@@ -70,7 +70,7 @@ Best current local stack. Load a large model in LM Studio (for example, a full-s
7070**Setup checklist**
71717272- Install LM Studio: [https://lmstudio.ai](https://lmstudio.ai)
73-- In LM Studio, download the **largest model build available** (avoid “small”/heavily quantized variants), start the server, confirm `http://127.0.0.1:1234/v1/models` lists it.
73+- In LM Studio, download the **largest model build available** (avoid "small"/heavily quantized variants), start the server, confirm `http://127.0.0.1:1234/v1/models` lists it.
7474- Replace `my-local-model` with the actual model ID shown in LM Studio.
7575- Keep the model loaded; cold-load adds startup latency.
7676- Adjust `contextWindow`/`maxTokens` if your LM Studio build differs.
@@ -317,7 +317,7 @@ If the model loads cleanly but full agent turns misbehave, work top-down — con
317317## Troubleshooting
318318319319- Gateway can reach the proxy? `curl http://127.0.0.1:1234/v1/models`.
320-- LM Studio model unloaded? Reload; cold start is a common “hanging” cause.
320+- LM Studio model unloaded? Reload; cold start is a common "hanging" cause.
321321- Local server says `terminated`, `ECONNRESET`, or closes the stream mid-turn?
322322 OpenClaw records a low-cardinality `model.call.error.failureKind` plus the
323323 OpenClaw process RSS/heap snapshot in diagnostics. For LM Studio/Ollama
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。