Use pyupgrade and ruff to modernize LMDeploy Python Code

[windreamer]

No description provided.

[Copilot]

Pull request overview

This PR introduces Ruff/pyupgrade-driven modernization across LMDeploy’s Python codebase (type hints, imports, small refactors) and wires Ruff into the project’s linting workflow.

Changes:

• Add Ruff configuration (targeting Python 3.10+) and enable pyupgrade-related linting via UP rules.
• Replace legacy typing.* generics/Optional/Union usage with Python 3.10+ syntax (list[...], dict[...], X | None, collections.abc).
• Apply small pyupgrade refactors (e.g., f-strings, comprehensions, simplified decorators / caching helpers) and update pre-commit hooks.

Reviewed changes

Copilot reviewed 272 out of 272 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
pyproject.toml	Add Ruff configuration (line length, target version, lint selects/ignores).
.pre-commit-config.yaml	Add ruff-check hook; remove yapf hook.
lmdeploy/vl/utils.py	Modernize type hints to `X
lmdeploy/vl/time_series_utils.py	Minor file open argument simplification.
lmdeploy/vl/model/yi.py	Modernize collection type hints.
lmdeploy/vl/model/utils.py	Switch Callable/MutableSequence imports to collections.abc.
lmdeploy/vl/model/qwen3.py	Modernize type hints; simplify typing imports.
lmdeploy/vl/model/qwen.py	Modernize type hints to built-in generics.
lmdeploy/vl/model/phi3_vision.py	Modernize type hints to built-in generics.
lmdeploy/vl/model/molmo.py	Modernize type hints and isinstance(..., list) checks.
lmdeploy/vl/model/mllama.py	Modernize type hints to built-in generics.
lmdeploy/vl/model/minicpmv.py	Modernize type hints; update dict/list annotations.
lmdeploy/vl/model/llava_next.py	Modernize type hints to built-in generics.
lmdeploy/vl/model/llava_hf.py	Modernize type hints to built-in generics.
lmdeploy/vl/model/llava.py	Modernize type hints to built-in generics.
lmdeploy/vl/model/llama4.py	Modernize type hints to built-in generics.
lmdeploy/vl/model/internvl_llava.py	Modernize type hints to built-in generics.
lmdeploy/vl/model/internvl3_hf.py	Modernize Optional[...] to `X
lmdeploy/vl/model/interns1_pro.py	Modernize type hints and optional parameters.
lmdeploy/vl/model/glm4_v.py	Modernize type hints and isinstance(..., list) checks.
lmdeploy/vl/model/glm4_1v.py	Modernize type hints to built-in generics.
lmdeploy/vl/model/gemma3_vl.py	Modernize type hints; update kwargs TypedDict fields.
lmdeploy/vl/model/deepseek_vl2.py	Modernize type hints to built-in generics.
lmdeploy/vl/model/deepseek.py	Modernize type hints to built-in generics.
lmdeploy/vl/model/cogvlm.py	Modernize type hints to built-in generics.
lmdeploy/vl/model/builder.py	Modernize union type for backend config parameter.
lmdeploy/version.py	Modernize return type annotation (Tuple -> tuple).
lmdeploy/turbomind/tokenizer_info.py	Modernize type hints (lists/unions).
lmdeploy/turbomind/deploy/source_model/base.py	Modernize type hints; switch Iterator import to collections.abc.
lmdeploy/turbomind/deploy/policy.py	Modernize list return annotation.
lmdeploy/turbomind/deploy/loader.py	Modernize type hints and file open call.
lmdeploy/turbomind/deploy/config.py	Modernize type hints for dataclass config objects.
lmdeploy/turbomind/init.py	Simplify Windows DLL-path bootstrap logic.
lmdeploy/serve/proxy/proxy.py	Minor typing + comprehension modernization.
lmdeploy/serve/openai/tool_parser/utils.py	Modernize list/tuple return types.
lmdeploy/serve/openai/tool_parser/tool_parser.py	Modernize typing; use collections.abc.Sequence.
lmdeploy/serve/openai/tool_parser/qwen3_parser.py	Modernize typing; simplify class declaration.
lmdeploy/serve/openai/tool_parser/qwen2d5_parser.py	Modernize typing; use collections.abc.Sequence.
