Upgrade VLLM dependency to support transformers 4.54.0+ (AIMv2 registration conflict)

[ExpressGradient]

Problem

ART currently uses VLLM 0.9.1, which conflicts with transformers 4.54.0+ due to duplicate model registration for "aimv2". This prevents using ART with the latest transformers version.

Error

Full traceback:

INFO 07-27 06:35:51 [init.py:244] Automatically detected platform cuda.
Traceback (most recent call last):
  File "/home/ubuntu/atreya/main.py", line 37, in <module>
    asyncio.run(train())
  File "/home/ubuntu/atreya/.venv/lib/python3.12/site-packages/nest_asyncio.py", line 30, in run
    return loop.run_until_complete(task)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/ubuntu/atreya/.venv/lib/python3.12/site-packages/nest_asyncio.py", line 98, in run_until_complete
    return f.result()
           ^^^^^^^^^^
  File "/home/ubuntu/.local/share/uv/python/cpython-3.12.11-linux-x86_64-gnu/lib/python3.12/asyncio/futures.py", line 202, in result
    raise self._exception.with_traceback(self._exception_tb)
  File "/home/ubuntu/.local/share/uv/python/cpython-3.12.11-linux-x86_64-gnu/lib/python3.12/asyncio/tasks.py", line 314, in step_run_and_handle_result
    result = coro.send(None)
             ^^^^^^^^^^^^^^^
  File "/home/ubuntu/atreya/main.py", line 14, in train
    await model.register(backend)
  File "/home/ubuntu/atreya/.venv/lib/python3.12/site-packages/art/model.py", line 307, in register
    base_url, api_key = await backend._prepare_backend_for_training(
                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/ubuntu/atreya/.venv/lib/python3.12/site-packages/art/local/backend.py", line 250, in preparebackend_for_training
    service = await self._get_service(model)
              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/ubuntu/atreya/.venv/lib/python3.12/site-packages/art/local/backend.py", line 113, in getservice
    from ..torchtune.service import TorchtuneService
  File "/home/ubuntu/atreya/.venv/lib/python3.12/site-packages/art/torchtune/service.py", line 14, in <module>
    from vllm import AsyncEngineArgs
  File "/home/ubuntu/atreya/.venv/lib/python3.12/site-packages/vllm/init.py", line 13, in <module>
    from vllm.engine.arg_utils import AsyncEngineArgs, EngineArgs
  File "/home/ubuntu/atreya/.venv/lib/python3.12/site-packages/vllm/engine/arg_utils.py", line 22, in <module>
    from vllm.config import (BlockSize, CacheConfig, CacheDType, CompilationConfig,
  File "/home/ubuntu/atreya/.venv/lib/python3.12/site-packages/vllm/config.py", line 43, in <module>
    from vllm.transformers_utils.config import (
  File "/home/ubuntu/atreya/.venv/lib/python3.12/site-packages/vllm/transformers_utils/config.py", line 33, in <module>
    from vllm.transformers_utils.configs import (ChatGLMConfig, Cohere2Config,
  File "/home/ubuntu/atreya/.venv/lib/python3.12/site-packages/vllm/transformers_utils/configs/init__.py", line 28, in <module>
    from vllm.transformers_utils.configs.ovis import OvisConfig
  File "/home/ubuntu/atreya/.venv/lib/python3.12/site-packages/vllm/transformers_utils/configs/ovis.py", line 76, in <module>
    AutoConfig.register("aimv2", AIMv2Config)
  File "/home/ubuntu/atreya/.venv/lib/python3.12/site-packages/transformers/models/auto/configuration_auto.py", line 1306, in register
    CONFIG_MAPPING.register(model_type, config, exist_ok=exist_ok)
  File "/home/ubuntu/atreya/.venv/lib/python3.12/site-packages/transformers/models/auto/configuration_auto.py", line 993, in register
    raise ValueError(f"'{key}' is already used by a Transformers config, pick another name.")
ValueError: 'aimv2' is already used by a Transformers config, pick another name.

Root Cause

• Transformers 4.54.0 officially added AIMv2 model support
• VLLM 0.9.1 tries to register "aimv2" itself, causing a conflict

Current Workaround

Downgrading to transformers<4.54.0 works.

Proposed Solution

Upgrade VLLM dependency to a version that's compatible with transformers 4.54.0+.

Environment

• OS: Ubuntu
• Python: 3.12.11
• ART version: 0.4.4
• Current VLLM: 0.9.1
• Transformers: 4.53.0

Additional Context

This affects anyone trying to use ART with the latest transformers features and new model architectures introduced in recent releases.

[zieen]

any progress

[corbt]

In the latest version 0.4.5 of this library we're now up to vllm 0.10.0, does that solve your problem?

[ExpressGradient]

closing the issue as its solved by upgrading art to latest version