GitLab

* can commit

* can exec sm_90a

* can exec < sm_90

* fix format

* fix format

* 增加测试，测试对标sglang test

* fix format 1

* fix format 2

* add compile option to disable cutlass

This reverts commit 25258029.

* issue/739 - support batched RoPE on Nvidia and CPU

* issue/739 - metax, moore batched rope

* issue/739 - adjust metax flags

* issue/739 - added a rope module interface to forward inplace in output tensor

Signed-off-by: Ceng23333 <441651826@qq.com>

* ISSUE/628 适配QY C610 GPU，增加编译选项，适配已有算子。添加bge类模型所需的算子，包括gelu,layer_norm，lp_norm(支持l1，l2 norm)，relu，softmax，tanh。

---------
Co-authored-by: xgqdut2016 <kenan_gewei@163.com>
Co-authored-by: xgqdut2016 <140036308+xgqdut2016@users.noreply.github.com>

* issue/571 - introduced the DeviceEvent feature

---------
Co-authored-by: Jiacheng Huang <huangjiacheng0709@outlook.com>

Signed-off-by: Ceng23333 <441651826@qq.com>

Signed-off-by: Ceng23333 <441651826@qq.com>

- Implement core modules: Linear, Embedding, RMSNorm
- Add PyTorch-like macros for module and parameter definition
  - INFINICORE_NN_MODULE for single module declaration
  - INFINICORE_NN_MODULE_VEC for module vectors
  - INFINICORE_NN_PARAMETER for parameter declaration
  - Corresponding INIT macros for initialization
- Implement hierarchical module system with dynamic path generation
- Add state_dict() and load_state_dict() support
- Refactor module design: protected registration methods, removed path_ member
- Add comprehensive test suite including TinyLlama integration
- All parameters are protected with public accessors