llama : advanced batch splits

This includes equal-sequence-length batch splits which are useful to simplify recurrent model operators. * llama : always make recurrent state slots contiguous * ggml : simplify mamba operators
ggerganov · Jul 17, 2024 · c51daef · c51daef
1 parent a38b884
commit c51daef
Show file tree

Hide file tree

Showing 3 changed files with 1,060 additions and 647 deletions.
diff --git a/ggml/include/ggml.h b/ggml/include/ggml.h
@@ -1760,10 +1760,8 @@ extern "C" {
 
     GGML_API struct ggml_tensor * ggml_ssm_conv(
             struct ggml_context * ctx,
-            struct ggml_tensor  * s,
-            struct ggml_tensor  * x,
-            struct ggml_tensor  * c,
-            struct ggml_tensor  * sq);
+            struct ggml_tensor  * sx,
+            struct ggml_tensor  * c);
 
     GGML_API struct ggml_tensor * ggml_ssm_scan(
             struct ggml_context * ctx,
@@ -1772,8 +1770,7 @@ extern "C" {
             struct ggml_tensor  * dt,
             struct ggml_tensor  * A,
             struct ggml_tensor  * B,
-            struct ggml_tensor  * C,
-            struct ggml_tensor  * sq);
+            struct ggml_tensor  * C);
 
     // partition into non-overlapping windows with padding if needed
     // example: