convert.py

#!pip install uv
#!uv pip install transformers torch onnx
# Импорты
from transformers import AutoTokenizer, AutoModel
import torch
import onnx

tokenizer = AutoTokenizer.from_pretrained("ai-forever/sbert_large_nlu_ru")
model = AutoModel.from_pretrained("ai-forever/sbert_large_nlu_ru")

# Экспорт в ONNX
dummy_input = tokenizer("Тест", return_tensors="pt", padding=True, truncation=True)
torch.onnx.export(model,
                  (dummy_input['input_ids'], dummy_input['attention_mask']),
                  "sbert_large_nlu_ru.onnx",
                  input_names=["input_ids", "attention_mask"],
                  output_names=["last_hidden_state"],
                  dynamic_axes={
                      "input_ids": {0: "batch", 1: "sequence"},
                      "attention_mask": {0: "batch", 1: "sequence"},
                      "last_hidden_state": {0: "batch", 1: "sequence", 2: "features"}
                  })
# from google.colab import files
# files.download("/content/sbert_large_nlu_ru.onnx")