I have searched through the source code with no luck on figuring out how to pass keep_alive. Am I missing something here?

Hello, how could I consumen the chat or any stream? since this request responds with an array and not a resource (having "stream" true by default)

$response = Ollama::agent("You are an awesome assistant")
  ->model('llama2')
  ->stream(true)
  ->chat($this->conversation);
                
  dd($response)
               
  (returns array)

Highly likely issue on my end, however:
[ec2-user@app www]$ curl http://192.168.1.149:11434/api/generate -d '{
"model": "llama3",
"prompt": "Hey",
"stream":false
}'
{"model":"llama3","created_at":"2024-06-06T01:27:27.256766425Z","response":"Hey! How's it going?","done":true,"done_reason":"stop","context":[128006,882,128007,271,19182,128009,128006,78191,128007,271,19182,0,2650,596,433,2133,30,128009],"total_duration":2320969362,"load_duration":1230493,"prompt_eval_count":6,"prompt_eval_duration":798696000,"eval_count":8,"eval_duration":1479250000}[ec2-user@app www]$

<?php

namespace App\Console\Commands;

use Illuminate\Console\Command;
use Illuminate\Support\Facades\Log;
use Cloudstudio\Ollama\Facades\Ollama;

class TestCommand extends Command
{
    protected $signature = 'test:command';
    protected $description = 'Test command';

    public function handle()
    {

        $response = Ollama::model('llama3')->show();
        print_r($response);
        }
}

shows:

[ec2-user@app www]$ php artisan test:command

Array
(
    [error] => unauthorized
)

Any thoughts?
Thanks!!

P.S. Config:

<?php

// Config for Cloudstudio/Ollama

return [
    'model' => env('OLLAMA_MODEL', 'llama3'),
    'url' => env('OLLAMA_URL', '192.168.1.149:11434'),
    'default_prompt' => env('OLLAMA_DEFAULT_PROMPT', 'Hello, how can I assist you today?'),
    'connection' => [
        'timeout' => env('OLLAMA_CONNECTION_TIMEOUT', 300),
    ],
];

The underlying llama.cpp has an --api-key option that I would like to utilize, but Ollama does not support. Their current recommendation appears to be setup your own proxy - which is fine - but I need to be able to pass the api key in the Guzzle request regardless if they support llama.cpp's option or my own proxy.

Is this something that makes sense to do a pull request for? I'm thinking a new .env variable and config key allowing the authorization header via Guzzle: Authorization: Bearer <apikey> added to https://github.com/cloudstudio/ollama-laravel/blob/main/src/Traits/MakesHttpRequests.php?

• ollama/ollama#849
• https://github.com/ggerganov/llama.cpp/blob/master/examples/server/README.md

Is this package support function call?

First of all, ollama is working fine when i use the terminal.

When i use this package, i mostly get connection time out in de browser.

Would it be possible to add support for the messages parameter in order to maintain history when creating a chatbot integration?

https://github.com/jmorganca/ollama/blob/main/docs/api.md#generate-a-chat-completion

thanks a lot for the package help me done so many things in my project, can you help to add more example to use Embeddings?

When I try to run just the default it returns null

        dd(\Cloudstudio\Ollama\Facades\Ollama::agent('You are a weather expert...')
            ->prompt('Why is the sky blue?')
            ->model('llama3')
            ->keepAlive('10m')
            ->options(['temperature' => 0.8])
            ->stream(false)
            ->ask());

It appears that OLLAMA_URL=http://ollama.test/ inside my .env gives enough information for the requests to proceed but the end result is not working. To fix it just remove the ending "/" so it will become - OLLAMA_URL=http://ollama.test

It could be fixed as a small amend on the code but there is still an easy fix with amending the url.

Hope that helps someone in the future.

Can you add L11 please