gahow/xserv
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Add Qwen3.6 MoE inference support

Browse Source
This commit is contained in:
Gahow Wang
2026-07-13 20:24:41 +08:00
parent 588bfd9df3
commit a2de146fb6
27 changed files with 3153 additions and 149 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
crates/xserv-model/src/decode_graph.rs
View File

@@ -98,9 +98,9 @@ impl DecodeGraphState {
let num_kv_heads = config.num_kv_heads();
let head_dim = config.head_dim();
let intermediate = config.ffn_hidden();
let vocab_size = config.vocab_size;
let vocab_size = config.vocab_size();
let num_layers = config.num_layers();
let eps = config.rms_norm_eps.unwrap_or(1e-6) as f32;
let eps = config.ln_eps();
let es = 2usize; // BF16 = 2 bytes
let stream = CudaStream::new().expect("create CUDA stream for graph");