fix: Fix typos in HumanEval evaluation code

[yurekami]

Summary

Fixes several typos in the HumanEval evaluation code that could cause confusion.

Changes

Evaluation/HumanEval/utils/utils.py

• Renamed languge_settings → language_settings (variable name typo)

Evaluation/HumanEval/eval_instruct.py

• Updated import to use corrected language_settings name
• Renamed function parameter languge → language in build_deepseekcoder_instruction()
• Fixed error message typo: "Not fond" → "Not found"
• Fixed argparse help text typo: "langauge" → "language"

Diff Summary

-languge_settings = {
+language_settings = {

-def build_deepseekcoder_instruction(languge: str, question: str):
+def build_deepseekcoder_instruction(language: str, question: str):

-'''.strip().format(languge.lower(), question.strip())
+'''.strip().format(language.lower(), question.strip())

-from utils.utils import extract_generation_code, languge_settings
+from utils.utils import extract_generation_code, language_settings

-    prompt = build_deepseekcoder_instruction(languge_settings[lang]['full_name'], ...)
+    prompt = build_deepseekcoder_instruction(language_settings[lang]['full_name'], ...)

-    assert os.path.exists(args.output_path), "Not fond output file: {}"
+    assert os.path.exists(args.output_path), "Not found output file: {}"

-    parser.add_argument('--language', type=str, help="langauge")
+    parser.add_argument('--language', type=str, help="language")

Test plan

• Verify Python syntax is valid
• Run HumanEval evaluation to verify no regressions

🤖 Generated with Claude Code