EvoWeb: the benchmark, its translations and the no-grounding ablation set, with a dataset card

README.md

@@ -1,3 +1,122 @@
 ---
 license: cc-by-4.0
+pretty_name: EvoWeb
+language:
+  - en
+  - fr
+  - de
+  - zh
+  - hi
+  - he
+  - es
+  - sw
+task_categories:
+  - text-generation
+tags:
+  - code
+  - coding-agents
+  - web-development
+  - benchmark
+  - agents
+  - multi-turn
+size_categories:
+  - n<1K
+configs:
+  - config_name: default
+    data_files: evoweb_benchmark.jsonl
+  - config_name: ablation_no_grounding
+    data_files: ablation_no_grounding.jsonl
+  - config_name: multilingual
+    data_files:
+      - split: fr
+        path: multilingual/evoweb_benchmark_fr.jsonl
+      - split: de
+        path: multilingual/evoweb_benchmark_de.jsonl
+      - split: zh
+        path: multilingual/evoweb_benchmark_zh.jsonl
+      - split: hi
+        path: multilingual/evoweb_benchmark_hi.jsonl
+      - split: he
+        path: multilingual/evoweb_benchmark_he.jsonl
+      - split: es
+        path: multilingual/evoweb_benchmark_es.jsonl
+      - split: sw
+        path: multilingual/evoweb_benchmark_sw.jsonl
 ---
+
+# EvoWeb: a benchmark for coding agents on evolving web applications
+
+EvoWeb gives a coding agent 32 web apps to build and then extend, request by request: a first
+build and nine follow-up requests per app, each adding a feature that may depend on earlier ones.
+After every request a browser agent tests the running app with five checks. **Grade** summarises
+how much working functionality survives the whole sequence; **Survival** counts how many requests
+in a row stay usable.
+
+The harness that runs the benchmark (OpenHands as the agent, WebVoyager as the judge), the
+scoring code and the released traces of the paper's runs are at
+**[github.com/two-delta/evo](https://github.com/two-delta/evo)**. This dataset is the benchmark
+itself, as one row per app.
+
+## Configurations
+
+| config | rows | what |
+|---|---|---|
+| `default` | 32 | the benchmark: 32 apps x 10 requests, with the judge's checks |
+| `multilingual` (splits `fr de zh hi he es sw`) | 32 each | the same apps with the user's requests translated; the app's UI labels, seeded data and the judge's checks stay in English |
+| `ablation_no_grounding` | 30 | 30 of the apps with follow-up requests written without the real-website corpus (paper App. C.1) |
+
+```python
+from datasets import load_dataset
+
+evoweb = load_dataset("TD-Labs/EvoWeb", split="train")                 # the benchmark
+french = load_dataset("TD-Labs/EvoWeb", "multilingual", split="fr")
+```
+
+**Row order matters.** A sample index in a run or a trace is the 1-based row position in the file
+the run used, and the scorer keys task counts by that position. Never re-sort a configuration.
+
+## Schema of one row
+
+```
+id                  "000001"  zero-padded WebGen-Bench app id (not the sample index)
+instruction         the first (build) request
+Category, application_type      WebGen-Bench metadata, unused by the harness
+ui_instruct         [ {task, expected_result, task_category} ]  the judge's checks for the first request
+follow_ups          [ 9 x {                                   request k = follow_ups[k-2], k = 2..10
+    difficulty      "hard"
+    archetype       derived-state | multi-surface-workflow | interactive-tool | design-restyle
+    instruction     the user's request
+    source_website, source_feature      the real product feature the request was grounded in
+                    (ablation_no_grounding carries feature_rationale instead)
+    ui_instruct     [ 5 x {kind, task, expected_result, task_category, [copied_from_turn]} ]
+  } ]
+dependency_graph    {nodes: [{id: "T1".."T10", label, archetype}], edges: [{from, to, what}]}
+global_rules        [ 3 x {rule, introduced_in_turn} ]   app-wide rules the requests introduce
+pinned_seed         {entity, introduced_in_turn, records_gist, [records]}
+                    the fixed demo data every exact-value check is written against
+                    (`records`, present for 3 apps, is a JSON-encoded string here)
+target_app_summary  prose description of the app the 10 requests build towards
+```
+
+Every follow-up request carries exactly five checks in a fixed order: three **functional**
+(`exact-value` or `feature`), one **integration** (the new feature combined with an earlier one,
+along an edge of `dependency_graph`, or one of the `global_rules` when the request has no incoming
+edge), and one **backward** (`regression`, a check copied from an earlier request).
+
+## How it was built
+
+The 32 apps are WebGen-Bench seeds. For each, an LLM workflow grounded in real similar websites
+designed the grown app and its dependency DAG, wrote the nine follow-up requests with their
+checks, and was audited and repaired; the generator, its audit rules and the assembly scripts are
+in the repository under `data/data_gen/`.
+
+## Citation
+
+```bibtex
+@inproceedings{evoweb2027,
+  title     = {EvoWeb: A Benchmark for Coding Agents on Evolving Web Applications},
+  author    = {Anonymous authors},
+  booktitle = {International Conference on Learning Representations (ICLR), under review},
+  year      = {2027}
+}
+```