{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/the-pile-an-800gb-dataset-of-diverse-text-for","title":"The Pile: An 800GB Dataset of Diverse Text for Language Modeling","arxiv_id":"2101.00027","date":"2020-12-31","proceeding":null,"authors":["Leo Gao","Stella Biderman","Sid Black","Laurence Golding","Travis Hoppe","Charles Foster","Jason Phang","Horace He","Anish Thite","Noa Nabeshima","Shawn Presser","Connor Leahy"],"abstract":"Recent work has demonstrated that increased training dataset diversity improves general cross-domain knowledge and downstream generalization capability for large-scale language models. With this in mind, we present \\textit{the Pile}: an 825 GiB English text corpus targeted at training large-scale language models. The Pile is constructed from 22 diverse high-quality subsets -- both existing and newly constructed -- many of which derive from academic or professional sources. Our evaluation of the untuned performance of GPT-2 and GPT-3 on the Pile shows that these models struggle on many of its components, such as academic writing. Conversely, models trained on the Pile improve significantly over both Raw CC and CC-100 on all components of the Pile, while improving performance on downstream evaluations. Through an in-depth exploratory analysis, we document potentially concerning aspects of the data for prospective users. We make publicly available the code used in its construction.","url_abs":"https://arxiv.org/abs/2101.00027v1","url_pdf":"https://arxiv.org/pdf/2101.00027v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"the-pile-an-800gb-dataset-of-diverse-text-for","repo_url":"https://github.com/EleutherAI/The-Pile","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"the-pile-an-800gb-dataset-of-diverse-text-for","repo_url":"https://github.com/EleutherAI/GPTNeo","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"unanswered"}},{"paper_slug":"the-pile-an-800gb-dataset-of-diverse-text-for","repo_url":"https://github.com/EleutherAI/gpt-neo","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"unanswered"}},{"paper_slug":"the-pile-an-800gb-dataset-of-diverse-text-for","repo_url":"https://github.com/RossNordby/SoftPromptsForEvaluation","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"the-pile-an-800gb-dataset-of-diverse-text-for","repo_url":"https://github.com/THUDM/GLM","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"the-pile-an-800gb-dataset-of-diverse-text-for","repo_url":"https://github.com/Wikidepia/indonesia_dataset","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"unanswered"}},{"paper_slug":"the-pile-an-800gb-dataset-of-diverse-text-for","repo_url":"https://github.com/ai21labs/lm-evaluation","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"unanswered"}},{"paper_slug":"the-pile-an-800gb-dataset-of-diverse-text-for","repo_url":"https://github.com/alrope123/prompt-waywardness","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"the-pile-an-800gb-dataset-of-diverse-text-for","repo_url":"https://github.com/codedotal/gpt-code-clippy","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"jax","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"the-pile-an-800gb-dataset-of-diverse-text-for","repo_url":"https://github.com/conceptofmind/LaMDA-pytorch","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"the-pile-an-800gb-dataset-of-diverse-text-for","repo_url":"https://github.com/conceptofmind/lamda-rlhf-pytorch","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"the-pile-an-800gb-dataset-of-diverse-text-for","repo_url":"https://github.com/ftramer/lm-extraction-benchmark","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"the-pile-an-800gb-dataset-of-diverse-text-for","repo_url":"https://github.com/glassroom/heinsen_attention","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"the-pile-an-800gb-dataset-of-diverse-text-for","repo_url":"https://github.com/google-research/lm-extraction-benchmark","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"the-pile-an-800gb-dataset-of-diverse-text-for","repo_url":"https://github.com/jackbandy/bookcorpus-datasheet","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"the-pile-an-800gb-dataset-of-diverse-text-for","repo_url":"https://github.com/ncoop57/gpt-code-clippy","