Update README.md
Browse files
README.md
CHANGED
@@ -60,17 +60,19 @@ User: {utterance}
|
|
60 |
Character: {utterance}
|
61 |
```
|
62 |
|
63 |
-
This has an immediately noticeable effect on
|
64 |
`tiny`, `short`, `medium`, `long`, `huge`, `humongous`, `extreme`, `unlimited`. The
|
65 |
-
recommended starting length is `medium` or `long`.
|
66 |
-
can occur with
|
67 |
|
68 |
-
The effect is reproducible, but the messages will not necessarily follow
|
69 |
-
precisely
|
70 |
-
reply at the beginning of the conversation:
|
71 |
|
72 |

|
73 |
|
|
|
|
|
74 |
## Suggested settings
|
75 |
You can follow these instruction format settings in SillyTavern. Replace `tiny` with
|
76 |
your desired response length:
|
|
|
60 |
Character: {utterance}
|
61 |
```
|
62 |
|
63 |
+
This has an immediately noticeable effect on bot responses. The available lenghts are:
|
64 |
`tiny`, `short`, `medium`, `long`, `huge`, `humongous`, `extreme`, `unlimited`. The
|
65 |
+
recommended starting length is `medium` or `long`. Keep in mind that the AI may ramble
|
66 |
+
and impersonation can occur with very long messages.
|
67 |
|
68 |
+
The length control effect is reproducible, but the messages will not necessarily follow
|
69 |
+
lengths very precisely, rather follow certain ranges on average, as seen in this table
|
70 |
+
with data from tests made with one reply at the beginning of the conversation:
|
71 |
|
72 |

|
73 |
|
74 |
+
Response length control appears to work well also deep into the conversation.
|
75 |
+
|
76 |
## Suggested settings
|
77 |
You can follow these instruction format settings in SillyTavern. Replace `tiny` with
|
78 |
your desired response length:
|