This table compares the performance of different edit formats (diff-fenced
and CEDARScript
)
for various benchmark tasks in Gemini 1.5 PRO
conversations.
Task | Edit Format | Lines | Result |
---|---|---|---|
doc_DocCLI_get_role_man_text | diff-fenced | 4948 | FAILED (failures=1) |
doc_DocCLI_get_role_man_text | CEDARScript | 45 | OK |
analyzer_cli_DebugAnalyzer__make_source_table | diff-fenced | 4497 | FAILED (failures=1) |
analyzer_cli_DebugAnalyzer__make_source_table | CEDARScript | 164 | OK |