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"jax","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"the-pile-an-800gb-dataset-of-diverse-text-for","repo_url":"https://github.com/neutralzz/billa","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"the-pile-an-800gb-dataset-of-diverse-text-for","repo_url":"https://github.com/nlpodyssey/verbaflow","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"unanswered"}},{"paper_slug":"the-pile-an-800gb-dataset-of-diverse-text-for","repo_url":"https://github.com/suu990901/InfoEntropy-Loss","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"jax","reach":{"status":"unanswered"}},{"paper_slug":"the-pile-an-800gb-dataset-of-diverse-text-for","repo_url":"https://github.com/suu990901/LLaMA-InfoEntropy-Loss","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"jax","reach":{"status":"unanswered"}},{"paper_slug":"the-pile-an-800gb-dataset-of-diverse-text-for","repo_url":"https://github.com/thoppe/personal_cv","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"unanswered"}},{"paper_slug":"the-pile-an-800gb-dataset-of-diverse-text-for","repo_url":"https://github.com/yuchuantian/dijiang","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"diversity","task_name":"Diversity"},{"task_slug":"language-modeling","task_name":"Language Modeling"},{"task_slug":"language-modelling","task_name":"Language Modelling"}],"methods":[{"method_slug":"adam","method_name":"Adam"},{"method_slug":"attention","method_name":"Attention"},{"method_slug":"attention-dropout","method_name":"Attention Dropout"},{"method_slug":"bpe","method_name":"BPE"},{"method_slug":"cosine-annealing","method_name":"Cosine Annealing"},{"method_slug":"dense-connections","method_name":"Dense Connections"},{"method_slug":"discriminative-fine-tuning","method_name":"Discriminative Fine-Tuning"},{"method_slug":"dropout","method_name":"Dropout"},{"method_slug":"gpt-2","method_name":"GPT-2"},{"method_slug":"gpt-3","method_name":"GPT-3"},{"method_slug":"layer-normalization","method_name":"Layer Normalization"},{"method_slug":"linear-layer","method_name":"Linear Layer"},{"method_slug":"linear-warmup-with-cosine-annealing","method_name":"Linear Warmup With Cosine Annealing"},{"method_slug":"multi-head-attention","method_name":"Multi-Head Attention"},{"method_slug":"residual-connection","method_name":"Residual Connection"},{"method_slug":"softmax","method_name":"Softmax"},{"method_slug":"weight-decay","method_name":"Weight Decay"}],"datasets_introduced":[{"slug":"the-pile","name":"The Pile","full_name":"The Pile"}],"methods_introduced":[],"results":[{"leaderboard":"/sota/language-modelling-on-the-pile","task":"Language Modelling","dataset":"The Pile","model":"GPT-3 Davinci 175B (pre-trained)","rank_in_archive_order":13,"of":39,"metrics":{"Bits per byte":"0.7177"},"uses_additional_data":false},{"leaderboard":"/sota/language-modelling-on-the-pile","task":"Language Modelling","dataset":"The Pile","model":"GPT-3 Curie 6.7B (pre-trained)","rank_in_archive_order":18,"of":39,"metrics":{"Bits per byte":"0.7980"},"uses_additional_data":false},{"leaderboard":"/sota/language-modelling-on-the-pile","task":"Language Modelling","dataset":"The Pile","model":"GPT-3 Babbage 1.3B (pre-trained)","rank_in_archive_order":22,"of":39,"metrics":{"Bits per byte":"0.8718"},"uses_additional_data":false},{"leaderboard":"/sota/language-modelling-on-the-pile","task":"Language Modelling","dataset":"The Pile","model":"GPT-3 Ada 350M (pre-trained)","rank_in_archive_order":23,"of":39,"metrics":{"Bits per byte":"0.9631"},"uses_additional_data":false},{"leaderboard":"/sota/language-modelling-on-the-pile","task":"Language Modelling","dataset":"The Pile","model":"GPT-2 XL 1.5B (pre-trained)","rank_in_archive_order":24,"of":39,"metrics":{"Bits per byte":"1.0468"},"uses_additional_data":false},{"leaderboard":"/sota/language-modelling-on-the-pile","task":"Language Modelling","dataset":"The Pile","model":"GPT-2 Large 774M (pre-trained)","rank_in_archive_order":25,"of":39,"metrics":{"Bits per byte":"1.0828"},"uses_additional_data":false},{"leaderboard":"/sota/language-modelling-on-the-pile","task":"Language Modelling","dataset":"The Pile","model":"GPT-2 Medium 355M (pre-trained)","rank_in_archive_order":26,"of":39,"metrics":{"Bits per