diff --git a/src/llama-model.cpp b/src/llama-model.cpp index ef311a96f..821c4ff1f 100644 --- a/src/llama-model.cpp +++ b/src/llama-model.cpp @@ -2026,7 +2026,9 @@ bool llama_model_base::load_tensors(llama_model_loader & ml) { auto & last_bufs = pimpl->ctxs_bufs.back().second; if (!host_ptr_path && !ml.no_alloc && last_bufs.size() == 1) { ggml_backend_buffer_t b = last_bufs[0].get(); - if (b != nullptr && !ggml_backend_buffer_is_host(b)) { + // CPU extra buffers (CPU_REPACK) are not host buffers, but they are in RAM and have no get_tensor + ggml_backend_dev_t buft_dev = ggml_backend_buft_get_device(buft); + if (b != nullptr && !ggml_backend_buffer_is_host(b) && (buft_dev == nullptr || ggml_backend_dev_type(buft_dev) != GGML_BACKEND_DEVICE_TYPE_CPU)) { slot.releasable = true; slot.buft = buft; }