"model" in args.load_from matched any path containing "model" (for
example a Hugging Face cache path or /data/models/...), so the scripts
loaded ./out/*.pth instead of the transformers checkpoint the user
passed. Compare the last path component with "model" instead.
Signed-off-by: YoungjuKim <fjvbn2003@gmail.com>
execute_tool() relied on signal.SIGALRM, which does not exist on Windows.
The AttributeError was swallowed by a bare except, so every tool call
silently returned None and the model always observed
{"error": "tool not found"} (issue #846).
calculate_math also ran eval() on model-generated text. Two of the three
call sites (scripts/eval_toolcall.py, scripts/web_demo.py) passed no
globals at all, leaving full builtins reachable from sampled output.
safe_math_eval() in trainer_utils.py parses the expression with ast and
walks it, allowing only arithmetic operators and a math whitelist, with
caps on expression length and exponent size. It cannot loop forever, so
the signal-based timeout is no longer needed and is removed. Both sqrt(x)
and math.sqrt(x) spellings keep working.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
YoungjuKim
tkgcc
jingyaogong