pub struct MimoAudioGeom {Show 16 fields
pub channels: usize,
pub group_size: usize,
pub local_dim: usize,
pub local_layers: usize,
pub local_heads: usize,
pub local_head_dim: usize,
pub local_intermediate: usize,
pub rope_theta: f64,
pub partial_rotary: f64,
pub full_attention: bool,
pub post_norm: bool,
pub projection_layers: usize,
pub out_hidden: usize,
pub segment_size: usize,
pub speech_vocab: Vec<usize>,
pub speech_zero: Vec<usize>,
}Expand description
Audio patch encoder geometry (config.audio_config, HF MiMoAudioEncoder).
Fields§
§channels: usizeRVQ channels summed per frame (20).
group_size: usizeFrames per LLM token (4).
local_dim: usizeLocal transformer width (1024).
local_layers: usize§local_heads: usize§local_head_dim: usize§local_intermediate: usize§rope_theta: f64§partial_rotary: f64§full_attention: boolBidirectional attention inside a group.
post_norm: bool§projection_layers: usizeLLM hidden (4096).
segment_size: usize§speech_vocab: Vec<usize>Rows of each speech_embeddings.{c} table.
speech_zero: Vec<usize>padding_idx of each table (the “empty” code).
Implementations§
Trait Implementations§
Source§impl Clone for MimoAudioGeom
impl Clone for MimoAudioGeom
Source§impl Debug for MimoAudioGeom
impl Debug for MimoAudioGeom
Source§impl PartialEq for MimoAudioGeom
impl PartialEq for MimoAudioGeom
impl StructuralPartialEq for MimoAudioGeom
Auto Trait Implementations§
impl Freeze for MimoAudioGeom
impl RefUnwindSafe for MimoAudioGeom
impl Send for MimoAudioGeom
impl Sync for MimoAudioGeom
impl Unpin for MimoAudioGeom
impl UnsafeUnpin for MimoAudioGeom
impl UnwindSafe for MimoAudioGeom
Blanket Implementations§
Source§impl<T> BorrowMut<T> for Twhere
T: ?Sized,
impl<T> BorrowMut<T> for Twhere
T: ?Sized,
Source§fn borrow_mut(&mut self) -> &mut T
fn borrow_mut(&mut self) -> &mut T
Mutably borrows from an owned value. Read more