Change logprobs to use int64 datatype in torch.gather

[pathorn]

Fix crash on DeepSeek R1 where torch.gather expects an int64 tensor.

I have confirmed that this fixes the crash on V1 on revision 233ffce , but I don't understand what caused the change in behavior. I suspect this regression happened after #12721 updating torch to 2.6.0, but I don't really understand the cause.

INFO 03-15 19:58:20 [logger.py:39] Received request cmpl-aca2df6d00644a15a3b1d0cbb1491f4a-0: prompt: '', params: SamplingParams(n=1, presence_penalty=0.0, frequency_penalty=0.0, repetition_penalty=1.0, temperature=0.6, top_p=0.95, top_k=-1, min_p=0.0, seed=3030081490969633974, stop=['<|user|>'], stop_token_ids=[], bad_words=[], include_stop_str_in_output=False, ignore_eos=False, max_tokens=8192, min_tokens=0, logprobs=1, prompt_logprobs=None, skip_special_tokens=True, spaces_between_special_tokens=True, truncate_prompt_tokens=None, guided_decoding=None, extra_args=None), prompt_token_ids: [], lora_request: None, prompt_adapter_request: None.

WorkerProc hit an exception: %s
Traceback (most recent call last):
  File "/opt/venv/lib/python3.12/site-packages/vllm/v1/executor/multiproc_executor.py", line 371, in worker_busy_loop
    output = func(*args, **kwargs)
             ^^^^^^^^^^^^^^^^^^^^^
  File "/opt/venv/lib/python3.12/site-packages/torch/utils/_contextlib.py", line 116, in decorate_context
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^
  File "/opt/venv/lib/python3.12/site-packages/vllm/v1/worker/gpu_worker.py", line 242, in execute_model
    output = self.model_runner.execute_model(scheduler_output)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/opt/venv/lib/python3.12/site-packages/torch/utils/_contextlib.py", line 116, in decorate_context
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^
  File "/opt/venv/lib/python3.12/site-packages/vllm/v1/worker/gpu_model_runner.py", line 1011, in execute_model
    sampler_output = self.model.sample(
                     ^^^^^^^^^^^^^^^^^^
  File "/opt/venv/lib/python3.12/site-packages/vllm/model_executor/models/deepseek_v2.py", line 706, in sample
    next_tokens = self.sampler(logits, sampling_metadata)
                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/opt/venv/lib/python3.12/site-packages/torch/nn/modules/module.py", line 1739, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/opt/venv/lib/python3.12/site-packages/torch/nn/modules/module.py", line 1750, in _call_impl
    return forward_call(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/opt/venv/lib/python3.12/site-packages/vllm/v1/sample/sampler.py", line 54, in forward
    self.gather_logprobs(raw_logprobs, num_logprobs, token_ids=sampled)
  File "/opt/venv/lib/python3.12/site-packages/vllm/v1/sample/sampler.py", line 163, in gather_logprobs
    token_logprobs = logprobs.gather(-1, token_ids)
                     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
RuntimeError: gather(): Expected dtype int64 for index

[github-actions]

👋 Hi! Thank you for contributing to the vLLM project.

💬 Join our developer Slack at https://slack.vllm.ai to discuss your PR in #pr-reviews, coordinate on features in #feat- channels, or join special interest groups in #sig- channels.

Just a reminder: PRs would not trigger full CI run by default. Instead, it would only run fastcheck CI which starts running only a small and essential subset of CI tests to quickly catch errors. You can run other CI tests on top of those by going to your fastcheck build on Buildkite UI (linked in the PR checks section) and unblock them. If you do not have permission to unblock, ping simon-mo or khluu to add you in our Buildkite org.

Once the PR is approved and ready to go, your PR reviewer(s) can run CI to test the changes comprehensively before merging.

To run CI, PR reviewers can either: Add ready label to the PR or enable auto-merge.

🚀

[mergify]

This pull request has merge conflicts that must be resolved before it can be
merged. Please rebase the PR, @pathorn.

https://docs.github.com/en/pull-requests/collaborating-with-pull-requests/working-with-forks/syncing-a-fork

[WoosukKwon]

@pathorn Superseded by #15049. Thanks for the PR!