Built on Qwen, these models incorporate our latest advances in post-training techniques. MindLink demonstrates strong performance across various common benchmarks and is widely applicable in diverse AI scenarios.

72B 32B

you are viewing a single comment's thread
view the rest of the comments
[–] 1 point 1 year ago (1 child)

reddit suggested "post training" means training on tests, rather than improving general programming outputs.

  • source
  • hideshow 1 child comment