Skip to content

can't load any model on nvidia dgx spark #163

Description

@Slach

🐛 Bug Description

i successfully install shimmy via cargo install shimmy --features gpu

~/.cargo/bin/shimmy list show me some already downloaded models

but i can't serve or probe any
it get always probe failed: null result from llama cpp

🔄 Steps to Reproduce

~/.cargo/bin/shimmy probe phi3-lora

ggml_cuda_init: GGML_CUDA_FORCE_MMQ: no
ggml_cuda_init: GGML_CUDA_FORCE_CUBLAS: no
ggml_cuda_init: found 1 CUDA devices:
Device 0: NVIDIA GB10, compute capability 12.1, VMM: yes
llama_model_load_from_file_impl: using device CUDA0 (NVIDIA GB10) (000f:01:00.0) - 73847 MiB free
gguf_init_from_file: failed to open GGUF file './models/phi3-mini.gguf'
llama_model_load: error loading model: llama_model_loader: failed to load model from ./models/phi3-mini.gguf
llama_model_load_from_file_impl: failed to load model
probe failed: null result from llama cpp

✅ Expected Behavior

loading successfully models

❌ Actual Behavior

probe failed: null result from llama cpp

📦 Shimmy Version

~/.cargo/bin/shimmy --version
shimmy 1.8.2

💻 Operating System

Linux (Ubuntu)

📥 Installation Method

cargo install shimmy

🌍 Environment Details

rustc --version
rustc 1.91.1 (ed61e7d7e 2025-11-07)

hardware NVIDIA GB10

Metadata

Metadata

Assignees

No one assigned

    Labels

    bugSomething isn't workingroadmapPlanned for a future release

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions