the largest operation is by far (q @ k) so just count that for simplicity |
||
|---|---|---|
| .. | ||
| ggml.go | ||
| ggml_test.go | ||
| gguf.go | ||
| type.go | ||
the largest operation is by far (q @ k) so just count that for simplicity |
||
|---|---|---|
| .. | ||
| ggml.go | ||
| ggml_test.go | ||
| gguf.go | ||
| type.go | ||