docs: clean README — drop references to non-public mirror
Browse files
README.md
CHANGED
|
@@ -22,11 +22,11 @@ configs:
|
|
| 22 |
|
| 23 |
# LexBench-Browser
|
| 24 |
|
| 25 |
-
|
| 26 |
|
| 27 |
## Dataset Description
|
| 28 |
|
| 29 |
-
|
| 30 |
|
| 31 |
### Dataset Statistics (v3.0, 2026-04-28)
|
| 32 |
|
|
@@ -143,15 +143,11 @@ huggingface-cli download Lexmount/LexBench-Browser --repo-type dataset --local-d
|
|
| 143 |
- **difficulty**: `easy` | `medium` | `hard`.
|
| 144 |
- **language**: task description language (`zh` for Chinese, `en` for English).
|
| 145 |
- **website_region**: target website region (`zh` for Chinese sites, `en` for international sites).
|
| 146 |
-
- **login_required**: whether the task is login-gated.
|
| 147 |
- **robustness_tags**: ordered, deduplicated list across 6 categories / 16 tags (popup interference, sequence complexity, content dynamics, anti-crawl, localization, complex interaction).
|
| 148 |
|
| 149 |
> **Migration note (v2.x → v3.0)**: the per-record `scenario_tier` field has been removed and the per-tier files (`l1.jsonl`, `l2.jsonl`, `l3-api.jsonl`, `l3-security.jsonl`, `tasks.jsonl`) have been replaced by a single `task.jsonl`. Slice the data with `login_required`, `domain`, or `risk_control` instead.
|
| 150 |
|
| 151 |
-
## Related Resources
|
| 152 |
-
|
| 153 |
-
- [Lexmount/LexBench-Browser-Private](https://huggingface.co/datasets/Lexmount/LexBench-Browser-Private) — full dataset including login-gated tasks (377 tasks).
|
| 154 |
-
|
| 155 |
## License
|
| 156 |
|
| 157 |
Apache 2.0
|
|
|
|
| 22 |
|
| 23 |
# LexBench-Browser
|
| 24 |
|
| 25 |
+
LexBench-Browser is a benchmark for evaluating AI browser agents on real-world web tasks across 50+ Chinese and English websites — e-commerce, social, video, tools/education, finance, and more.
|
| 26 |
|
| 27 |
## Dataset Description
|
| 28 |
|
| 29 |
+
This release contains **220 tasks** spanning common multi-step browsing flows. None of the tasks gate on user login, so they can be benchmarked end-to-end without manual sign-in.
|
| 30 |
|
| 31 |
### Dataset Statistics (v3.0, 2026-04-28)
|
| 32 |
|
|
|
|
| 143 |
- **difficulty**: `easy` | `medium` | `hard`.
|
| 144 |
- **language**: task description language (`zh` for Chinese, `en` for English).
|
| 145 |
- **website_region**: target website region (`zh` for Chinese sites, `en` for international sites).
|
| 146 |
+
- **login_required**: whether the task is login-gated.
|
| 147 |
- **robustness_tags**: ordered, deduplicated list across 6 categories / 16 tags (popup interference, sequence complexity, content dynamics, anti-crawl, localization, complex interaction).
|
| 148 |
|
| 149 |
> **Migration note (v2.x → v3.0)**: the per-record `scenario_tier` field has been removed and the per-tier files (`l1.jsonl`, `l2.jsonl`, `l3-api.jsonl`, `l3-security.jsonl`, `tasks.jsonl`) have been replaced by a single `task.jsonl`. Slice the data with `login_required`, `domain`, or `risk_control` instead.
|
| 150 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 151 |
## License
|
| 152 |
|
| 153 |
Apache 2.0
|