The Pile corpus for measuring lanugage model performance across various domains [(Gao et al., 2020)](https://arxiv.org/pdf/2101.00027.pdf).

the_pile

The Pile corpus for measuring lanugage model performance across various domains [(Gao et al., 2020)](https://arxiv.org/pdf/2101.00027.pdf).

Bits/byte: Average number of bits per byte according to model probabilities.

The Pile corpus for measuring lanugage model performance across various domains [(Gao et al., 2020)](https://arxiv.org/pdf/2101.00027.pdf).

Denoised inference runtime (s): Average time to process a request to the model minus performance contention by using profiled runtimes from multiple trials of SyntheticEfficiencyScenario.

The Pile corpus for measuring lanugage model performance across various domains [(Gao et al., 2020)](https://arxiv.org/pdf/2101.00027.pdf).

# eval: Number of evaluation instances.

The Pile corpus for measuring lanugage model performance across various domains [(Gao et al., 2020)](https://arxiv.org/pdf/2101.00027.pdf).

# train: Number of training instances (e.g., in-context examples).

The Pile corpus for measuring lanugage model performance across various domains [(Gao et al., 2020)](https://arxiv.org/pdf/2101.00027.pdf).

truncated: Fraction of instances where the prompt itself was truncated (implies that there were no in-context examples).

The Pile corpus for measuring lanugage model performance across various domains [(Gao et al., 2020)](https://arxiv.org/pdf/2101.00027.pdf).

# prompt tokens: Number of tokens in the prompt.

The Pile corpus for measuring lanugage model performance across various domains [(Gao et al., 2020)](https://arxiv.org/pdf/2101.00027.pdf).

# output tokens: Actual number of output tokens.

The Pile corpus for measuring lanugage model performance across various domains [(Gao et al., 2020)](https://arxiv.org/pdf/2101.00027.pdf).

# trials: Number of trials, where in each trial we choose an independent, random set of training instances.

min=0.28, mean=0.545, max=0.745, sum=3.268 (6)

min=0.422, mean=0.848, max=1.13, sum=5.085 (6)

min=361.567, mean=1374.206, max=2046.982, sum=8245.234 (6)

min=0.358, mean=0.645, max=0.833, sum=3.87 (6)

min=0.35, mean=0.642, max=0.835, sum=3.852 (6)

min=0.314, mean=0.6, max=0.792, sum=3.603 (6)

min=0.426, mean=0.753, max=0.969, sum=4.52 (6)

min=0.333, mean=0.602, max=0.767, sum=3.611 (6)

min=0.241, mean=0.533, max=0.715, sum=3.198 (6)

min=352.627, mean=3256.852, max=5999.966, sum=19541.11 (6)

min=0.322, mean=0.593, max=0.762, sum=3.555 (6)

min=0.343, mean=0.639, max=0.81, sum=3.832 (6)

Anthropic is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2112.00861.

min=0.336, mean=0.597, max=0.874, sum=3.583 (6)
☠ Anthropic is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2112.00861.

min=0.771, mean=2.911, max=9.758, sum=17.464 (6)
☠ Anthropic is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2112.00861.

min=28, mean=492, max=1000, sum=2952 (6)
☠ Anthropic is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2112.00861.

min=0, mean=0, max=0, sum=0 (6)
☠ Anthropic is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2112.00861.

min=426.794, mean=4398.18, max=8191.98, sum=26389.078 (6)
☠ Anthropic is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2112.00861.

min=0.011, mean=24.407, max=128.85, sum=146.444 (6)
☠ Anthropic is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2112.00861.

min=1, mean=1, max=1, sum=6 (6)
☠ Anthropic is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2112.00861.

