Incompatible with transformers>=4.39

[jvamvas]

The decoding code is currently not compatible with versions of Hugging Face transformers >= v4.39:

======================================================================
ERROR: test_generate (test_generate.DecoderOnlyTestCase.test_generate)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/runner/work/mbr/mbr/tests/test_generate.py", line 28, in test_generate
    output = self.model.generate(
             ^^^^^^^^^^^^^^^^^^^^
  File "/opt/hostedtoolcache/Python/3.11.9/x64/lib/python3.11/site-packages/torch/utils/_contextlib.py", line 115, in decorate_context
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^
  File "/opt/hostedtoolcache/Python/3.11.9/x64/lib/python3.11/site-packages/mbr/generation/utils.py", line 344, in generate
    generation_mode = self._get_generation_mode(generation_config, assistant_model)
                      ^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/opt/hostedtoolcache/Python/3.11.9/x64/lib/python3.11/site-packages/torch/nn/modules/module.py", line 1709, in __getattr__
    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
AttributeError: 'MBRGPT2LMHeadModel' object has no attribute '_get_generation_mode'

[cgr71ii]

Hi! I could make it work using transformers==4.43.3

diff src/mbr/generation/utils.py local_installation/src/mbr/generation/utils.py:

18d17
< from mbr.metrics import load_metric_runner
227,229d225
<         if generation_config.num_return_sequences > 1:
<             raise ValueError("MBR decoding does not support `num_return_sequences` > 1.")
< 
347c343
<         generation_mode = self._get_generation_mode(generation_config, assistant_model)
---
>         generation_mode = generation_config.get_generation_mode(assistant_model)
387a384,388
>         if not hasattr(generation_config, "_eos_token_tensor"):
>             setattr(generation_config, "_eos_token_tensor", torch.tensor([generation_config.eos_token_id]).to(inputs_tensor.device))
>         if not hasattr(generation_config, "_pad_token_tensor"):
>             setattr(generation_config, "_pad_token_tensor", torch.tensor([generation_config.eos_token_id]).to(inputs_tensor.device))
> 
414c415
<             logits_warper = self._get_logits_warper(generation_config)
---
>             logits_warper = self._get_logits_warper(generation_config, inputs_tensor.device,)
432c433
<                 samples.append(self.sample(
---
>                 samples.append(self._sample(
436a438
>                     generation_config=generation_config,
457c459
<                 logits_warper = self._get_logits_warper(references_config)
---
>                 logits_warper = self._get_logits_warper(references_config, inputs_tensor.device)
467c469
<                     references.append(self.sample(
---
>                     references.append(self._sample(
471a474
>                         generation_config=generation_config,
483c486
<             metric_runner = load_metric_runner(mbr_config, tokenizer)
---
>             metric_runner = MetricRunner(mbr_config, tokenizer)