Tag BFCL argument mismatch causes

[SonAIengine]

Summary

• add stable failure_tags for BFCL argument_value_mismatch cases
• classify unexpected arguments, optional value mismatches, structured missing values, percentage scale errors, and data-reference substitutions
• document the argument-tag subset loop for faster next-step experiments

Tags

• unexpected_argument
• optional_value_mismatch
• structured_value_missing
• percentage_scale_mismatch
• data_reference_substitution

Validation

• poetry run pytest tests/test_bfcl_tool_selection_llm_loop.py -q
• poetry run pytest tests/test_bfcl_tool_selection_llm_loop.py tests/test_bfcl_tool_selection_hard_cases.py tests/test_bfcl_tool_selection_sweep.py -q
• poetry run ruff check .
• poetry run ruff format --check .
• poetry run pytest tests/ -q
• Qwen3.6-27B argument subset smoke:
  • command: poetry run python -m benchmarks.bfcl_tool_selection.llm_loop --case-ids-file /tmp/gtc-bfcl-qwen027-tag-subsets/failure_argument_value_mismatch.txt --top-k 5 --tool-source retrieved --model qwen3.6-27b --llm-url http://127.0.0.1:18000/v1 --disable-thinking --candidate-selection-guidance --cohesive-namespace-candidates --cache-dir /tmp/gtc-bfcl-argument-tags-cache --cache-namespace argument-tags --timeout 180 --progress --progress-every 1 --output /tmp/gtc-bfcl-argument-tags-subset.json --json
  • cases: 6
  • evaluator_exact_match: 0.166667
  • failure_tag_breakdown: data_reference_substitution=1, optional_value_mismatch=1, percentage_scale_mismatch=1, structured_value_missing=1, unexpected_argument=1
• Hard-case bundle produced tag files under /tmp/gtc-bfcl-qwen027-argument-tags-hardcases/ for each tag.

[SonAIengine]

Closing this internal stacked PR as superseded by the current main branch. The branch is preserved for history.