Enum VisionTower 

pub enum VisionTower {
    Qwen(Box<VisionEncoder>),
    Glm(Box<GlmVit>),
}

Variants

Qwen(Box<VisionEncoder>)

2026-09-25: Built by weight_loader::qwen3_vl and weight_loader::qwen35 (which the qwen35_dense and qwen4_exp loaders call).

Glm(Box<GlmVit>)

2026-09-25: GLM-5.3-Flash (glm5_next).

Implementations

impl VisionTower

pub fn qwen(encoder: VisionEncoder) -> Self

pub fn glm(encoder: GlmVit) -> Self

pub fn forward_batched( &self, images: &[(&[f32], usize, usize)], gpu: &dyn GpuBackend, stream: u64, ) -> Result<Vec<(usize, usize, usize)>>

2026-09-25: Encode a batch of (pixels, grid_h, grid_w) temporal groups. Returns (post_merge_h, post_merge_w, merged_patches) per group, in order.

pub fn out_hidden_size(&self) -> usize

2026-09-25: Width of one merged embedding row.

pub fn ensure_scratch(&self, gpu: &dyn GpuBackend) -> Result<()>

2026-09-25: Allocate the encoder’s device scratch if it is not allocated yet.

Normally the first image does this. A tensor-parallel worker receives the merged rows by broadcast instead of encoding, so it calls this before out_row names a destination.

pub fn out_row(&self, row: usize) -> DevicePtr

2026-09-25: Device pointer to merged row row of the packed output. The Qwen arm panics unless forward_batched or ensure_scratch has allocated the scratch.