Add support for HRM-Text

[xenova]

import { AutoModelForCausalLM, AutoTokenizer, TextStreamer } from "@huggingface/transformers";

const model_id = "onnx-community/HRM-Text-1B-ONNX";

const tokenizer = await AutoTokenizer.from_pretrained(model_id);
const model = await AutoModelForCausalLM.from_pretrained(model_id, {
  device: "webgpu",
  dtype: "fp16",
});

// synth,cot composite — reasoning / CoT style
const condition = "<|quad_end|><|object_ref_end|>";
const prompt = `<|im_start|>${condition}Explain why the sky is blue.<|im_end|>`;

const inputs = tokenizer(prompt);

// Generate a response
const start = performance.now();
const output = await model.generate({
  ...inputs,
  max_new_tokens: 256,
  do_sample: false,
  streamer: new TextStreamer(tokenizer, {
    skip_prompt: false,
    skip_special_tokens: false,
  }),
});

const decoded = tokenizer.batch_decode(output, { skip_special_tokens: false });
console.log(decoded[0]);

const end = performance.now();
console.log(`Inference took ${(end - start).toFixed(2)} ms`);

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.