Replace per-project files with combined data/train.jsonl (746 rows) + dataset card

README.md

@@ -29,9 +29,9 @@ A benchmark for **test-case generation** in evolving Java codebases. Each row ca
 
 ## Dataset summary
 
-`teb-generation` targets the **fail-error-old** subset of TestEvo-Bench: pairs where the *new* test, when run against the *old* production code (revision `rev1`), fails or errors. These are the cases where authoring a new test genuinely requires the changes that landed in `rev2` — the test cannot be satisfied by old code, so a model must read the production-side diff to write a correct test.
+`teb-generation` is a benchmark for generating new test methods that exercise newly-introduced or changed production code. Every example was mined from a real commit (`rev1` → `rev2`) on a public Java project, and the new test is one that depends on the production-code change to behave correctly: it cannot be written or satisfied by reading only the pre-change codebase.
 
-Each row corresponds to a single (project, `rev1` → `rev2`) pair. Inside the row, `test_changes` is a list of one or more individual test methods that were added or modified in that revision, each with its own focal-method analysis.
+Each row corresponds to a single (project, `rev1` → `rev2`) pair. Inside the row, `test_changes` is a list of one or more individual test methods authored in that commit, each paired with its focal production method, the production-side before/after, candidate dependency methods, and the dependencies that changed in the same commit.
 
 | | |
 |---|---|