61 transformer blocks › 58 MoE blocks › Sparse latent attention › Value heads
Value up-projection
×learnedself.wkv_b(kv)
Map the latent to 128 × 128 value features.
- in
- [B batch, T positions, 512 kv latent]
- out
- [B batch, T positions, 16384 heads × value]
- parameters
- 8,388,608 · × 58 MoE blocks = 486,539,264
When decoding, the reference dequantizes this matrix once and uses it unquantized (model.py 591–593).
Parameters
8,388,608 parameters per instance.
| Component | Calculation | Parameters |
|---|---|---|
| Weight matrix | 8,388,608 | |
| Per-instance total | 8,388,608 | |
This projection has no bias parameter.
The reference module holds the weight output by input, [out, in].
Across 58 independent MoE blocks: parameters.
Derived from the configured matrix dimensions; checkpoint tensors have not been independently counted.
Explanation
is output features 128–255 of each head’s 256 in wkv_b.
Codemodel.py 523–539, 574–589
self.qk_head_dim = args.qk_nope_head_dim + args.qk_rope_head_dim
self.v_head_dim = args.v_head_dim
self.wq_a = Linear(self.dim, self.q_lora_rank)
self.q_norm = RMSNorm(self.q_lora_rank)
self.wq_b = ColumnParallelLinear(self.q_lora_rank, self.n_heads * self.qk_head_dim)
self.wkv_a = Linear(self.dim, self.kv_lora_rank + self.qk_rope_head_dim)
self.kv_norm = RMSNorm(self.kv_lora_rank)
self.wkv_b = ColumnParallelLinear(self.kv_lora_rank, self.n_heads * (self.qk_nope_head_dim + self.v_head_dim))
self.wo = RowParallelLinear(self.n_heads * self.v_head_dim, self.dim)
self.softmax_scale = self.qk_head_dim ** -0.5
self.scale_fmt = args.scale_fmt
if args.max_seq_len > args.original_seq_len:
mscale = 0.1 * args.mscale * math.log(args.rope_factor) + 1.0
self.softmax_scale = self.softmax_scale * mscale * mscale
self.indexer = Indexer(args) if mask is not None: # MHA prefill
q = torch.cat([q_nope, q_pe], dim=-1)
kv = self.wkv_b(kv)
kv = kv.view(bsz, seqlen, self.n_local_heads, self.qk_nope_head_dim + self.v_head_dim)
k_nope, v = torch.split(kv, [self.qk_nope_head_dim, self.v_head_dim], dim=-1)
k = torch.cat([k_nope, k_pe.expand(-1, -1, self.n_local_heads, -1)], dim=-1)
scores = torch.einsum("bshd,bthd->bsht", q, k).mul_(self.softmax_scale)
# indexer
topk_indices = self.indexer(x, qr, start_pos, freqs_cis, mask)
index_mask = torch.full((bsz, seqlen, seqlen), float("-inf"), device=x.device).scatter_(-1, topk_indices, 0)
index_mask += mask
scores += index_mask.unsqueeze(2)
scores = scores.softmax(dim=-1)
x = torch.einsum("bsht,bthd->bshd", scores, v)Sources2
Direct evidence
MLA owns a query down-projection, query RMSNorm and up-projection, a joint key–value down-projection that also emits the shared rotary key, a latent RMSNorm, a key–value up-projection and an output projection.
mla-init · deepseek-ai/DeepSeek-V3.2-Exp / inference/model.py
Pinned revision 87e509a2e5a1 ↗ · lines 523–539
Read code excerpt
self.qk_head_dim = args.qk_nope_head_dim + args.qk_rope_head_dim
self.v_head_dim = args.v_head_dim
self.wq_a = Linear(self.dim, self.q_lora_rank)
self.q_norm = RMSNorm(self.q_lora_rank)
self.wq_b = ColumnParallelLinear(self.q_lora_rank, self.n_heads * self.qk_head_dim)
self.wkv_a = Linear(self.dim, self.kv_lora_rank + self.qk_rope_head_dim)
self.kv_norm = RMSNorm(self.kv_lora_rank)
self.wkv_b = ColumnParallelLinear(self.kv_lora_rank, self.n_heads * (self.qk_nope_head_dim + self.v_head_dim))
self.wo = RowParallelLinear(self.n_heads * self.v_head_dim, self.dim)
self.softmax_scale = self.qk_head_dim ** -0.5
self.scale_fmt = args.scale_fmt
if args.max_seq_len > args.original_seq_len:
mscale = 0.1 * args.mscale * math.log(args.rope_factor) + 1.0
self.softmax_scale = self.softmax_scale * mscale * mscale
self.indexer = Indexer(args)Direct evidence
In the full-sequence (prefill) path, the latent is up-projected into per-head keys and values; each head’s key joins its non-rotated part with the shared rotary key; scores are scaled, masked by the indexer’s selection and the causal mask, normalized and used to mix values.
mla-prefill · deepseek-ai/DeepSeek-V3.2-Exp / inference/model.py
Pinned revision 87e509a2e5a1 ↗ · lines 574–589
Read code excerpt
if mask is not None: # MHA prefill
q = torch.cat([q_nope, q_pe], dim=-1)
kv = self.wkv_b(kv)
kv = kv.view(bsz, seqlen, self.n_local_heads, self.qk_nope_head_dim + self.v_head_dim)
k_nope, v = torch.split(kv, [self.qk_nope_head_dim, self.v_head_dim], dim=-1)
k = torch.cat([k_nope, k_pe.expand(-1, -1, self.n_local_heads, -1)], dim=-1)
scores = torch.einsum("bshd,bthd->bsht", q, k).mul_(self.softmax_scale)
# indexer
topk_indices = self.indexer(x, qr, start_pos, freqs_cis, mask)
index_mask = torch.full((bsz, seqlen, seqlen), float("-inf"), device=x.device).scatter_(-1, topk_indices, 0)
index_mask += mask
scores += index_mask.unsqueeze(2)
scores = scores.softmax(dim=-1)
x = torch.einsum("bsht,bthd->bshd", scores, v)