Struct BertEncoder 

pub struct BertEncoder {
    pub hp: BertHparams,
    pub tok_embd: WeightMatrix,
    pub type_embd_row0: Option<Vec<f32>>,
    pub pos_embd: WeightMatrix,
    pub tok_norm_w: Vec<f32>,
    pub tok_norm_b: Vec<f32>,
    pub layers: Vec<BertLayer>,
}

Fields

hp: BertHparams

tok_embd: WeightMatrix

type_embd_row0: Option<Vec<f32>>

Row 0 of token_types.weight — “Sentence A”. See the module docs.

pos_embd: WeightMatrix

tok_norm_w: Vec<f32>

tok_norm_b: Vec<f32>

layers: Vec<BertLayer>

Implementations

impl BertEncoder

pub fn vocab_size(&self) -> usize

Trait Implementations

impl TextEncoder for BertEncoder

fn wrap_special(&self, pieces: &[u32]) -> Vec<u32>

[CLS] … [SEP], which is what llama.cpp’s WPM branch adds when add_special is set: it pushes special_bos_id before the pieces and special_sep_id after, unconditionally — the add_bos/add_eos flags are not consulted on that path (llama-vocab.cpp, case LLAMA_VOCAB_TYPE_WPM).

fn wrap_special_pair(&self, a: &[u32], b: &[u32]) -> Option<Vec<u32>>

[CLS] a [SEP] b [SEP] — what HuggingFace’s tokenizer(query, document) builds for a BERT cross-encoder, which is the input these checkpoints were trained on.

The segment ids that pairing normally carries are not here, and that is deliberate rather than forgotten: this graph adds row 0 of token_types.weight to every position, because llama.cpp does the same (token types are hardcoded to zero, its bert.cpp; see BertEncoder::type_embd_row0). A reranker scored here therefore differs from the HF reference by the second segment’s type embedding, on both engines equally. It is recorded in the ferrox rerank docs and is the first thing to check if an ordering disagrees with transformers.

fn n_embd(&self) -> usize

Width of one hidden-state row, and of the pooled embedding.

fn n_ctx_train(&self) -> usize

Longest sequence this checkpoint can represent. For a learned position table this is the table’s height, and exceeding it is an error rather than a degradation.

fn pooling_type(&self) -> PoolingType

What the checkpoint’s own {arch}.pooling_type said.

fn encode_tokens(&self, tokens: &[u32]) -> Result<Vec<f32>, EncodeError>

n_tokens × n_embd hidden states, in row order.

fn embed_tokens(&self, tokens: &[u32]) -> Result<Vec<f32>, EncodeError>

Self::encode_tokens followed by the checkpoint’s own pooling. Not L2-normalized — see crate::pooling::pool.