valentimarco

Update: on my pc (that one with the 3070ti) i get this error when try to inference mistral:latest ```bash 2023/12/20 12:44:41 images.go:737: total blobs: 10 2023-12-20T12:44:41.088356611Z 2023/12/20 12:44:41 images.go:744: total...

Update 2: I tested "zephyr:7b-beta-q4_0" in my server with ollama 0.1.9 and no problems of allocations. I saw that the new version (0.1.17) use the latest version of llama.cpp, could...

So with the last version of ollama: - Same problems in the server with the same error, cannot run zephyr:7b-beta-q4_0 and phi. Mistral worked, but output no sense tokens or...

how we can help? i have a gtx 960 (4g version) with a compute capability of 5.2 in my little server

i tested the main repo of ollama with only 2 modificationn: change the CudaComputevalue to 5 and apply the change from #1880. i tested phi and mistral and works just...

If you need some dumps of connection, feel free to ask me!

Maybe the error below occurs for a memory leak issue? ``` cuBLAS error 15 at /go/src/github.com/jmorganca/ollama/llm/llama.cpp/ggml-cuda.cu:8458 current device: 0 GGML_ASSERT: /go/src/github.com/jmorganca/ollama/llm/llama.cpp/ggml-cuda.cu:8458: !"cuBLAS error" ``` I tried to understand this asserts,...

@pieroit With the new agent structure, should we discuss a solution?

I have the same issue in my country(italy): ```js const inputFromBackend = { start: '14:30:00', end: '18:45:00', } // 2. let startHours = inputFromBackend.start.split(':').map(Number) let endHours = inputFromBackend.end.split(':').map(Number) let time...

There was a request similar: #694 This way the cat is more like a Product/Runner (Ollama for example is one of those, bash script and docker image) rather than a...