"docs/en/user_guides/useful_tools.md" did not exist on "92d5a87e97b058621fa55a6f073a287e5ecd1a27"
  • Daniël de Kok's avatar
    Test Marlin MoE with `desc_act=true` (#2622) · 7f54b733
    Daniël de Kok authored
    Update the Mixtral GPTQ test to use a model with `desc_act=true` and
    `group_size!=-1` to ensure that we are checking activation
    sorting/non-full K (with tensor parallelism). The `desc_act=false` case
    is already checked by the Mixtral AWQ test.
    7f54b733
test_flash_mixtral_gptq.py 2.14 KB