lmdeploy/serve/openai/tool_parser/llama3_parser.py	Modernize typing; use built-in generics.
lmdeploy/serve/openai/tool_parser/internlm2_parser.py	Modernize typing; use collections.abc.Sequence.
lmdeploy/serve/openai/reasoning_parser/reasoning_parser.py	Modernize typing; use collections.abc.Sequence.
lmdeploy/serve/openai/reasoning_parser/qwen_qwq_reasoning_parser.py	Modernize typing; use collections.abc.Sequence.
lmdeploy/serve/openai/reasoning_parser/deepseek_r1_reasoning_parser.py	Modernize typing; use collections.abc.Sequence.
lmdeploy/serve/openai/launch_server.py	Modernize typing (list unions, backend config union).
lmdeploy/serve/openai/harmony_utils.py	Modernize typing for token lists.
lmdeploy/serve/openai/api_server.py	Switch AsyncGenerator import to collections.abc.
lmdeploy/serve/managers/session_manager.py	Modernize history type annotation.
lmdeploy/pytorch/weight_loader/model_weight_loader.py	Simplify file open args (mode default).
lmdeploy/pytorch/utils.py	Modernize typing; use collections.abc.Sequence; optional types to `
lmdeploy/pytorch/transformers/init.py	Simplify @lru_cache() to @lru_cache.
lmdeploy/pytorch/tools/utils.py	Modernize list annotation.
lmdeploy/pytorch/strategies/dllm/sequence.py	Modernize typing aliases and optional annotations.
lmdeploy/pytorch/strategies/dllm/sampling.py	Modernize typing alias to built-in list.
lmdeploy/pytorch/strategies/dllm/model_agent.py	Modernize typing aliases and dict annotations.
lmdeploy/pytorch/strategies/base/sequence.py	Modernize typing alias and optional migration request.
lmdeploy/pytorch/strategies/base/sampling.py	Modernize typing alias to built-in list.
lmdeploy/pytorch/strategies/base/model_agent.py	Modernize typing aliases and dict annotations.
lmdeploy/pytorch/strategies/ar_spec/sequence.py	Modernize typing aliases and optional migration request.
lmdeploy/pytorch/strategies/ar_spec/model_agent.py	Modernize typing aliases and dict annotations.
lmdeploy/pytorch/strategies/ar/sequence.py	Modernize typing aliases and dict annotations.
lmdeploy/pytorch/strategies/ar/sampling.py	Modernize typing alias to built-in list.
lmdeploy/pytorch/strategies/ar/model_inputs.py	Modernize optional annotations to `
lmdeploy/pytorch/strategies/ar/model_agent.py	Modernize typing aliases and dict annotations.
lmdeploy/pytorch/spec_decode/reject_sampler.py	Modernize optional tensor typing to `Tensor
lmdeploy/pytorch/spec_decode/proposers/eagle3.py	Modernize dict annotation.
lmdeploy/pytorch/spec_decode/proposers/deepseek_mtp.py	Modernize dict annotation.
lmdeploy/pytorch/spec_decode/proposers/base.py	Modernize optional annotations and list typing.
lmdeploy/pytorch/spec_decode/base.py	Modernize dict annotation.
lmdeploy/pytorch/ray.py	Modernize dict/list annotations.
lmdeploy/pytorch/paging/scheduler.py	Modernize map/list type aliases.
lmdeploy/pytorch/paging/eviction_helper/recompute_eviction_helper.py	Modernize list typing.
lmdeploy/pytorch/paging/eviction_helper/base_eviction_helper.py	Modernize list typing.
lmdeploy/pytorch/paging/block_trie.py	Modernize dict/set annotations.
lmdeploy/pytorch/paging/block_manager/base_block_manager.py	Modernize dict annotation.
lmdeploy/pytorch/nn/norm.py	Modernize dict optional annotation.
lmdeploy/pytorch/nn/moe/w8a8.py	Modernize typing (lists/dicts/optionals).
lmdeploy/pytorch/nn/moe/route.py	Modernize tuple return annotation.
lmdeploy/pytorch/nn/moe/default.py	Modernize dict checks/types; Callable from collections.abc.
lmdeploy/pytorch/nn/moe/blocked_fp8.py	Modernize typing; dict checks/types.
lmdeploy/pytorch/nn/moe/init.py	Modernize typing; Callable from collections.abc.
lmdeploy/pytorch/nn/linear/base.py	Modernize typing; list generics.
