Struct ModelLevers 

pub struct ModelLevers {

    pub gdn_regresident: bool,
    pub gdn_batched_fla: bool,
    pub gdn_wy17: bool,
    pub gdn_wyn: bool,
    pub gemv_sw: bool,
    pub decode_ffn_via_gemm: bool,
    pub ffn_small_m: bool,
    pub holo_moe_down_fp4: bool,
    pub holo_moe_gateup_fp4: bool,
    pub moe_union_stats: bool,
    pub dflash_contig_attn: bool,
    pub lora_eager: bool,
    pub lora_rotate: bool,
    pub k4_diag: bool,
    pub gemma4_diag: bool,
    pub bf16_tc_proj: bool,
    pub max_decode_seqs: u32,
    pub shadow_topk: usize,
    pub kv_poison: bool,
    pub drafter: DrafterContext,
}

Kernel-path levers for one loaded model.

Plain Copy data resolved from the environment at model construction. Group membership follows the subsystem the lever steers, so a reader can see at a glance which part of the forward pass a flag reaches.

Fields

gdn_regresident: bool

Keep GDN recurrent state in registers across the prefill chunk loop. Default ON (the fold that shipped in PR #369, −7.25 % wall); the env var is an opt-OUT, which is why the field is stored positively and the resolution inverts it.

gdn_batched_fla: bool

Batched FLA path for multi-sequence GDN decode.

gdn_wy17: bool

WY17 GDN recurrence variant. Ships ON; ATLAS_GDN_WY17=0 opts out.

gdn_wyn: bool

WY-N GDN recurrence variant. Ships ON; ATLAS_GDN_WYN=0 opts out.

gemv_sw: bool

Lossless single-warp decode GEMV (w4a16_gemv_sw, w4a16_gemv_dual_sw). Ships ON; ATLAS_NO_GEMV_SW=1 restores the 64-thread kernels.

decode_ffn_via_gemm: bool

Route decode FFN through the tile GEMM rather than the scalar GEMV.

ffn_small_m: bool

Small-M FFN GEMM tile shape. Ships ON; ATLAS_FFN_SMALLM=0 opts out.

holo_moe_down_fp4: bool

FP4 holo layout for the MoE down projection.

holo_moe_gateup_fp4: bool

FP4 holo layout for the MoE gate/up projections.

moe_union_stats: bool

Collect per-layer MoE expert-union statistics. Diagnostic.

dflash_contig_attn: bool

Contiguous-attention path for the DFlash head.

lora_eager: bool

Apply LoRA eagerly at load instead of at each forward.

lora_rotate: bool

Allow hot rotation of LoRA adapters.

k4_diag: bool

K=4 chain-widening diagnostics.

gemma4_diag: bool

Per-layer hidden-state norm dumps on the Gemma-4 decode path. Heavy — one device-to-host copy per layer.

bf16_tc_proj: bool

BF16 tensor-core attention projections: dequant FP4 to BF16 and use a BF16 MMA instead of the default path, which crushes activations to FP8 E4M3. Removes the FP8 prefill perturbation on those projections.

max_decode_seqs: u32

Configured max decode batch (--max-batch-size), the reference count the split-K attention split count is pinned to. Not from the environment: TransformerModel::new writes it from the serve arg.

It pins DETERMINISM — the online-softmax split-merge is non-associative, so a sequence decoded alone must see the same reduction tree as one co-batched with fifteen others. Held in a OnceLock it was also idempotent, so a second model with a different max batch would silently keep the first model’s split count.

shadow_topk: usize

ATLAS_MTP_SHADOW_TOPK=k (0 = off, clamped to 8): the drafter D2Hs its logits and logs the top-k candidates. Observational only.

kv_poison: bool

ATLAS_KV_POISON=1 — fill a fresh KV block with NaN instead of zero, the discriminator for the “unwritten fresh tail block read” hypothesis. A diagnostic that changes what the kernels READ, so it must not leak across a swap.

drafter: DrafterContext

MTP drafter context policy (ATLAS_NO_DRAFTER_CONTEXT / ATLAS_DRAFTER_PREFILL_ONLY), resolved and logged once per model. The two halves are coupled — prefill without carry is a measured −927 ms/turn loss — so they travel as one value.

Implementations

impl ModelLevers

pub fn from_env() -> Self

Resolve from the environment. Called once, when the model is built.

pub fn defaults() -> Self

What a build resolves to with no ATLAS_* set — every opt-in off, the one opt-out lever on. Tests construct a context with this instead of mutating the process environment.

Trait Implementations

impl Clone for ModelLevers

fn clone(&self) -> ModelLevers

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for ModelLevers

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for ModelLevers

fn default() -> ModelLevers

Returns the “default value” for a type.

impl PartialEq for ModelLevers

fn eq(&self, other: &ModelLevers) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl Copy for ModelLevers

impl Eq for ModelLevers

impl StructuralPartialEq for ModelLevers