PolyPythias
- Preview • Updated • 964
EleutherAI/pile-preshuffled-seeds
Updated • 938 • 1Note Training data information for each seed.
-
EleutherAI/pythia-14m-deduped
Text Generation • 39.2M • Updated • 5.33k • 32 -
EleutherAI/pythia-14m-seed1
Text Generation • Updated • 2.31k • 1 -
EleutherAI/pythia-14m-seed2
Text Generation • Updated • 1.98k -
EleutherAI/pythia-14m-seed3
Text Generation • Updated • 1.49k -
EleutherAI/pythia-14m-seed4
Text Generation • Updated • 1.37k -
EleutherAI/pythia-14m-seed5
Text Generation • Updated • 1.5k -
EleutherAI/pythia-14m-seed6
Text Generation • Updated • 1.45k -
EleutherAI/pythia-14m-seed7
Text Generation • Updated • 1.37k -
EleutherAI/pythia-14m-seed8
Text Generation • Updated • 1.27k -
EleutherAI/pythia-14m-seed9
Text Generation • Updated • 1.36k -
EleutherAI/pythia-31m-deduped
Text Generation • 55.7M • Updated • 1.73k • 5 -
EleutherAI/pythia-31m-seed1
Text Generation • Updated • 1.75k -
EleutherAI/pythia-31m-seed2
Text Generation • Updated • 1.53k -
EleutherAI/pythia-31m-seed3
Text Generation • Updated • 1.41k -
EleutherAI/pythia-31m-seed4
Text Generation • Updated • 1.32k -
EleutherAI/pythia-31m-seed5
Text Generation • Updated • 1.36k -
EleutherAI/pythia-31m-seed6
Text Generation • Updated • 1.41k -
EleutherAI/pythia-31m-seed7
Text Generation • Updated • 1.2k -
EleutherAI/pythia-31m-seed8
Text Generation • Updated • 1.22k -
EleutherAI/pythia-31m-seed9
Text Generation • Updated • 1.25k -
EleutherAI/pythia-70m
95.6M • Updated • 231k • 88 -
EleutherAI/pythia-70m-seed1
Text Generation • Updated • 4.12k -
EleutherAI/pythia-70m-seed2
Text Generation • Updated • 3.43k -
EleutherAI/pythia-70m-seed3
Text Generation • Updated • 2.63k -
EleutherAI/pythia-70m-seed4
Text Generation • Updated • 2.11k -
EleutherAI/pythia-70m-seed5
Text Generation • Updated • 1.86k -
EleutherAI/pythia-70m-seed6
Text Generation • Updated • 2.05k -
EleutherAI/pythia-70m-seed7
Text Generation • Updated • 1.92k -
EleutherAI/pythia-70m-seed8
Text Generation • Updated • 1.72k -
EleutherAI/pythia-70m-seed9
Text Generation • Updated • 1.91k -
EleutherAI/pythia-160m
Text Generation • 0.2B • Updated • 3.45M • 66 -
EleutherAI/pythia-160m-seed1
Text Generation • 0.2B • Updated • 18k -
EleutherAI/pythia-160m-seed2
Text Generation • 0.2B • Updated • 16.2k -
EleutherAI/pythia-160m-seed3
Text Generation • 0.2B • Updated • 15.4k -
EleutherAI/pythia-160m-seed4
Text Generation • Updated • 2.94k • 1 -
EleutherAI/pythia-160m-seed5
Text Generation • Updated • 2.69k -
EleutherAI/pythia-160m-seed6
Text Generation • Updated • 2.07k -
EleutherAI/pythia-160m-seed7
Text Generation • Updated • 2.11k -
EleutherAI/pythia-160m-seed8
Text Generation • Updated • 2.07k -
EleutherAI/pythia-160m-seed9
Text Generation • Updated • 2.13k -
EleutherAI/pythia-410m
Text Generation • 0.5B • Updated • 94.1k • 41 -
EleutherAI/pythia-410m-seed1
Text Generation • Updated • 3.92k -
EleutherAI/pythia-410m-seed2
Text Generation • Updated • 3.03k -
EleutherAI/pythia-410m-seed3
Text Generation • Updated • 3.01k -
EleutherAI/pythia-410m-seed4
Text Generation • Updated • 2.78k -
EleutherAI/pythia-410m-seed5
Text Generation • Updated • 2.35k -
EleutherAI/pythia-410m-seed6
Text Generation • Updated • 2.21k • 1 -
EleutherAI/pythia-410m-seed7
Text Generation • Updated • 2.14k -
EleutherAI/pythia-410m-seed8
Text Generation • Updated • 1.94k -
EleutherAI/pythia-410m-seed9
Text Generation • Updated • 2.11k
EleutherAI/pythia-160m-data-seed1
Text Generation • Updated • 2kNote 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 • 1.79kNote 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 • 1.71kNote 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 • 2.07kNote 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 • 1.96kNote 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 • 1.4kNote 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
EleutherAI/pythia-160m-weight-seed1-corrected
Text Generation • Updated • 386Note Corrected weight-seed 1: independent initialization with the original Pythia data order.
EleutherAI/pythia-160m-weight-seed2-corrected
Text Generation • Updated • 711Note Corrected weight-seed 1: independent initialization with the original Pythia data order.