vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Update special token handling in conversion scripts for gpt2 derived tokenizers (#3746)

Browse source 
We still have the heads up in `README.md` regarding `bpe` tokenizers and this patch is needed for 

- a couple of tokenizer tests
- some more `special` and `non-special` added tokens handling (as far as I understand it)

* Update special token handling

* Add mpt
This commit is contained in:
Galunid 2023-10-23 21:46:00 +02:00 • committed by GitHub
parent 5be6c803fa
commit 69a6735087
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
5 changed files with 56 additions and 19 deletions
Download patch file Download diff file Expand all files Collapse all files

6
convert-mpt-hf-to-gguf.py
View file

@ -136,9 +136,11 @@ for i in range(vocab_size):
tokens.append(f"[PAD{i}]")
toktypes.append(gguf.TokenType.USER_DEFINED)
elif reverse_vocab[i] in added_vocab:
# NOTE: wouldn't we like to distinguish CONTROL tokens here?
tokens.append(reverse_vocab[i])
toktypes.append(gguf.TokenType.USER_DEFINED)
if tokenizer.added_tokens_decoder[i].special:
toktypes.append(gguf.TokenType.CONTROL)
else:
toktypes.append(gguf.TokenType.USER_DEFINED)
else:
tokens.append(reverse_vocab[i])
toktypes.append(gguf.TokenType.NORMAL)