Directories
¶
| Path | Synopsis |
|---|---|
|
Package argmax generates the WGSL of a greedy sampler in two stages, so that a large vocabulary is not walked by one workgroup: Groups workgroups each reduce a slice of the logits to its (largest, lowest index among equals), then one workgroup reduces the partials and writes the token.
|
Package argmax generates the WGSL of a greedy sampler in two stages, so that a large vocabulary is not walked by one workgroup: Groups workgroups each reduce a slice of the logits to its (largest, lowest index among equals), then one workgroup reduces the partials and writes the token. |
|
Package q4 generates the WGSL of a matrix-vector product for ggml Q4_0 and Q4_1 weights, f32 activations, with no subgroups and no f16.
|
Package q4 generates the WGSL of a matrix-vector product for ggml Q4_0 and Q4_1 weights, f32 activations, with no subgroups and no f16. |
Click to show internal directories.
Click to hide internal directories.