PolyPythias
- Preview • Updated • 93
EleutherAI/pile-preshuffled-seeds
Updated • 309 • 1Note Training data information for each seed.
-
EleutherAI/pythia-14m-deduped
Text Generation • 39.2M • Updated • 65.9k • 28 -
EleutherAI/pythia-14m-seed1
Text Generation • Updated • 183 -
EleutherAI/pythia-14m-seed2
Text Generation • Updated • 238 -
EleutherAI/pythia-14m-seed3
Text Generation • Updated • 107 -
EleutherAI/pythia-14m-seed4
Text Generation • Updated • 104 -
EleutherAI/pythia-14m-seed5
Text Generation • Updated • 140 -
EleutherAI/pythia-14m-seed6
Text Generation • Updated • 124 -
EleutherAI/pythia-14m-seed7
Text Generation • Updated • 136 -
EleutherAI/pythia-14m-seed8
Text Generation • Updated • 131 -
EleutherAI/pythia-14m-seed9
Text Generation • Updated • 128 -
EleutherAI/pythia-31m-deduped
Text Generation • 55.7M • Updated • 3.4k • 5 -
EleutherAI/pythia-31m-seed1
Text Generation • Updated • 205 -
EleutherAI/pythia-31m-seed2
Text Generation • Updated • 205 -
EleutherAI/pythia-31m-seed3
Text Generation • Updated • 95 -
EleutherAI/pythia-31m-seed4
Text Generation • Updated • 90 -
EleutherAI/pythia-31m-seed5
Text Generation • Updated • 97 -
EleutherAI/pythia-31m-seed6
Text Generation • Updated • 101 -
EleutherAI/pythia-31m-seed7
Text Generation • Updated • 98 -
EleutherAI/pythia-31m-seed8
Text Generation • Updated • 95 -
EleutherAI/pythia-31m-seed9
Text Generation • Updated • 98 -
EleutherAI/pythia-70m
95.6M • Updated • 89.4k • 78 -
EleutherAI/pythia-70m-seed1
Text Generation • Updated • 981 -
EleutherAI/pythia-70m-seed2
Text Generation • Updated • 660 -
EleutherAI/pythia-70m-seed3
Text Generation • Updated • 575 -
EleutherAI/pythia-70m-seed4
Text Generation • Updated • 544 -
EleutherAI/pythia-70m-seed5
Text Generation • Updated • 559 -
EleutherAI/pythia-70m-seed6
Text Generation • Updated • 530 -
EleutherAI/pythia-70m-seed7
Text Generation • Updated • 560 -
EleutherAI/pythia-70m-seed8
Text Generation • Updated • 541 -
EleutherAI/pythia-70m-seed9
Text Generation • Updated • 526 -
EleutherAI/pythia-160m
Text Generation • Updated • 1.69M • 38 -
EleutherAI/pythia-160m-seed1
Text Generation • 0.2B • Updated • 1.22k -
EleutherAI/pythia-160m-seed2
Text Generation • 0.2B • Updated • 1.16k -
EleutherAI/pythia-160m-seed3
Text Generation • 0.2B • Updated • 1.06k -
EleutherAI/pythia-160m-seed4
Text Generation • Updated • 878 • 1 -
EleutherAI/pythia-160m-seed5
Text Generation • Updated • 625 -
EleutherAI/pythia-160m-seed6
Text Generation • Updated • 618 -
EleutherAI/pythia-160m-seed7
Text Generation • Updated • 623 -
EleutherAI/pythia-160m-seed8
Text Generation • Updated • 620 -
EleutherAI/pythia-160m-seed9
Text Generation • Updated • 606 -
EleutherAI/pythia-410m
Text Generation • 0.5B • Updated • 71k • 36 -
EleutherAI/pythia-410m-seed1
Text Generation • Updated • 682 -
EleutherAI/pythia-410m-seed2
Text Generation • Updated • 1.32k -
EleutherAI/pythia-410m-seed3
Text Generation • Updated • 618 -
EleutherAI/pythia-410m-seed4
Text Generation • Updated • 571 -
EleutherAI/pythia-410m-seed5
Text Generation • Updated • 548 -
EleutherAI/pythia-410m-seed6
Text Generation • Updated • 858 • 1 -
EleutherAI/pythia-410m-seed7
Text Generation • Updated • 587 -
EleutherAI/pythia-410m-seed8
Text Generation • Updated • 577 -
EleutherAI/pythia-410m-seed9
Text Generation • Updated • 730
EleutherAI/pythia-160m-data-seed1
Text Generation • Updated • 131Note Version where the data order and weight initialization seeds are decoupled. Here, only the data seed is different from pythia-160m ("seed 0").
EleutherAI/pythia-160m-data-seed2
Text Generation • Updated • 132Note Version where the data order and weight initialization seeds are decoupled. Here, only the data seed is different from pythia-160m ("seed 0").
EleutherAI/pythia-160m-data-seed3
Text Generation • Updated • 135Note Version where the data order and weight initialization seeds are decoupled. Here, only the data seed is different from pythia-160m ("seed 0").
EleutherAI/pythia-160m-weight-seed1
Text Generation • Updated • 247Note Version where the data order and weight initialization seeds are decoupled. Here, only the weight initialization seed is different from pythia-160m ("seed 0").
EleutherAI/pythia-160m-weight-seed2
Text Generation • Updated • 218Note Version where the data order and weight initialization seeds are decoupled. Here, only the weight initialization seed is different from pythia-160m ("seed 0").
EleutherAI/pythia-160m-weight-seed3
Text Generation • Updated • 211Note Version where the data order and weight initialization seeds are decoupled. Here, only the weight initialization seed is different from pythia-160m ("seed 0").
-
PolyPythias: Stability and Outliers across Fifty Language Model Pre-Training Runs
Paper • 2503.09543 • Published