Update README.md
Browse files
README.md
CHANGED
@@ -32,8 +32,8 @@ llama-cli --model <model name>.gguf --control-vector-scaled <model name>-dark.gg
|
|
32 |
**NOTE:**
|
33 |
|
34 |
- Use ***positive scale factors*** to make the model "***more dark***" or "***more chaotic***".
|
35 |
-
- I suggest you use `--control-vector-scaled
|
36 |
-
- The
|
37 |
-
|
38 |
-
- You can use one control vector file alone if you want, or alternatively set the scale factor to `0.0` for
|
39 |
- You can use the same "`--control-vector-scaled`" command line arguments for "`llama-server`" as in the above "`llama-cli`" examples.
|
|
|
32 |
**NOTE:**
|
33 |
|
34 |
- Use ***positive scale factors*** to make the model "***more dark***" or "***more chaotic***".
|
35 |
+
- I suggest you use `--control-vector-scaled 0.5` and `--control-vector-scaled 0.5` to start and then test the effect.
|
36 |
+
- The "chaos" control vectors generally seems less effective than the "dark" control vectors.
|
37 |
+
- Some models like `command-r:35b` and `command-r-plus:104b` need lower scale factors, whereas `miqu-1:70b` seems to need (much) higher scale factors to stamp out pesky redemption arcs.
|
38 |
+
- You can use one control vector file alone if you want, or alternatively set the scale factor to `0.0` for traits you don't want to use.
|
39 |
- You can use the same "`--control-vector-scaled`" command line arguments for "`llama-server`" as in the above "`llama-cli`" examples.
|