sleepy/llama.cpp
1
0
Fork 0
Code Issues 8 Pull Requests 1 Actions 175 Packages Projects Releases Wiki Activity

Optimization: Qwen3 next autoregressive pass (#17996)

Browse Source 
* It's Qwen3 Next, the lean mean token generation machine!

* Apply patches from thread

* Remove recurrent version, only keep chunked and autoregressive

* Remove unnecessary conts and asserts

* Remove more extra conts and asserts

* Cleanup masking
This commit is contained in:
Piotr Wilkin (ilintar)
2025-12-16 11:59:53 +01:00
committed by GitHub
parent fb644247de
commit a5251ca11d
2 changed files with 85 additions and 270 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/models/models.h
+11 -11
View File
@@ -441,23 +441,13 @@ private:
ggml_tensor * cur,
ggml_tensor * causal_mask,
ggml_tensor * identity,
ggml_tensor * diag_mask,
int il);
ggml_tensor * build_layer_ffn(
ggml_tensor * cur,
int il);
ggml_tensor * build_delta_net_recurrent(
ggml_tensor * q,
ggml_tensor * k,
ggml_tensor * v,
ggml_tensor * g,
ggml_tensor * beta,
ggml_tensor * state,
ggml_tensor * causal_mask,
ggml_tensor * identity,
int il);
ggml_tensor * build_delta_net_chunking(
ggml_tensor * q,
ggml_tensor * k,
@@ -467,8 +457,18 @@ private:
ggml_tensor * state,
ggml_tensor * causal_mask,
ggml_tensor * identity,
ggml_tensor * diag_mask,
int il);
ggml_tensor * build_delta_net_autoregressive(
ggml_tensor * q,
ggml_tensor * k,
ggml_tensor * v,
ggml_tensor * g,
ggml_tensor * beta,
ggml_tensor * state,
int il);
ggml_tensor * build_norm_gated(
ggml_tensor * input,
ggml_tensor * weights,