MMLU
Harbor adapterMMLU evaluates English language models on knowledge and reasoning across 57 diverse subjects using a multiple-choice format with over 14k samples.
- Samples
- 14,042
- Released
- 2020
- Scores
- 209
WebLINX
Real-World Website Navigation with Multi-Turn
WebLINX is a large-scale benchmark of 100K interactions across 2300 expert demonstrations of conversational web navigation.
How to run
ltb · weblinx
$ npm i -g lettokenburn-cli$ ltb bench info weblinx
External harness. No Harbor adapter has been generated for this benchmark yet. `ltb bench info` prints the known links.
Runs happen on your machine with your own subscription; only the score summary is pushed. CLI docs
Scores
8 sourced entries| Harness | Source | Confidence | Link | |||
|---|---|---|---|---|---|---|
| LLaMA 2 13B | 25.2 | N/A | Independent | medium | Feb 8, 2024 | arxiv.org |
| LLaMA 2 7B | 24.6 | N/A | Independent | medium | Feb 8, 2024 | arxiv.org |
| Flan-T5-3B | 23.8 | N/A | Independent | medium | Feb 8, 2024 | arxiv.org |
| GPT-3.5F | 21.2 | N/A | Independent | medium | Feb 8, 2024 | arxiv.org |
| Flan-T5-780M | 17.3 | N/A | Independent | medium | Feb 8, 2024 | arxiv.org |
| Flan-T5-Base 250M | 15.0 | N/A | Independent | medium | Feb 8, 2024 | arxiv.org |
| GPT-4V | 10.4 | N/A | Independent | medium | Feb 8, 2024 | arxiv.org |
| GPT-3.5T | 8.5 | N/A | Independent | medium | Feb 8, 2024 | arxiv.org |
Every number links to where it was published. Low-confidence rows are muted and need per-cell verification.
Reproduced by the community
All runsLoading community runs…
About
WebLINX is a large-scale benchmark of 100K interactions across 2300 expert demonstrations of conversational web navigation. It covers a broad range of patterns on over 150 real-world websites and can be used to train and evaluate agents in diverse scenarios.