BLOOM is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://huggingface.co/spaces/bigscience/BigScienceCorpus.

min=0.269, mean=0.571, max=0.828, sum=3.423 (6)
☠ BLOOM is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://huggingface.co/spaces/bigscience/BigScienceCorpus.

min=0.628, mean=0.969, max=1.172, sum=5.813 (6)
☠ BLOOM is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://huggingface.co/spaces/bigscience/BigScienceCorpus.

min=28, mean=492, max=1000, sum=2952 (6)
☠ BLOOM is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://huggingface.co/spaces/bigscience/BigScienceCorpus.

min=0, mean=0, max=0, sum=0 (6)
☠ BLOOM is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://huggingface.co/spaces/bigscience/BigScienceCorpus.

min=378.692, mean=1376.41, max=2048.999, sum=8258.459 (6)
☠ BLOOM is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://huggingface.co/spaces/bigscience/BigScienceCorpus.

min=1, mean=1, max=1, sum=6 (6)
☠ BLOOM is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://huggingface.co/spaces/bigscience/BigScienceCorpus.

min=0.565, mean=0.757, max=0.936, sum=2.271 (3)

min=0.471, mean=0.654, max=0.944, sum=1.961 (3)

min=28, mean=376.667, max=1000, sum=1130 (3)

min=446.848, mean=1058.016, max=2039.941, sum=3174.048 (3)

min=0.64, mean=0.811, max=0.984, sum=2.432 (3)

min=0.31, mean=0.456, max=0.69, sum=1.367 (3)

min=0.68, mean=0.845, max=1.017, sum=2.534 (3)

min=0.268, mean=0.37, max=0.533, sum=1.11 (3)

min=0.837, mean=0.996, max=1.182, sum=2.987 (3)

min=0.283, mean=0.394, max=0.572, sum=1.181 (3)

min=0.544, mean=0.741, max=0.932, sum=2.224 (3)

min=0.711, mean=0.858, max=1.049, sum=2.574 (3)

min=0.735, mean=0.875, max=1.053, sum=2.625 (3)

min=445.514, mean=1049.308, max=2012.12, sum=3147.923 (3)

min=0.998, mean=1.001, max=1.004, sum=3.002 (3)

min=0.611, mean=0.781, max=0.961, sum=2.344 (3)

GPT-J is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arankomatsuzaki.wordpress.com/2021/06/04/gpt-j/.

min=0.234, mean=0.49, max=0.701, sum=2.942 (6)
☠ GPT-J is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arankomatsuzaki.wordpress.com/2021/06/04/gpt-j/.

min=0.45, mean=0.455, max=0.458, sum=2.727 (6)
☠ GPT-J is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arankomatsuzaki.wordpress.com/2021/06/04/gpt-j/.

min=28, mean=492, max=1000, sum=2952 (6)
☠ GPT-J is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arankomatsuzaki.wordpress.com/2021/06/04/gpt-j/.

min=0, mean=0, max=0, sum=0 (6)
☠ GPT-J is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arankomatsuzaki.wordpress.com/2021/06/04/gpt-j/.

min=438.952, mean=1442.192, max=2048.976, sum=8653.149 (6)
☠ GPT-J is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arankomatsuzaki.wordpress.com/2021/06/04/gpt-j/.

min=1, mean=1, max=1, sum=6 (6)
☠ GPT-J is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arankomatsuzaki.wordpress.com/2021/06/04/gpt-j/.

GPT-NeoX is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2204.06745.

min=0.212, mean=0.466, max=0.677, sum=2.796 (6)
☠ GPT-NeoX is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2204.06745.

min=0.85, mean=0.902, max=0.94, sum=5.412 (6)
☠ GPT-NeoX is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2204.06745.

min=28, mean=492, max=1000, sum=2952 (6)
☠ GPT-NeoX is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2204.06745.

min=0, mean=0, max=0, sum=0 (6)
☠ GPT-NeoX is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2204.06745.

min=408.865, mean=1407.518, max=2048.969, sum=8445.108 (6)
☠ GPT-NeoX is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2204.06745.

min=1, mean=1, max=1, sum=6 (6)
☠ GPT-NeoX is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2204.06745.

OPT is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2205.01068.

min=0.443, mean=0.592, max=0.73, sum=3.554 (6)
☠ OPT is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2205.01068.

min=0.751, mean=0.829, max=0.884, sum=4.975 (6)
☠ OPT is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2205.01068.

min=28, mean=492, max=1000, sum=2952 (6)
☠ OPT is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2205.01068.

min=0, mean=0, max=0, sum=0 (6)
☠ OPT is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2205.01068.

min=438.952, mean=1442.191, max=2048.976, sum=8653.145 (6)
☠ OPT is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2205.01068.

min=1, mean=1, max=1, sum=6 (6)
☠ OPT is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2205.01068.

min=0.468, mean=0.618, max=0.756, sum=3.706 (6)
☠ OPT is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2205.01068.

min=0.323, mean=0.495, max=0.782, sum=2.973 (6)
☠ OPT is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2205.01068.

