Crash: Assertion '!this->empty()' failed

[moritztim]

System Info

aur/gpt4all-chat 2.5.4-1 (+0 0.00) (Installed)
Linux pc 6.6.3-arch1-1 #1 SMP PREEMPT_DYNAMIC Wed, 29 Nov 2023 00:37:40 +0000 x86_64 GNU/Linux

Information

• The official example notebooks/scripts
• My own modified scripts
• GUI

Reproduction

1. open gpt4all
2. start download of mistral-7b-openorca.Q4_0.gguf
3. download finishes, app crashes

[Debug] (Thu Nov 30 10:55:58 2023): deserializing chats took: 0 ms
[Warning] (Thu Nov 30 10:56:52 2023): Opening temp file for writing: "/home/moti/AI/Text Generation/gpt4all/incomplete-mistral-7b-openorca.Q4_0.gguf"
[Warning] (Thu Nov 30 10:57:18 2023): Opening temp file for writing: "/home/moti/AI/Text Generation/gpt4all/incomplete-mistral-7b-openorca.Q4_0.gguf"
[Warning] (Thu Nov 30 10:59:05 2023): stream 3 finished with error: "Internal server error"
[Warning] (Thu Nov 30 10:59:05 2023): Opening temp file for writing: "/home/moti/AI/Text Generation/gpt4all/incomplete-mistral-7b-openorca.Q4_0.gguf"
[Warning] (Thu Nov 30 10:59:05 2023): "ERROR: Downloading failed with code 401 \"Internal server error\""
llama_new_context_with_model: max tensor size =   102.55 MB
llama.cpp: using Vulkan on /usr/include/c++/13.2.1/bits/stl_vector.h:1208: constexpr std::vector<_Tp, _Alloc>::reference std::vector<_Tp, _Alloc>::front() [with _Tp = ggml_vk_device; _Alloc = std::allocator<ggml_vk_device>; reference = ggml_vk_device&]: Assertion '!this->empty()' failed.

[Debug] (Thu Nov 30 11:01:34 2023): deserializing chats took: 6 ms
[Warning] (Thu Nov 30 11:01:35 2023): ERROR: Previous attempt to load model resulted in crash for `mistral-7b-openorca.Q4_0.gguf` most likely due to insufficient memory. You should either remove this model or decrease your system RAM by closing other applications. id "0eabbb6e-765e-4319-af82-c03c81e9e303"
llama_new_context_with_model: max tensor size =   102.55 MB
llama.cpp: using Vulkan on /usr/include/c++/13.2.1/bits/stl_vector.h:1208: constexpr std::vector<_Tp, _Alloc>::reference std::vector<_Tp, _Alloc>::front() [with _Tp = ggml_vk_device; _Alloc = std::allocator<ggml_vk_device>; reference = ggml_vk_device&]: Assertion '!this->empty()' failed.
[2]    36718 IOT instruction (core dumped)  gpt4all-chat

Expected behavior

...

[moritztim]

This issue template is incomplete. Usually there's steps, expected behavior and actual behavior. Also there section about scripts is required but not necessarily applicable. There's also no * to indicate it is required but I can't submit without it.

[moritztim]

most likely due to insufficient memory

16% usage.

[cebtenzzre]

This issue will be fixed in the next release (I wrote a fix before I read this issue). You're the first user to notice it because you're building with -D_GLIBCXX_ASSERTIONS, which Arch Linux enables by default. We should probably enable that in our build script for at least debug builds... I only caught it because of AddressSanitizer.