Jim Gu

（狗头保命 pkg_resources 库已废弃，应替换为 importlib

In Chapter 18, we have the `train` method as: ```Python def train( self, X: np.ndarray, y: np.ndarray, *, validation: tuple[np.ndarray, np.ndarray] = None, epochs: int = 100, print_every: int =...

Thanks for the code! In `models.py` line 104 and line 118, I saw two missing `nn.LeakyReLU` after convolution blocks. To my knowledge, the conventions write it that each convolution block...

[Feature] Optimize Requirements File

1

### Motivation Currently, all requirements for InternVL codebase rely on standalone `requirements.txt` files, in which, the version information is mostly incomplete. I would strongly suggest completing the version information for...

Thanks for open-sourcing this wonderful work. I attempt to reproduce the results but get stuck in running the demo. The error info suggests the GPU architecture is not compatible with...

## Describe the bug A clear and concise description of what the bug is. 使用中文也可以。 As the title suggests, `nn.AdaptiveAvgPool2d` does not accept 3D `jittor.Var` as opposed to `torch.nn.functional.adaptive_avg_pool_2d`. ##...

## Describe the bug A clear and concise description of what the bug is. 使用中文也可以。 The `nn.AdaptiveAvgPool2d` does not accept input spatial size (H, W) smaller than output shape (`output_shape`)....

## Describe the bug A clear and concise description of what the bug is. 使用中文也可以。 Even though the documentation specifies the correct usage of `jt.transpose`, for convenience though, hovering over...

## Describe the bug A clear and concise description of what the bug is. 使用中文也可以。 The type hint for `x.shape` indicates it's a method (`x` is a `jt.Var`). However, in...

Thanks for the great work! The core logic of the proposed GRPB sits in the function `global_position_index`, about which I have a few questions. The intention of this function is...