MT-NLG is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2201.11990.

min=0.319, mean=0.61, max=0.872, sum=3.661 (6)
☠ MT-NLG is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2201.11990.

min=28, mean=492, max=1000, sum=2952 (6)
☠ MT-NLG is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2201.11990.

min=0, mean=0, max=0, sum=0 (6)
☠ MT-NLG is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2201.11990.

min=438.905, mean=1441.58, max=2047.976, sum=8649.479 (6)
☠ MT-NLG is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2201.11990.

min=0, mean=0.007, max=0.02, sum=0.044 (6)
☠ MT-NLG is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2201.11990.

min=1, mean=1, max=1, sum=6 (6)
☠ MT-NLG is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2201.11990.

min=0.418, mean=0.704, max=0.981, sum=4.226 (6)
☠ MT-NLG is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2201.11990.

min=0.515, mean=0.713, max=0.979, sum=4.278 (6)

min=0.205, mean=0.226, max=0.239, sum=1.359 (6)

min=438.952, mean=1442.192, max=2048.976, sum=8653.149 (6)

min=0.587, mean=0.789, max=1.054, sum=4.735 (6)

min=0.091, mean=0.109, max=0.12, sum=0.657 (6)

min=0.687, mean=0.866, max=1.123, sum=5.194 (6)

min=0.118, mean=0.128, max=0.134, sum=0.766 (6)

min=0.823, mean=0.96, max=1.212, sum=5.763 (6)

min=0.14, mean=0.142, max=0.144, sum=0.854 (6)

min=0.205, mean=0.576, max=0.863, sum=3.455 (6)

min=447.971, mean=2476.022, max=4000.966, sum=14856.129 (6)

min=0.214, mean=0.578, max=0.827, sum=3.466 (6)

min=0.189, mean=0.279, max=0.347, sum=1.675 (6)

min=0.677, mean=0.969, max=1.275, sum=5.811 (6)

min=0.133, mean=0.147, max=0.156, sum=0.885 (6)

min=0.814, mean=1.103, max=1.435, sum=6.62 (6)

min=0.132, mean=0.155, max=0.169, sum=0.93 (6)

min=1.156, mean=1.61, max=2.003, sum=9.661 (6)

min=0.087, mean=0.108, max=0.12, sum=0.645 (6)

min=408.865, mean=1407.518, max=2048.969, sum=8445.108 (6)

The Pile

the_pile_subset:ArXiv

min=0.618, mean=0.618, max=0.618, sum=0.618 (1)

min=1.126, mean=1.126, max=1.126, sum=1.126 (1)

min=2036.998, mean=2036.998, max=2036.998, sum=2036.998 (1)

min=0.679, mean=0.679, max=0.679, sum=0.679 (1)

min=0.832, mean=0.832, max=0.832, sum=0.832 (1)

min=0.643, mean=0.643, max=0.643, sum=0.643 (1)

min=0.966, mean=0.966, max=0.966, sum=0.966 (1)

min=0.628, mean=0.628, max=0.628, sum=0.628 (1)

min=0.537, mean=0.537, max=0.537, sum=0.537 (1)

min=5765.889, mean=5765.889, max=5765.889, sum=5765.889 (1)

min=0.661, mean=0.661, max=0.661, sum=0.661 (1)

min=0.529, mean=0.529, max=0.529, sum=0.529 (1)
☠ Anthropic is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2112.00861.

min=2.381, mean=2.381, max=2.381, sum=2.381 (1)
☠ Anthropic is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2112.00861.

min=238, mean=238, max=238, sum=238 (1)
☠ Anthropic is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2112.00861.

min=0, mean=0, max=0, sum=0 (1)
☠ Anthropic is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2112.00861.

min=7832.227, mean=7832.227, max=7832.227, sum=7832.227 (1)
☠ Anthropic is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2112.00861.

min=0.011, mean=0.011, max=0.011, sum=0.011 (1)
☠ Anthropic is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2112.00861.