lmdeploy/pytorch/multimodal/image_type.py	Modernize meta dict annotation.
lmdeploy/pytorch/multimodal/data_type.py	Modernize multimodal type aliases.
lmdeploy/pytorch/models/utils/multimodal.py	Modernize prepared input tuple alias.
lmdeploy/pytorch/models/utils/model.py	Modernize iterable/tuple typing.
lmdeploy/pytorch/models/siglip.py	Modernize iterable/set/tuple typing.
lmdeploy/pytorch/models/sdar.py	Modernize optional and iterable/tuple typing.
lmdeploy/pytorch/models/qwen3_vl_moe.py	Modernize typing for expert loading helpers.
lmdeploy/pytorch/models/qwen3_5_moe.py	Modernize typing for expert loading helpers.
lmdeploy/pytorch/models/qwen3.py	Modernize optional/iterable/tuple typing.
lmdeploy/pytorch/models/qwen2_reward.py	Modernize optional/iterable/tuple typing.
lmdeploy/pytorch/models/qwen2.py	Modernize optional/iterable/tuple typing.
lmdeploy/pytorch/models/qwen.py	Modernize optional/iterable/tuple typing.
lmdeploy/pytorch/models/phi3_v.py	Modernize optional/iterable typing; multimodal preprocess typing.
lmdeploy/pytorch/models/phi3.py	Modernize optional/iterable/tuple typing.
lmdeploy/pytorch/models/patch.py	Modernize dict typing and dict isinstance checks.
lmdeploy/pytorch/models/mistral.py	Modernize optional/iterable/tuple typing.
lmdeploy/pytorch/models/minicpm3.py	Modernize optional/iterable/tuple typing.
lmdeploy/pytorch/models/llama_eagle.py	Modernize optional/iterable/tuple typing.
lmdeploy/pytorch/models/internvl_patch.py	Modernize optional tensor typing.
lmdeploy/pytorch/models/interns1_pro_ts.py	Modernize optional/tuple typing; small call simplification.
lmdeploy/pytorch/models/internlm2_reward.py	Modernize iterable/tuple typing.
lmdeploy/pytorch/models/internlm.py	Modernize optional/iterable/tuple typing.
lmdeploy/pytorch/models/glm4.py	Modernize optional/iterable/tuple typing.
lmdeploy/pytorch/models/gemma3_vl.py	Modernize iterable/tuple typing; multimodal preprocess typing.
lmdeploy/pytorch/models/deepseek_vl2.py	Modernize optional/iterable/tuple typing; multimodal preprocess typing.
lmdeploy/pytorch/models/deepseek_v32.py	Modernize Sequence import and tuple typing.
lmdeploy/pytorch/kernels/dlinfer/pagedattention.py	Modernize optional/sequence typing for kernel wrappers.
lmdeploy/pytorch/kernels/dlinfer/moe_gating_topk_softmax.py	Modernize tuple return typing.
lmdeploy/pytorch/kernels/dlinfer/linear.py	Modernize optional tensor typing.
lmdeploy/pytorch/kernels/dlinfer/fill_kv_cache.py	Modernize optional tensor sequence typing.
lmdeploy/pytorch/kernels/dlinfer/awq_kernels.py	Modernize optional tensor typing.
lmdeploy/pytorch/kernels/dlinfer/apply_rotary_pos_emb.py	Modernize optional/tuple typing.
lmdeploy/pytorch/kernels/dispatcher.py	Switch Callable import to collections.abc.
lmdeploy/pytorch/kernels/cuda/rms_norm.py	Modernize print formatting to f-strings.
lmdeploy/pytorch/kernels/cuda/pagedattention.py	Switch Sequence import to collections.abc.
lmdeploy/pytorch/kernels/cuda/gated_delta_rule.py	Switch Sequence import to collections.abc.
lmdeploy/pytorch/kernels/cuda/fused_moe_ep.py	Modernize optional list typing.
lmdeploy/pytorch/kernels/cuda/fused_moe.py	Minor tensor allocation call simplification.
lmdeploy/pytorch/kernels/cuda/flashattention.py	Minor constant expression cleanup.
lmdeploy/pytorch/kernels/cuda/fill_kv_cache.py	Modernize optional tensor typing and optional string typing.
lmdeploy/pytorch/kernels/cuda/blocked_gemm_fp8.py	Modernize optional string typing.
lmdeploy/pytorch/kernels/cuda/blocked_fp8_fused_moe.py	Switch Callable import to collections.abc.
lmdeploy/pytorch/envs.py	Modernize union typing for parameter defaults.
lmdeploy/pytorch/engine/mp_engine/ray_engine.py	Modernize dict typing.
