Add transformers v4.49 support

[echarlaix]

No description provided.

[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.

[echarlaix]

Failing tests

onnxruntime.capi.onnxruntime_pybind11_state.NotImplemented: [ONNXRuntimeError] : 9 : NOT_IMPLEMENTED : Could not find an implementation for ArgMax(13) node with name '/ArgMax'

will be fixed in huggingface/transformers#36332

[xenova]

I've tested converting LlamaForCausalLM models, but I'm running into this error:

Detailed error: Required inputs (['onnx::Gather_7']) are missing from input feed (['input_ids', 'attention_mask', 'position_ids', 'past_key_values.0.key', 'past_key_values.0.value', 'past_key_values.1.key', 'past_key_values.1.value'])..

Command:

optimum-cli export onnx -m hf-internal-testing/tiny-random-LlamaForCausalLM /tmp

I assume onnx::Gather_7 is related to logits_to_keep?

[xenova]

Ahh nevermind!! I was on an old version of Optimum 🤦