min=1, mean=1, max=1, sum=1 (1)
☠ Anthropic is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2112.00861.

min=0.577, mean=0.577, max=0.577, sum=0.577 (1)
☠ BLOOM is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://huggingface.co/spaces/bigscience/BigScienceCorpus.

min=1.168, mean=1.168, max=1.168, sum=1.168 (1)
☠ BLOOM is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://huggingface.co/spaces/bigscience/BigScienceCorpus.

min=238, mean=238, max=238, sum=238 (1)
☠ BLOOM is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://huggingface.co/spaces/bigscience/BigScienceCorpus.

min=0, mean=0, max=0, sum=0 (1)
☠ BLOOM is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://huggingface.co/spaces/bigscience/BigScienceCorpus.

min=2039.713, mean=2039.713, max=2039.713, sum=2039.713 (1)
☠ BLOOM is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://huggingface.co/spaces/bigscience/BigScienceCorpus.

min=1, mean=1, max=1, sum=1 (1)
☠ BLOOM is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://huggingface.co/spaces/bigscience/BigScienceCorpus.

min=0.525, mean=0.525, max=0.525, sum=0.525 (1)
☠ GPT-J is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arankomatsuzaki.wordpress.com/2021/06/04/gpt-j/.

min=0.456, mean=0.456, max=0.456, sum=0.456 (1)
☠ GPT-J is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arankomatsuzaki.wordpress.com/2021/06/04/gpt-j/.

min=238, mean=238, max=238, sum=238 (1)
☠ GPT-J is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arankomatsuzaki.wordpress.com/2021/06/04/gpt-j/.

min=0, mean=0, max=0, sum=0 (1)
☠ GPT-J is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arankomatsuzaki.wordpress.com/2021/06/04/gpt-j/.

min=2041.731, mean=2041.731, max=2041.731, sum=2041.731 (1)
☠ GPT-J is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arankomatsuzaki.wordpress.com/2021/06/04/gpt-j/.

min=1, mean=1, max=1, sum=1 (1)
☠ GPT-J is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arankomatsuzaki.wordpress.com/2021/06/04/gpt-j/.

min=0.503, mean=0.503, max=0.503, sum=0.503 (1)
☠ GPT-NeoX is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2204.06745.

min=0.933, mean=0.933, max=0.933, sum=0.933 (1)
☠ GPT-NeoX is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2204.06745.

min=238, mean=238, max=238, sum=238 (1)
☠ GPT-NeoX is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2204.06745.

min=0, mean=0, max=0, sum=0 (1)
☠ GPT-NeoX is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2204.06745.

min=2040.215, mean=2040.215, max=2040.215, sum=2040.215 (1)
☠ GPT-NeoX is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2204.06745.

min=1, mean=1, max=1, sum=1 (1)
☠ GPT-NeoX is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2204.06745.

min=0.73, mean=0.73, max=0.73, sum=0.73 (1)
☠ OPT is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2205.01068.

min=0.881, mean=0.881, max=0.881, sum=0.881 (1)
☠ OPT is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2205.01068.

min=238, mean=238, max=238, sum=238 (1)
☠ OPT is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2205.01068.

min=0, mean=0, max=0, sum=0 (1)
☠ OPT is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2205.01068.

min=2041.731, mean=2041.731, max=2041.731, sum=2041.731 (1)
☠ OPT is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2205.01068.

min=1, mean=1, max=1, sum=1 (1)
☠ OPT is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2205.01068.

min=0.756, mean=0.756, max=0.756, sum=0.756 (1)
☠ OPT is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2205.01068.

min=0.33, mean=0.33, max=0.33, sum=0.33 (1)
☠ OPT is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2205.01068.

min=0.608, mean=0.608, max=0.608, sum=0.608 (1)
☠ MT-NLG is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2201.11990.

min=238, mean=238, max=238, sum=238 (1)
☠ MT-NLG is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2201.11990.

min=0, mean=0, max=0, sum=0 (1)
☠ MT-NLG is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2201.11990.

min=2040.741, mean=2040.741, max=2040.741, sum=2040.741 (1)
☠ MT-NLG is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2201.11990.

min=0.004, mean=0.004, max=0.004, sum=0.004 (1)
☠ MT-NLG is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2201.11990.

