lemonilia commited on
Commit
cdb3ec3
1 Parent(s): f170976

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +10 -0
README.md CHANGED
@@ -11,6 +11,16 @@ and a 2-pass training procedure. The first pass includes unsupervised finetuning
11
  For more details about LimaRP, see the model page for the [previously released version](https://huggingface.co/lemonilia/limarp-llama2-v2).
12
  Most details written there apply for this version as well.
13
 
 
 
 
 
 
 
 
 
 
 
14
  ## Prompt format
15
  Same as before. It uses the [extended Alpaca format](https://github.com/tatsu-lab/stanford_alpaca),
16
  with `### Input:` immediately preceding user inputs and `### Response:` immediately preceding
 
11
  For more details about LimaRP, see the model page for the [previously released version](https://huggingface.co/lemonilia/limarp-llama2-v2).
12
  Most details written there apply for this version as well.
13
 
14
+ ## Important notes on generation settings
15
+ It's recommended not to go overboard with low tail-free-sampling (TFS) values. From testing, decreasing it too much
16
+ appears to easily yield rather repetitive responses. Suggested starting generation settings are:
17
+
18
+ - TFS = 0.95
19
+ - Temperature = 0.70~0.85
20
+ - Repetition penalty = 1.05~1.10
21
+ - top-k = 0 (disabled)
22
+ - top-p = 1 (disabled)
23
+
24
  ## Prompt format
25
  Same as before. It uses the [extended Alpaca format](https://github.com/tatsu-lab/stanford_alpaca),
26
  with `### Input:` immediately preceding user inputs and `### Response:` immediately preceding