fixed no max output tokens

[Puchaczov]

Hey, thanks a lot for creating this project! I think I found a bug related to missing MaxOutputTokens field mapping, I fixed it both in the code and in the unit test.

I think i found another one but it's within Semantic Kernel connector library. The OllamaPromptExecutionSettings doesn't contain and thus you cannot send max output tokens, I will try to report it them

    [JsonPropertyName("num_predict")]
    [JsonIgnore(Condition = JsonIgnoreCondition.WhenWritingNull)]
    public double? NumPredict
    {
        get => this._numPredict;

        set
        {
            this.ThrowIfFrozen();
            this._numPredict = value;
        }
    }

[awaescher]

Nice spot. Would you add a line to TryAddOllamaOption(...) for it as well? Just in case these settings are delivered as untyped argument.

[Puchaczov]

Not sure if that's what you meant but I've extended TryAddOllamaOption to support max_output_tokens and added new test to cover all the options included

[awaescher]

I understand you prefer jumping out here. I am okay with that but the combination of not and the coalesce operator is not that easy to recognize so far. Also I personally don't like having the return in the same line.

I would prefer something like this, what do you think?

var hasAdditionalProperties = options?.AdditionalProperties?.Any() ?? false;
if (!hasAdditionalProperties)
    return request;

[Puchaczov]

I totally agree with you! fixed!

[awaescher]

Thanks Jakub