lmdeploy/pytorch/engine/mp_engine/base_worker.py	Modernize optional list typing.
lmdeploy/pytorch/engine/mp_engine/base.py	Modernize optional list typing.
lmdeploy/pytorch/engine/model_agent/init.py	Modernize adapters dict typing.
lmdeploy/pytorch/engine/logits_process.py	Modernize typing of internal dataclasses and lists/dicts.
lmdeploy/pytorch/engine/input_process.py	Modernize type aliases and dataclass field typing.
lmdeploy/pytorch/engine/guided_process.py	Modernize typing; dict checks/types.
lmdeploy/pytorch/engine/executor/uni_executor.py	Modernize adapters and list typing.
lmdeploy/pytorch/engine/executor/base_worker.py	Modernize adapters, dict output typing, and optional list typing.
lmdeploy/pytorch/engine/executor/base.py	Modernize adapters and optional list typing.
lmdeploy/pytorch/engine/executor/init.py	Modernize adapters dict typing.
lmdeploy/pytorch/engine/engine_instance.py	Modernize multimodal typing and input id list typing.
lmdeploy/pytorch/distributed.py	Modernize optional/list typing throughout dist helpers.
lmdeploy/pytorch/disagg/messages.py	Modernize nested list/tuple typing; optional config typing.
lmdeploy/pytorch/disagg/conn/proxy_conn.py	Modernize set/dict/tuple typing for connection pool.
lmdeploy/pytorch/disagg/conn/protocol.py	Modernize optional/list typing for protocol models.
lmdeploy/pytorch/disagg/conn/engine_conn.py	Modernize dict/list typing for sockets/contexts.
lmdeploy/pytorch/disagg/config.py	Modernize optional typing.
lmdeploy/pytorch/disagg/backend/mooncake.py	Modernize dict typing for endpoint/link bookkeeping.
lmdeploy/pytorch/disagg/backend/dlslime.py	Modernize dict typing for endpoints/links.
lmdeploy/pytorch/devices/device_manager.py	Switch Callable import to collections.abc.
lmdeploy/pytorch/configurations/init.py	Switch to f-string module import format.
lmdeploy/pytorch/check_env/base.py	Modernize list typing for checker list.
lmdeploy/pytorch/backends/token_dispatcher.py	Modernize tuple return typing.
lmdeploy/pytorch/backends/rotary_embedding.py	Modernize list typing for scaling parameters.
lmdeploy/pytorch/backends/qmodules.py	Modernize optional typing for bias/group args.
lmdeploy/pytorch/backends/moe_router.py	Modernize tuple return typing.
lmdeploy/pytorch/backends/moe.py	Modernize typing; optional string typing.
lmdeploy/pytorch/backends/linear.py	Modernize optional/list typing.
lmdeploy/pytorch/backends/graph_runner.py	Modernize list typing for past key values and batch sizes.
lmdeploy/pytorch/backends/dlinfer/qmodules.py	Modernize optional typing for bias/group args.
lmdeploy/pytorch/backends/dlinfer/op_backend.py	Modernize tuple return typing.
lmdeploy/pytorch/backends/dlinfer/moe.py	Modernize list typing and callable import.
lmdeploy/pytorch/backends/dlinfer/maca/op_backend.py	Modernize tuple return typing.
lmdeploy/pytorch/backends/dlinfer/linear.py	Modernize optional/list typing.
lmdeploy/pytorch/backends/dlinfer/camb/op_backend.py	Modernize tuple return typing.
lmdeploy/pytorch/backends/dlinfer/awq_modules.py	Modernize optional typing for bias/group args.
lmdeploy/pytorch/backends/dlinfer/attention.py	Modernize optional typing; Sequence from collections.abc.
lmdeploy/pytorch/backends/dlinfer/ascend/op_backend.py	Modernize dict/tuple typing and file open args.
lmdeploy/pytorch/backends/default/token_dispatcher.py	Modernize tuple return typing.
lmdeploy/pytorch/backends/default/op_backend.py	Modernize tuple return typing.
lmdeploy/pytorch/backends/default/moe_router.py	Modernize tuple return typing.
lmdeploy/pytorch/backends/default/linear.py	Modernize optional/list typing.
lmdeploy/pytorch/backends/default/awq_modules.py	Modernize optional typing for bias/group args.
