tanaymehta
commited on
Commit
•
26de81c
1
Parent(s):
7f22f0a
Add Fill-in-Middle specific tokens
Browse filesRequest for adding FIM-specific tokens (prefix, middle and suffix) as discussed in the PR for adding FIM pipeline to transformers (https://github.com/huggingface/transformers/pull/27828#issuecomment-1970847921)
- tokenizer_config.json +4 -1
tokenizer_config.json
CHANGED
@@ -30,5 +30,8 @@
|
|
30 |
"normalized": true,
|
31 |
"rstrip": false,
|
32 |
"single_word": false
|
33 |
-
}
|
|
|
|
|
|
|
34 |
}
|
|
|
30 |
"normalized": true,
|
31 |
"rstrip": false,
|
32 |
"single_word": false
|
33 |
+
},
|
34 |
+
"fim_prefix": "▁<PRE>",
|
35 |
+
"fim_middle": "▁<MID>",
|
36 |
+
"fim_suffix": "▁<SUF>"
|
37 |
}
|