Error processing text: Error: Embed model not set when using ollama.

[remisharrock]

• I'm submitting a ... [ ] bug report [ ] feature request [ ] question about the decisions made in the repository [X ] question about how to use this project

• Summary I modified the example vectordb.ts to use ollama:

  const ai = axAI('ollama', {
  model: 'llama3:latest',
  url: 'http://localhost:11434'
  } as AxOllamaArgs);

  but got this error:

  PS C:\Users\remis\ax\src\examples> bun run .\vectordb.ts
  98 | 
  99 |         // Batch upsert embeddings
  100 |         await this.db.batchUpsert(embeddings);
  101 |       }
  102 |     } catch (error) {
  103 |       throw new Error(`Error processing text: ${error}`);
                ^
  error: Error processing text: Error: Embed model not set
    at C:\Users\remis\ax\src\docs\manager.ts:103:13

• Other information (e.g. detailed explanation, stack traces, related issues, suggestions how to fix, links for us to have context, eg. StackOverflow, personal fork, etc.)

Indeed in ollama i have only one model: 'llama3:latest' installed. Should I install an embedding model like explained here https://ollama.com/blog/embedding-models

But then how do I configure this model in here:

export type AxOllamaArgs = {
  model: string;
  url?: string;
  apiKey?: string;
  config?: Readonly<Omit<AxOllamaAIConfig, 'model'>>;
  options?: Readonly<AxAIServiceOptions>;
};

thanks for the help!

[dosco]

I will set two default models on ollama in the default config but you are free to change them as below

  model: 'nous-hermes2',
  embedModel: 'all-minilm'

const ai = axAI('ollama', { model: 'nous-hermes2' , embedModel: 'mxbai-embed-large' });