Edit model card

image/png

NOTE: See creative-writing-control-vectors-v3.0 for the current main control-vector repo.


UPDATE: (06/07/24) Added a "show_dont_tell" control vector for each of the models.

Use the '--control-vector-scaled' option like this:

llama-cli --model <model name>.gguf [the rest of your CLI arguments...] \
    --control-vector-scaled <model name>-dark.gguf 0.5 \
    --control-vector-scaled <model name>-chaos.gguf 0.5 \
    --control-vector-scaled <model name>-show_dont_tell.gguf 0.5

or:

llama-server --model <model name>.gguf [the rest of your CLI arguments...] \
    --control-vector-scaled <model name>-dark.gguf 0.5 \
    --control-vector-scaled <model name>-chaos.gguf 0.5 \
    --control-vector-scaled <model name>-show_dont_tell.gguf 0.5

NOTE:

  • Use positive scale factors to make the model "more dark" or "more chaotic" or "more show don't tell".
  • I forgot to change the default scale factor of these to 1.0 (to match the old creative-writing-control-vectors-BETA-v0.1 files... oops) - these have a default of 0.5!
  • I suggest you use --control-vector-scaled <model name>-XXX.gguf 0.5 to start off with and test the effect it has.
  • You can use one or all of the files together. For traits you don't want: either set the scale factor to 0.0 or just don't include it on the command line.
  • Make sure you have a version of llama.cpp from after 27/06/24 - the PR to use multiple control vectors together was only added on that date. Older versions of llama.cpp will just silently load the first control vector (and none of the others) if the layer index of the final direction in each file does not match!!!
Downloads last month
20
GGUF
Model size
338k params
Architecture
controlvector
Inference API
Unable to determine this model's library. Check the docs .

Collection including jukofyork/creative-writing-control-vectors-BETA-v0.2