Skip to content

Fix: Added check the type of tokenizer.json's model_merges #987

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

Open
wants to merge 9 commits into
base: main
Choose a base branch
from
Open

Fix: Added check the type of tokenizer.json's model_merges #987

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
94bdcea
Enhance handling of merges data in HFTokenizerConverter for improved …
N-E-W-T-O-N Aug 9, 2025
deb1721
Add test for Microsoft Phi-4 mini instruct tokenizer
N-E-W-T-O-N Aug 13, 2025
275b820
Update Test Case . Make Ammends as Requested
N-E-W-T-O-N Aug 14, 2025
e797229
Merge branch 'microsoft:main' into merges
N-E-W-T-O-N Aug 19, 2025
a1e8f92
Merge branch 'microsoft:main' into merges
N-E-W-T-O-N Aug 30, 2025
a95f673
Merge branch 'microsoft:main' into merges
N-E-W-T-O-N Sep 7, 2025
c17c8f8
Merge branch 'main' into merges
sayanshaw24 Sep 10, 2025
fd02717
Merge branch 'main' into merges
N-E-W-T-O-N Sep 18, 2025
d1e8286
Merge branch 'main' into merges
N-E-W-T-O-N Sep 26, 2025
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
9 changes: 8 additions & 1 deletion onnxruntime_extensions/_hf_cvt.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -54,8 +54,15 @@ def convert_json_vocab(hf_tokenizer):
# get vocab object from json file
vocab = tokenizer_json.get("model", {}).get("vocab", {})
sorted_merges = tokenizer_json.get("model", {}).get("merges", [])
sorted_merges = [v_.replace("\n", "<0x0A>") for v_ in sorted_merges]

attrs = {"vocab": json.dumps(vocab, separators=(",", ":"))}

# merges data can be a list of string or list of list of string
if (all(isinstance(v_,(list,tuple))) for v_ in sorted_merges) :
sorted_merges = [ " ".join(v if v != "\n" else "<0x0A>" for v in v_ ) for v_ in sorted_merges]
else :
sorted_merges = [v_.replace("\n", "<0x0A>") for v_ in sorted_merges]

attrs["merges"] = "\n".join(sorted_merges)
if hf_tokenizer.added_tokens_encoder:
token_map = [f"{_k}={_v}" for _k,
Expand Down
Loading