Pietro Lesci PRO
pietrolesci
AI & ML interests
I like developing and applying causal methods to study the effect of training choices on models’ behaviour, including memorisation, shortcut learning, and tokenisation.
Recent Activity
updated
a dataset
8 days ago
pietrolesci/pile-deduped
updated
a collection
22 days ago
The Pile Companion
updated
a collection
22 days ago
The Pile Companion
Organizations
Collections
10
spaces
1
models
21

pietrolesci/me100M_finewebedu-20B_bpe32000minipile
Updated

pietrolesci/me100M-tied_finewebedu-20B_bpe32000minipile
Updated

pietrolesci/me850M_minipile_bpe32000minipile
Updated

pietrolesci/me340M-tied_minipile_bpe32000minipile
Updated

pietrolesci/me57M-tied_minipile_bpe2wp32000minipile
Updated

pietrolesci/me57M-tied_minipile_bpe128000minipile
Updated

pietrolesci/me57M-tied_minipile_wordpiece32000minipile
Updated

pietrolesci/me57M-tied_minipile_bpe8064minipile
Updated

pietrolesci/me57M-tied_minipile_bpe32000minipile
Updated

pietrolesci/tokenisers
Updated
datasets
55
pietrolesci/pile-deduped
Viewer
•
Updated
•
748M
•
9.87k
pietrolesci/pythia-deduped-memorisation-profiles
Viewer
•
Updated
•
2.13M
•
57
pietrolesci/pile-validation
Viewer
•
Updated
•
429k
•
225
pietrolesci/pile-deduped-subset
Viewer
•
Updated
•
16.3k
•
31
pietrolesci/pythia-deduped-stats
Viewer
•
Updated
•
16.3M
•
53
pietrolesci/pythia-deduped-stats-raw
Viewer
•
Updated
•
14.9M
•
14.4k
pietrolesci/agnews
Viewer
•
Updated
•
510k
•
92
pietrolesci/amazoncat-13k
Viewer
•
Updated
•
5.99M
•
349
•
1
pietrolesci/wikitoxic
Viewer
•
Updated
•
894k
•
178
•
1
pietrolesci/multiwoz_all_versions
Viewer
•
Updated
•
82k
•
75
•
1