lmdeploy/pytorch/backends/cuda/qmodules.py	Modernize optional typing for bias/group args.
lmdeploy/pytorch/backends/cuda/op_backend.py	Modernize tuple return typing.
lmdeploy/pytorch/backends/cuda/moe_router.py	Modernize tuple return typing.
lmdeploy/pytorch/backends/cuda/moe/w8a8.py	Modernize list typing.
lmdeploy/pytorch/backends/cuda/moe/ep_utils.py	Modernize list typing.
lmdeploy/pytorch/backends/cuda/moe/default.py	Modernize list typing and optional typing.
lmdeploy/pytorch/backends/cuda/moe/blocked_fp8.py	Modernize list typing and callable import.
lmdeploy/pytorch/backends/cuda/graph_runner.py	Modernize dict/list typing and tuple typing.
lmdeploy/pytorch/backends/cuda/blockedf8_modules.py	Modernize optional/list typing.
lmdeploy/pytorch/backends/cuda/awq_modules.py	Modernize optional typing for bias/group args.
lmdeploy/pytorch/backends/cuda/attention/mla.py	Replace lru_cache(maxsize=None) with functools.cache.
lmdeploy/pytorch/backends/blockedf8_modules.py	Modernize optional/list typing.
lmdeploy/pytorch/backends/base.py	Modernize tuple return typing.
lmdeploy/pytorch/backends/awq_modules.py	Modernize optional typing for bias/group args.
lmdeploy/pytorch/adapter/adapter.py	Modernize dict/list typing; iterable typing.
lmdeploy/profiler.py	Modernize list typing for collected metrics.
lmdeploy/metrics/stats.py	Modernize list typing and optional float typing.
lmdeploy/metrics/metrics_processor.py	Simplify class declaration syntax.
lmdeploy/metrics/loggers.py	Modernize list typing for bucket builders.
lmdeploy/logger.py	Modernize optional/list typing for request logging.
lmdeploy/lite/utils/memory_efficient.py	Modernize list typing in helper functions.
lmdeploy/lite/utils/global_avail.py	Modernize nested dict typing and unions.
lmdeploy/lite/utils/collect.py	Modernize dict/list/tuple typing throughout helpers.
lmdeploy/lite/utils/cal_qparams.py	Modernize optional typing in NamedTuple.
lmdeploy/lite/utils/batch_split.py	Modernize unions/tuples and Tuple -> tuple checks.
lmdeploy/lite/quantization/weight/quantizer.py	Modernize typing; nested dict typing for function maps.
lmdeploy/lite/quantization/weight/quant_utils.py	Modernize optional/union typing for helpers.
lmdeploy/lite/quantization/modules/linear.py	Modernize typing; update classmethod cls annotation.
lmdeploy/lite/quantization/calibration.py	Simplify class declarations; modernize union typing.
lmdeploy/lite/quantization/awq.py	Modernize list typing for smoothing helpers.
lmdeploy/lite/apis/get_small_sharded_hf.py	Replace .format(...) with f-string.
lmdeploy/lite/apis/calibrate.py	Modernize union typing; minor set construction simplification.
lmdeploy/cli/utils.py	Modernize list typing for CLI utilities.
lmdeploy/cli/lite.py	Simplify class declaration syntax.
lmdeploy/cli/cli.py	Simplify subprocess call using capture_output=True.
lmdeploy/archs.py	Modernize tuple return typing; simplify set literal; fix type comparison to is.
lmdeploy/api.py	Modernize typing by removing string-literal types where possible; update list typing.

Comments suppressed due to low confidence (1)

lmdeploy/lite/quantization/modules/linear.py:36

• The bias parameter is used as a boolean flag (if bias: and set to True/False), but it is annotated as torch.Tensor | None and defaults to True. This annotation is incorrect and makes the API misleading for callers/type checkers. Consider changing the type to bool (and updating the docstring accordingly).

        self,
        in_features: int,
        out_features: int,
        bias: torch.Tensor | None = True,
        w_bit: int = 4,
        symmetry: bool = False,
        group_size: int = 128,
    ) -> None:

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.