Struct Qwen35WeightLoader 

pub struct Qwen35WeightLoader;

Trait Implementations

impl ModelWeightLoader for Qwen35WeightLoader

fn reads_expert_table_plan(&self) -> bool

2026-10-02: Its MoE layers build the transposed NVFP4 prefill tables only when the serve’s memory plan says they fit (load_layers.rs).

fn load_vision_encoder( &self, store: &WeightStore, config: &ModelConfig, gpu: &dyn GpuBackend, ) -> Result<Option<VisionTower>>

2026-09-25: Loads the ViT tower from model.visual or model.language_model.visual. Returns None when config.vision is None or neither prefix holds patch_embed.proj.weight. Reads depth blocks, one deepstack merger per deepstack_visual_indexes entry, and the final merger. Linear weights may be BF16, FP32 or FP8E4M3 (dequantized to BF16); biases, norms and embeddings BF16 or FP32.

fn supports_tp(&self) -> bool

2026-09-25: Whether this loader’s weight slicing honours config.tp_world_size / config.tp_rank. There is no default, so every loader declares it.

fn load_layers( &self, store: &WeightStore, config: &ModelConfig, gpu: &dyn GpuBackend, layer_kv_dtypes: &[KvCacheDtype], ) -> Result<Vec<Box<dyn TransformerLayer>>>

2026-09-25: Load all transformer layers from the weight store.

fn load_embedding( &self, store: &WeightStore, config: &ModelConfig, gpu: &dyn GpuBackend, ) -> Result<DenseWeight>

fn load_final_norm( &self, store: &WeightStore, config: &ModelConfig, gpu: &dyn GpuBackend, ) -> Result<DenseWeight>

2026-09-25: Load the final RMSNorm weight used before the LM head.

fn load_lm_head( &self, store: &WeightStore, config: &ModelConfig, gpu: &dyn GpuBackend, ) -> Result<DenseWeight>

fn load_mtp_weights( &self, store: &WeightStore, config: &ModelConfig, gpu: &dyn GpuBackend, ) -> Result<Option<MtpWeights>>

2026-09-25: Load single-module MTP head weights, or None.

fn prune_after_load( &self, _store: &mut WeightStore, _config: &ModelConfig, _gpu: &dyn GpuBackend, ) -> Result<()>

2026-09-25: Drop store tensors this loader has finished with. factory::build calls it after every load_* reader and before the KV budget is computed, so the freed memory counts.

fn precision_schedule(&self, _config: &ModelConfig) -> PrecisionSchedule

2026-09-25: Per-(layer, role) weight precision schedule. The default is PrecisionSchedule::default(), in which every role is Dtype::Inherit.

fn load_ngram_embedding( &self, _store: &WeightStore, _config: &ModelConfig, _gpu: &dyn GpuBackend, _max_tokens: usize, ) -> Result<Option<NgramEmbedding>>

2026-09-25: Build the n-gram embedding, for an architecture that fuses hashed n-gram lookups into the input embedding (the LongCat loader). The default None leaves the plain embed_tokens gather in place.

fn load_mtp_weights_multi( &self, store: &WeightStore, config: &ModelConfig, gpu: &dyn GpuBackend, ) -> Result<Vec<MtpWeights>>

2026-09-25: Load every MTP module; factory::build calls this one. The default wraps load_mtp_weights (zero or one module). The MiniMax and Step3p7 loaders override it.

fn kv_layer_dims(&self, _config: &ModelConfig) -> Vec<(usize, usize)>

2026-09-25: Per-attention-layer (num_kv_heads, head_dim), indexed like layer_kv_dtypes, for models whose attention geometry differs per layer (the Gemma-4 loader). The default is empty: every layer uses the global dims.

fn load_dflash_weights( &self, _drafter_store: &WeightStore, _config: &ModelConfig, _gpu: &dyn GpuBackend, _tp_size: usize, ) -> Result<Option<DflashWeights>>

2026-09-25: Load DFlash drafter weights from the drafter’s own WeightStore. The default returns None, and no loader overrides it: factory::build loads the drafter with the free function dflash_loader::load_dflash_weights.

fn load_lora_adapters( &self, adapters: &[LoraAdapterInput<'_>], config: &ModelConfig, gpu: &dyn GpuBackend, max_loras: usize, max_lora_rank: usize, ) -> Result<Option<LoraWeights>>

2026-09-25: Load startup-static PEFT LoRA adapters into the LoRA pool, through metrale_model_layers::lora::load_lora_adapters_multi. factory::build calls it before the buffer arena and the KV budget, so the pool counts as used memory.

fn binds_vision_encoder(&self) -> bool

2026-09-25: Will this loader bind a vision encoder for a multimodal checkpoint?

fn defer_predicate(&self, _config: &ModelConfig) -> Option<DeferHook>

2026-09-25: Which checkpoint tensors will this loader read from the host at bind time, so the checkpoint loader records their location instead of uploading them?