docs: explain slow nvidia custom providers · openclaw/openclaw@759965a
steipete
·
2026-05-07
·
via Recent Commits to openclaw:main
| Original file line number | Diff line number | Diff line change |
|---|
@@ -88,6 +88,38 @@ openclaw onboard --auth-choice nvidia-api-key --nvidia-api-key "nvapi-..."
|
88 | 88 | NVIDIA uses the standard `/v1` completions endpoint. Any OpenAI-compatible |
89 | 89 | tooling should work out of the box with the NVIDIA base URL. |
90 | 90 | </Accordion> |
| 91 | + |
| 92 | +<Accordion title="Slow custom provider responses"> |
| 93 | +Some NVIDIA-hosted custom models can take longer than the default model idle |
| 94 | +watchdog before they emit a first response chunk. For custom NVIDIA provider |
| 95 | +entries, raise the provider timeout instead of raising the whole agent |
| 96 | +runtime timeout: |
| 97 | + |
| 98 | +```json5 |
| 99 | +{ |
| 100 | + models: { |
| 101 | + providers: { |
| 102 | + "custom-integrate-api-nvidia-com": { |
| 103 | + baseUrl: "https://integrate.api.nvidia.com/v1", |
| 104 | + api: "openai-completions", |
| 105 | + apiKey: "NVIDIA_API_KEY", |
| 106 | + timeoutSeconds: 300, |
| 107 | + }, |
| 108 | + }, |
| 109 | + }, |
| 110 | + agents: { |
| 111 | + defaults: { |
| 112 | + models: { |
| 113 | + "custom-integrate-api-nvidia-com/meta/llama-3.1-70b-instruct": { |
| 114 | + params: { thinking: "off" }, |
| 115 | + }, |
| 116 | + }, |
| 117 | + }, |
| 118 | + }, |
| 119 | +} |
| 120 | +``` |
| 121 | + |
| 122 | +</Accordion> |
91 | 123 | </AccordionGroup> |
92 | 124 | |
93 | 125 | <Tip> |
|
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。
