Mastodon Feed
Boosted by jwz:
xgranade@wandering.shop ("Cassandra is only carbon now") wrote:
And yes, this is once again critiquing Dash's whole "I trained this model myself from open weights." That's fine-tuning. You didn't "train it yourself," you took a model someone gave you *in its compiled form* and tweaked it. Unless you have provenance for all the training data that went into it and can assert that the authors of that data agreed to have their labor used in that way, you cannot use that fine-tuning as an argument about labor ethics.