Ollama Support

[evrenyal]

Hi,

Are you going to add Ollama support or maybe I missed it?

Thanks.

[laurentboutet]

How can we configure the URL and port of the Ollama?
By default, it tests localhost and default port 11434. I need to send to my ollama server.
No way to found how to manage configuration
Thanks

[laurentboutet]

Hello, I found. It is outside the solution in another package. I resolve by this way langchain-ai/langchain#15365

But it would be confortable to be able to manage it directly
thanks

[guy-ps]

Hi @laurentboutet, what do you mean by manage it directly?

[laurentboutet]

I mean have the possibility to configure the URL and port in the configuration or with environnement in ps-fuzz.

[toolmanxyz]

I found a way to manually specify the URL and port in the code.

In the file ./ps_fuzz/prompt_injection_fuzzer.py, append base_url="http://IP:11434" at the end of line 170:

target_client = ClientLangChain(app_config.target_provider, model=app_config.target_model, temperature=0, base_url="http://IP:11434")

And append base_url="http://IP:11434" at the end of line 178:

attack_client = ClientLangChain(app_config.attack_provider, model=app_config.attack_model, temperature=app_config.attack_temperature, base_url="http://IP:11434"),