Ollama Provider

Prerequisites

• You must have ollama installed and running eg. (ollama serve).
• You must have a model installed and run it eg. (ollama run qwen2.5vl:7b).

Gowall assumes that ollama is running in the default host/port http://127.0.0.1:11434 if you are running it on anything different you can specify it with this env, see here on how to setup env's.

OLLAMA_HOST=your_host

tip

If the model isn't offloading to the GPU, you see any 100% of CPU usage meanwhile it hasn't even taken all of the VRAM you can tweak the following env's before you do ollama run ...

export OLLAMA_GPU_LAYERS=999
export OLLAMA_NUM_PARALLEL=1

You must absolutely make sure ollama does not really offload to the CPU, because its insanely slow. If you see the following message when you do ollama run ...

msg="offloaded 36/37 layers to GPU"
msg="offloading output layer to CPU" # <-- use the env's above, do not let it offload to both the CPU and GPU.  

Most Basic Schema

~/.config/gowall/schema.yml

schemas:
  - name: "oll"
    config:
      ocr:
        provider: "ollama"
        model: "qwen2.5vl:7b"

gowall ocr img.png -s oll

Please refer to the schemas section for more information on how to use schemas with rate limiting,pre/post processing options,text correction etc...