Add DeepSeek R1 WebGPU demo

[Jiawei-Shao]

No description provided.

[Jiawei-Shao]

@ibelem PTAL, thanks!

[NingW101]

@Jiawei-Shao I found code not format following the rules defined in prettierrc, could you please format them, thanks

[Jiawei-Shao]

@Jiawei-Shao I found code not format following the rules defined in prettierrc, could you please format them, thanks

Done @NingW101

[NingW101]

@Jiawei-Shao I found code not format following the rules defined in prettierrc, could you please format them, thanks

Done @NingW101

Thanks. I'll update UI based on your commits

[NingW101]

Suggested change

	if (!VITE_ENV_USE_REMOTE_MODELS) {
	env.backends.onnx.wasm.wasmPaths = `${baseUrl}/models/frameworks/ort-web/ort-web@transformers_js_3_3_1/`;
	env.allowLocalModels = true;
	env.allowRemoteModels = false;
	env.localModelPath = `${baseUrl}/models/`;
	} else {
	env.allowLocalModels = false;
	env.allowRemoteModels = true;
	}
	env.allowLocalModels = true;
	env.localModelPath = `${baseUrl}/models/`;

These options enable users to fetch .json and .onnx files from both local and remote sources. The priority is to fetch from local storage first, and if unavailable, fall back to the remote source. Since users will be uploading .onnx models themselves, we only need to allow fetching .json files (such as tokenizer configurations) from local server.

[Jiawei-Shao]

Done

[NingW101]

Since the entire models directory will not be copied only in --use-remote-models mode, we need to explicitly copy the resources required by the deepseek sample. Otherwise, deepseek won’t be able to fetch them from the deployed github.io server.

You can push "models/onnx-communify/DeepSeek-R1.." folder string into the RESOURCES_ARRAY or define another array variable const EXTRA_RESOURCE_ARRAY= ["models/onnx-communify/DeepSeek-R1.."] then execute the copy command as the

for (let path of EXTRA_RESOURCE_ARRAY) {
      if (path)
        execSync(
          `powershell -Command "Copy-Item -Path "${path}" -Destination "dist/${path}" -Recurse -Force"`,
          { stdio: "inherit" }
        );
    }

[Jiawei-Shao]

Done

[ibelem]

Thanks @Jiawei-Shao and @NingW101 's great efforts!

[ibelem]

@Jiawei-Shao Could you please fix the two linting issues? Thanks

[ibelem]

@Jiawei-Shao Thanks for fixing the @transformers_js_3_3_1/ort-wasm-simd-threaded.jsep.mjs linting issue, please also fix:

/home/runner/work/web-ai-showcase/web-ai-showcase/samples/deepseek-r1-webgpu/src/worker.js
  52:5   error  'past_key_values_cache' is assigned a value but never used                                                                     no-unused-vars
  64:[10](https://github.com/intel/web-ai-showcase/actions/runs/13406308241/job/37446789576?pr=74#step:4:11)  error  'START_THINKING_TOKEN_ID' is assigned a value but never used. Allowed unused elements of array destructuring must match /^_/u  no-unused-vars

✖ 2 problems (2 errors, 0 warnings)

[ibelem]

LGTM