Data Acquisition and Curation Intelligence

Estimate the return on external datasets before you spend on full-scale training. Select which dataset(s) you want to try, launch small proxy training runs, and get a performance forecast on full-scale training.

Training task specifications

Task type

classification qa language model pretraining speech recognition

Available external datasets for you to choose

glue/sst2 ag_news

Base model

If you have any existing training data, please upload

Optional training set upload

Hub dataset id (user/dataset)

Evaluation specifications

Eval Metric

loss f1 exact_match

Available public benchmark datasets

If you have any existing benchmark dataset, please upload

Optional test set upload

Test dataset id (user/dataset[:split])

Scaling prediction specifications

Target dataset size for full-scale training

Monitor jobs and open artifacts in the configured results repo once complete.