.dev/en/r/Graphify-Labs/graphify.md
3. **ai-job-search-dev/ai-job-search** — 7.0 · Peaking · AI agents · Python · +286 stars measured 2026-10-22, 00:37–11:25 UTC, ≈ 884 by evening
   A job application framework built on Claude Code: evaluates postings, tailors CVs, and writes cover letters. Runs locally, focused on the Danish job market.
   Full card: https://gitnova.dev/en/r/ai-job-search-dev/ai-job-search.md
4. **omnara-ai/omnara** — 6.5 · Breakout · AI agents · Go · +133 stars measured 2026-10-22, 00:01–11:26 UTC, ≈ 230 by evening
   Omnara is an open-source, self-hostable platform for running managed AI agents on any model and any machine, with tools, secrets, permissions, and streaming built in. It serves as an alternative to Claude Managed Agents and OpenAI's…
   Full card: https://gitnova.dev/en/r/omnara-ai/omnara.md
5. **bowenliang123/dsh-context** — 6.2 · Breakout · AI agents · TypeScript · +185 stars measured 2026-10-22, 00:37–11:26 UTC, ≈ 396 by evening
   A DeepSeek Harness plugin that visualizes an agent's context: dashboard, browser, sidebar panel and /context command showing composition, stats and evolution of the context window.
   Full card: https://gitnova.dev/en/r/bowenliang123/dsh-context.md

---

Magnitude (0–10) measures how fast and how unusually interest in a repository is growing right now. It is not a quality score.
Days are UTC. “So far today” is a fact; “expected by the end of the day” is a forecast.
Summaries and use cases are written by an LLM (DeepSeek V4.1 Flash) from the README and may be inaccurate: verify specific claims (benchmarks, speed, hardware) in the repository itself.
Data as of 2026-10-22 11:38 UTC, updated every 30 minutes.