Skip to content

Scanning 1 #4044

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

Closed
girichinna27 wants to merge 2 commits into from
+3 −4
Closed

Scanning 1 #4044

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
722fa4b
fix: subprocess-shell-true-2694
Feb 13, 2026
d7db609
Merge pull request #1 from girichinna27/fix/semgrep-subprocess-shell-…
girichinna27 Feb 13, 2026
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
7 changes: 3 additions & 4 deletions unsloth/chat_templates.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2682,16 +2682,15 @@ def test_hf_gguf_equivalence(tokenizer, gguf_model = "./model-unsloth.F16.gguf")

if tokenizer.chat_template is not None:
prompt = tokenizer.apply_chat_template(messages, tokenize = False, add_generation_prompt = True)
prompt = prompt.replace("'", "") # Subprocess does not like ''
prompt = remove_special_tokens(tokenizer, prompt)
Comment on lines 2684 to 2685
Copy link
Copy Markdown

@chatgpt-codex-connector chatgpt-codex-connector Bot Feb 13, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P2 Badge Restore apostrophe normalization for GGUF comparison

This change stops removing ' from chat-template prompts, but test_hf_gguf_equivalence still parses llama-cli output with re.findall(r"([\d]{1,}) \-\> \'([^\']{1,})\'", gguf_tokens), which cannot capture tokens that contain apostrophes. With the current default messages (e.g., It's, I'm), those GGUF tokens are skipped, gguf_tokenized becomes misaligned, and the HF vs GGUF check can fail even when tokenization is actually correct.

Useful? React with 👍 / 👎.

prompts.append(prompt)

for prompt in prompts:
command = f"./llama.cpp/llama-cli -m {gguf_model} -n 0 --temp 0.0 --verbose-prompt "\
f"--check-tensors -p '{prompt}'"
command = ["./llama.cpp/llama-cli", "-m", gguf_model, "-n", "0", "--temp", "0.0", "--verbose-prompt",
"--check-tensors", "-p", prompt]

datas = []
with subprocess.Popen(command, shell = True, stdout = subprocess.PIPE, stderr = subprocess.STDOUT, bufsize = 1) as sp:
with subprocess.Popen(command, shell = False, stdout = subprocess.PIPE, stderr = subprocess.STDOUT, bufsize = 1) as sp:
for line in sp.stdout:
datas.append(line.decode("utf-8", errors = "replace"))
gguf_tokens = "".join(datas)
Expand Down