Skip to content

Make convert-pt-to-ggml.py backwards compatible with older vocab.json tokenizer files #1001

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
ggerganov merged 2 commits into from
Jun 25, 2023
Merged

Make convert-pt-to-ggml.py backwards compatible with older vocab.json tokenizer files #1001

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
e979dbb
patch checkpoint convert script to keep compatibility with older hf_t…
akashmjn Jun 11, 2023
06022a1
typo fix
akashmjn Jun 11, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
35 changes: 28 additions & 7 deletions models/convert-pt-to-ggml.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -224,16 +224,39 @@ def bytes_to_unicode():

#code.interact(local=locals())

# load tokenizer
# for backwards compatibility, also check for older hf_transformers format tokenizer files
# old format: dir_whisper/whisper/assets/[multilingual/gpt2]/vocab.json
# new format: dir_whisper/whisper/assets/[multilingual/gpt2].tiktoken
multilingual = hparams["n_vocab"] == 51865
tokenizer = dir_whisper / "whisper" / "assets" / (multilingual and "multilingual.tiktoken" or "gpt2.tiktoken")
tokenizer_type = "tiktoken"
if not tokenizer.is_file():
tokenizer = dir_whisper / "whisper" / "assets" / (multilingual and "multilingual" or "gpt2") / "vocab.json"
tokenizer_type = "hf_transformers"
if not tokenizer.is_file():
print("Error: failed to find either tiktoken or hf_transformers tokenizer file:", tokenizer)
sys.exit(1)

byte_encoder = bytes_to_unicode()
byte_decoder = {v:k for k, v in byte_encoder.items()}

if tokenizer_type == "tiktoken":
with open(tokenizer, "rb") as f:
contents = f.read()
tokens = {base64.b64decode(token): int(rank) for token, rank in (line.split() for line in contents.splitlines() if line)}
elif tokenizer_type == "hf_transformers":
with open(tokenizer, "r", encoding="utf8") as f:
_tokens_raw = json.load(f)
if '<|endoftext|>' in _tokens_raw:
# ensures exact same model as tokenizer_type == tiktoken
# details: https://github.com/ggerganov/whisper.cpp/pull/725
del _tokens_raw['<|endoftext|>']
tokens = {bytes([byte_decoder[c] for c in token]): int(idx) for token, idx in _tokens_raw.items()}

# output in the same directory as the model
fname_out = dir_out / "ggml-model.bin"

with open(tokenizer, "rb") as f:
contents = f.read()
tokens = {base64.b64decode(token): int(rank) for token, rank in (line.split() for line in contents.splitlines() if line)}

# use 16-bit or 32-bit floats
use_f16 = True
if len(sys.argv) > 4:
Expand Down Expand Up @@ -262,9 +285,7 @@ def bytes_to_unicode():
for j in range(filters.shape[1]):
fout.write(struct.pack("f", filters[i][j]))

byte_encoder = bytes_to_unicode()
byte_decoder = {v:k for k, v in byte_encoder.items()}

# write tokenizer
fout.write(struct.pack("i", len(tokens)))

for key in tokens:
Expand Down