byte":"1.0928"},"uses_additional_data":false},{"leaderboard":"/sota/language-modelling-on-the-pile","task":"Language Modelling","dataset":"The Pile","model":"GPT-2 Small 124M (pre-trained)","rank_in_archive_order":27,"of":39,"metrics":{"Bits per byte":"1.2253"},"uses_additional_data":false}],"syntology":{"syntology_url":"https://syntology.ai/paper/2101.00027","atlas_url":"https://app.syntology.ai/?focus=2101.00027","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2101.00027"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-25T09:33:49+00:00","read_at_is":"when the build read Syntology's graph, not when any sample ran","claim":"Per-sample execution status on synthesized fixtures; not a correctness claim about the paper. Samples come from repositories linked to the paper, official or community; repo_kind says which.","repos":[{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/ai21labs/lm-evaluation","reach":{"status":"unanswered"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/conceptofmind/lamda-rlhf-pytorch","reach":{"status":"ok","spdx":"MIT"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/suu990901/LLaMA-InfoEntropy-Loss","reach":{"status":"unanswered"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/conceptofmind/LaMDA-pytorch","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/EleutherAI/gpt-neo","reach":{"status":"unanswered"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/Wikidepia/indonesia_dataset","reach":{"status":"unanswered"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/ftramer/lm-extraction-benchmark","reach":{"status":"ok","spdx":"Apache-2.0"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/RossNordby/SoftPromptsForEvaluation","reach":{"status":"ok","spdx":"MIT"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/suu990901/InfoEntropy-Loss","reach":{"status":"unanswered"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/alrope123/prompt-waywardness","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/google-research/lm-extraction-benchmark","reach":{"status":"ok","spdx":"Apache-2.0"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/EleutherAI/GPTNeo","reach":{"status":"unanswered"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/thoppe/personal_cv","reach":{"status":"unanswered"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/ncoop57/gpt-code-clippy","reach":{"status":"ok","spdx":"Apache-2.0"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/neutralzz/billa","reach":{"status":"unanswered"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/THUDM/GLM","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/EleutherAI/The-Pile","reach":{"status":"ok","spdx":"MIT"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/jackbandy/bookcorpus-datasheet","reach":{"status":"ok","spdx":"MIT"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/codedotal/gpt-code-clippy","reach":{"status":"ok","spdx":"Apache-2.0"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/yuchuantian/dijiang","reach":{"status":"ok"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/nlpodyssey/verbaflow","reach":{"status":"unanswered"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/glassroom/heinsen_attention","reach":{"status":"unanswered"}}],"summary":{"ran_honours":1},"by_repo_kind":{"official":{"samples":1,"ran":1,"repositories":1}},"repo_kind_vocabulary":{"official":"The archive marks this repository official for the paper","named_in_paper":"The archive records that the paper mentions this repository; it is not marked official","listed":"In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper","found_in_text":"Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted","community":"Not in the archive's code links for this paper; a community repository Syntology harvested"},"n_pointer_only_for_licence":0,"samples":[{"code_sha256_prefix":"7ed0726edbb13bb5","entry":"take","repo":"EleutherAI/The-Pile","repo_kind":"official","path":"the_pile/pile.py","file_url":"https://github.com/EleutherAI/The-Pile/blob/HEAD/the_pile/pile.py","link_basis":"first_harvest_node","language":"python","status":"ran_honours","verification_level":1,"contract_check":"HONOURS","metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"7ed0726edbb13bb5"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}