Did someone transpose the SOLVE_TRI result matrix? Perhaps...

Author: pwilkin

ggml/src/ggml-cpu/ops.cpp

@@ -9698,7 +9698,7 @@ static void ggml_compute_forward_solve_tri_f32(const struct ggml_compute_params
             for (int64_t t = 0; t < i00; ++t) {
                 sum += A_batch[i00 * n + t] * X_batch[t * k + i01];
             }
-
+            
             const float diag = A_batch[i00 * n + i00];
             GGML_ASSERT(diag != 0.0f && "Zero diagonal in triangular matrix");
             X_batch[i00 * k + i01] = (B_batch[i00 * k + i01] - sum) / diag;