rohitg00/ai-engineering-from-scratch · error · ValueError

sinusoidal_2d dim must be divisible by 4, got {dim}

Error message

sinusoidal_2d dim must be divisible by 4, got {dim}

What it means

Error "sinusoidal_2d dim must be divisible by 4, got {dim}" thrown in rohitg00/ai-engineering-from-scratch.

Source

Thrown at phases/19-capstone-projects/58-vision-encoder-patches/code/main.py:50

            raise ValueError(
                f"patch_size {self.patch_size} must divide image_size {self.image_size}"
            )
        return self.image_size // self.patch_size

    @property
    def num_patches(self) -> int:
        return self.grid_size * self.grid_size


def sinusoidal_2d(grid_h: int, grid_w: int, dim: int) -> torch.Tensor:
    """Build a deterministic 2D sinusoidal position table of shape (grid_h * grid_w, dim).

    Half of dim encodes row position, half encodes column position. Within each
    half, frequencies span the standard Transformer sin/cos band. Identical
    inputs always produce identical outputs, with no learned state.
    """
    if dim % 4 != 0:
        raise ValueError(f"sinusoidal_2d dim must be divisible by 4, got {dim}")
    half = dim // 2
    quarter = half // 2

    freq = torch.arange(quarter, dtype=torch.float32)
    inv = torch.exp(-math.log(10000.0) * freq / max(1, quarter))

    rows = torch.arange(grid_h, dtype=torch.float32).unsqueeze(1) * inv.unsqueeze(0)
    cols = torch.arange(grid_w, dtype=torch.float32).unsqueeze(1) * inv.unsqueeze(0)

    row_emb = torch.cat([torch.sin(rows), torch.cos(rows)], dim=1)
    col_emb = torch.cat([torch.sin(cols), torch.cos(cols)], dim=1)

    table = torch.zeros(grid_h, grid_w, dim)
    table[:, :, :half] = row_emb.unsqueeze(1).expand(-1, grid_w, -1)
    table[:, :, half:] = col_emb.unsqueeze(0).expand(grid_h, -1, -1)
    return table.reshape(grid_h * grid_w, dim)

View on GitHub (pinned to 39ea8a1c6d)

When it happens

Trigger: Thrown at phases/19-capstone-projects/58-vision-encoder-patches/code/main.py:50 when the library encounters an invalid state.

Common situations: See trigger scenarios.

AI-assisted analysis of rohitg00/ai-engineering-from-scratch@39ea8a1c6d (2026-08-26). Data as JSON: /api/errors/f6094dd9ededf5f2. Report an issue: GitHub.