A small-model question for running Hermes Agent

is being considered for such as running . A test did not produce the expected output, even though the model has been getting strong interest.

The main question is whether something was missed in setup or whether a 3B model is simply too small for this kind of task. The practical limit being considered is another at 12B or smaller.

Key points

  • is being tested for -style tasks.
  • The result did not match the expected output.
  • The open question is whether the issue is setup or model capability.
  • The search is limited to models of 12B or smaller.
Read original