Problem with JSONL when running the 3 lines of code

Short Course Q&A Finetuning Large Language Models
1

I am running the training lab on colab, the step where the model is finetuned using 3 lines of code fails with the below:

ValueError: Each object must have input_key={input_key} (and optionally output_key={output_key}). You can pass in different input_key and output_keys

I do have the correct jsonl file. Wondering if the modules I have are of a different version and maybe expects the input file in a different format.

Any thoughts?

Thanks.

2

Hi @sanimesa

There is already a topic regards this issue open

3

Did you check this?

1 Like
4

Thanks, will check it out. So it confirms that the JSONL format does not work with the code except in the course environment for some reason.

Thanks.

5

Looks like this is due to a version mismatch - the current lamini version is 0.0.21 while the training lab version is 0.0.20.