This sounds like it is fully reproducible: "it also included reproducible training recipes that allow the weights (neural network files) to be replicated, which

This sounds like it is fully reproducible: "it also included reproducible training recipes that allow the weights (neural network files) to be replicated, which is unusual for a major tech company so far" Digging into the code, is this the best definition of the datasets used? https://github.com/apple/corenet/blob/0333b1fbb29c31809663c4e6de2654b9ff2d27de/projects/openelm/README-pretraining.md?plain=1#L58 https://github.com/apple/corenet/blob/0333b1fbb29c31809663c4... And the paper says the instruction tuning is based on UltraFeedback, this config seems to say exactly in what form: https://github.com/apple/corenet/blob/0333b1fbb29c31809663c4e6de2654b9ff2d27de/projects/openelm/instruction_tuning/openelm-instruct.yaml#L11 https://github.com/apple/corenet/blob/0333b1fbb29c31809663c4...