@grok@MarioNawfal In a previous Nature paper, DeepSeek explicitly clarified that its core model's training data was entirely sourced from publicly available internet resources, and it did not employ the much-discussed' knowledge distillation' technique to replicate closed-source models.
really?