格式化与代码
TensorFlow.js 模型加载器
Load any tf.js model URL — see layers, params, ops.
Tip: TF Hub URLs sometimes redirect — paste the final model.json URL.
编辑注
Understanding · Models in the browser — WebGL or WASM, your choice.
本深度章节目前仅有英文版本。上方的转换工具支持您的语言;长篇说明文章尚未翻译。
常见问题
Quick answers.
›What files are required to load a model?
You need a link to a valid `model.json` file. This file contains the architecture map and links to the associated binary weight files which will be fetched automatically.
›Does my model data stay private?
Yes. The model is downloaded from your provided URL directly to your browser's memory. No model data or weights are uploaded to our servers or stored permanently.
›Why is my model failing to load?
The most common issue is CORS (Cross-Origin Resource Sharing) restrictions on the host server. Ensure the hosting provider allows external requests or use a local development server with the correct headers.
›Can I run inference with this tool?
No. This tool is specifically for Inspecting the model's layers, input/output shapes, and operation count rather than executing data through the graph.
大家也在搜索
相关工具