3 ms·

Which models would you find useful?

• baobabKoodaa 3y ago
  Depends on the task, but generally, LLaMA 2 finetunes of at least 13B params, 4-bit quantized
  • celso 3y ago
    did you see we have lamma2-7b int8?
    • baobabKoodaa 3y ago
      Can you provide an example of a task where the 7b model is useful?