min=1, mean=1, max=1, sum=1 (1)
☠ MT-NLG is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2201.11990.

min=0.701, mean=0.701, max=0.701, sum=0.701 (1)
☠ MT-NLG is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2201.11990.

min=0.788, mean=0.788, max=0.788, sum=0.788 (1)

min=0.239, mean=0.239, max=0.239, sum=0.239 (1)

min=2041.731, mean=2041.731, max=2041.731, sum=2041.731 (1)

min=0.865, mean=0.865, max=0.865, sum=0.865 (1)

min=0.12, mean=0.12, max=0.12, sum=0.12 (1)

min=0.95, mean=0.95, max=0.95, sum=0.95 (1)

min=0.134, mean=0.134, max=0.134, sum=0.134 (1)

min=1.055, mean=1.055, max=1.055, sum=1.055 (1)

min=0.144, mean=0.144, max=0.144, sum=0.144 (1)

min=0.555, mean=0.555, max=0.555, sum=0.555 (1)

min=3965.679, mean=3965.679, max=3965.679, sum=3965.679 (1)

min=0.563, mean=0.563, max=0.563, sum=0.563 (1)

min=0.346, mean=0.346, max=0.346, sum=0.346 (1)

min=1.071, mean=1.071, max=1.071, sum=1.071 (1)

min=0.156, mean=0.156, max=0.156, sum=0.156 (1)

min=1.229, mean=1.229, max=1.229, sum=1.229 (1)

min=0.169, mean=0.169, max=0.169, sum=0.169 (1)

min=1.771, mean=1.771, max=1.771, sum=1.771 (1)

min=2040.215, mean=2040.215, max=2040.215, sum=2040.215 (1)

subset: ArXiv

the_pile_subset:BookCorpus2

min=0.745, mean=0.745, max=0.745, sum=0.745 (1)

min=1.13, mean=1.13, max=1.13, sum=1.13 (1)

min=2046.982, mean=2046.982, max=2046.982, sum=2046.982 (1)

min=0.833, mean=0.833, max=0.833, sum=0.833 (1)

min=0.835, mean=0.835, max=0.835, sum=0.835 (1)

min=0.792, mean=0.792, max=0.792, sum=0.792 (1)

min=0.969, mean=0.969, max=0.969, sum=0.969 (1)

min=0.767, mean=0.767, max=0.767, sum=0.767 (1)

min=0.715, mean=0.715, max=0.715, sum=0.715 (1)

min=5999.966, mean=5999.966, max=5999.966, sum=5999.966 (1)

min=0.762, mean=0.762, max=0.762, sum=0.762 (1)

min=0.803, mean=0.803, max=0.803, sum=0.803 (1)

min=0.709, mean=0.709, max=0.709, sum=0.709 (1)
☠ Anthropic is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2112.00861.

min=9.758, mean=9.758, max=9.758, sum=9.758 (1)
☠ Anthropic is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2112.00861.

min=28, mean=28, max=28, sum=28 (1)
☠ Anthropic is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2112.00861.

min=8191.98, mean=8191.98, max=8191.98, sum=8191.98 (1)
☠ Anthropic is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2112.00861.

min=128.85, mean=128.85, max=128.85, sum=128.85 (1)
☠ Anthropic is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://arxiv.org/abs/2112.00861.

min=0.691, mean=0.691, max=0.691, sum=0.691 (1)
☠ BLOOM is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://huggingface.co/spaces/bigscience/BigScienceCorpus.

min=1.172, mean=1.172, max=1.172, sum=1.172 (1)
☠ BLOOM is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://huggingface.co/spaces/bigscience/BigScienceCorpus.

min=28, mean=28, max=28, sum=28 (1)
☠ BLOOM is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://huggingface.co/spaces/bigscience/BigScienceCorpus.

min=2048.999, mean=2048.999, max=2048.999, sum=2048.999 (1)
☠ BLOOM is explicitly trained on the Pile, i.e. data from the same distribution as the test set. See https://huggingface.co/spaces/bigscience/BigScienceCorpus.

min=0.769, mean=0.769, max=0.769, sum=0.769 (1)

min=0.944, mean=0.944, max=0.944, sum=0.944 (1)