NLP Lineage

3 April 2023 · Language modelling

Pythia

Pythia: A Suite for Analyzing Large Language Models Across Training and Scaling

S. Biderman, H. Schoelkopf, Q. Anthony, et al. (13 authors)

Date
3 April 2023
What the date marks
arXiv submission
Kind
Model
Organisation
EleutherAI
Access
Apache 2.0
Other dates
ICML 2023
Threads
In the guided tour
Chapter 8. The weights get out
Learning phases
8. Training from scratch

A milestone: 1 later milestone descends directly from it, and its descendants reach 3 of the 10 threads. How levels are decided

Read the primary source (arxiv.org)

Full summary in progress. Until then, this page shows the milestone's record from the research data.

What it changed

Before it

Open models were released as final checkpoints with no training history.

The problem

You cannot study how capabilities emerge if you only see the finished model.

What it introduced

Sixteen models across scales, trained on identical data in identical order, with 154 checkpoints each released.

Why it matters

It turned training dynamics into something outsiders could actually study, including memorisation and bias formation.

What modern models inherit

Open checkpoints and data order as research infrastructure.

What came next

OLMo extended the fully-open approach to frontier-adjacent scale.

Before and after

The milestones it descends from, and the milestones that descend from it. Work developed alongside it is not counted. Weak links are left out unless you include them.

Before: where it came from

Pythia has 20 documented ancestors across 5 chapters, reaching back to Recurrent networks in 1990. Showing its direct parents and the 10 turning points among them.

Pythia has 20 documented ancestors across 5 chapters, reaching back to Recurrent networks in 1990. Showing all of them.

Pythia has 29 recorded ancestors, including indirect ones, across 5 chapters, reaching back to Recurrent networks in 1990. Showing its direct parents and the 10 turning points among them.

Pythia has 29 recorded ancestors, including indirect ones, across 5 chapters, reaching back to Recurrent networks in 1990. Showing all of them.

  1. 2023PythiaThis milestone.
  2. 2020The PilePythia built directly on thisChapter 5. Scale becomes a science
  3. 2020GPT-3 Turning pointThe Pile was influenced by thisChapter 5. Scale becomes a science
  4. 2020Scaling laws Turning pointPythia was influenced by thisChapter 5. Scale becomes a science
  5. 2019C4 Turning pointThe Pile was influenced by thisChapter 4. Pretrain once, use everywhere
  6. 2019Sparse Transformer Turning pointGPT-3 built directly on thisChapter 4. Pretrain once, use everywhere
  7. 2019GPT-2 Turning pointScaling laws was influenced by thisChapter 4. Pretrain once, use everywhere
  8. 2019WebTextGPT-2 was influenced by thisChapter 4. Pretrain once, use everywhere
  9. 2018GPT-1GPT-2 built directly on thisChapter 4. Pretrain once, use everywhere
  10. 2017Transformer Turning pointScaling laws built directly on thisChapter 3. Attention, then only attention
  11. 2016Layer normTransformer built directly on this
    GPT-2 built directly on this    Chapter 3. Attention, then only attention
  12. 2015Residual connectionsTransformer was influenced by thisChapter 3. Attention, then only attention
  13. 2015LM pretraining Turning pointGPT-1 was influenced by thisChapter 2. Words become vectors
  14. 2015BPE tokenizationGPT-2 built directly on thisChapter 2. Words become vectors
  15. 2015Luong attentionTransformer was influenced by thisChapter 3. Attention, then only attention
  16. 2014AdamTransformer was influenced by thisChapter 2. Words become vectors
  17. 2014Seq2seq Turning pointTransformer built directly on thisChapter 2. Words become vectors
  18. 2014Attention Turning pointTransformer built directly on thisChapter 3. Attention, then only attention
  19. 2014GRU / encoder-decoderAttention built directly on thisChapter 2. Words become vectors
  20. 1997LSTM Turning pointSeq2seq built directly on thisChapter 1. Language as probability
  21. 1990Recurrent networksLSTM was influenced by thisChapter 1. Language as probability
  1. 2023PythiaThis milestone.
  2. 2020The PilePythia built directly on thisChapter 5. Scale becomes a science
  3. 2020GPT-3 Turning pointThe Pile was influenced by thisChapter 5. Scale becomes a science
  4. 2020Scaling laws Turning pointPythia was influenced by thisChapter 5. Scale becomes a science
  5. 2019C4 Turning pointThe Pile was influenced by thisChapter 4. Pretrain once, use everywhere
  6. 2019Sparse Transformer Turning pointGPT-3 built directly on thisChapter 4. Pretrain once, use everywhere
  7. 2019GPT-2 Turning pointScaling laws was influenced by thisChapter 4. Pretrain once, use everywhere
  8. 2019WebTextGPT-2 was influenced by this
    C4 was influenced by this (indirect influence)    Chapter 4. Pretrain once, use everywhere
  9. 2019Transformer-XLSparse Transformer was influenced by this (indirect influence)Chapter 4. Pretrain once, use everywhere
  10. 2018GPT-1GPT-2 built directly on thisChapter 4. Pretrain once, use everywhere
  11. 2018ELMoGPT-1 was influenced by this (indirect influence)Chapter 4. Pretrain once, use everywhere
  12. 2018ULMFiTGPT-1 was influenced by this (indirect influence)Chapter 4. Pretrain once, use everywhere
  13. 2017CoVeELMo was influenced by thisChapter 3. Attention, then only attention
  14. 2017Transformer Turning pointScaling laws built directly on thisChapter 3. Attention, then only attention
  15. 2017ConvS2STransformer was influenced by this (indirect influence)Chapter 3. Attention, then only attention
  16. 2016Layer normTransformer built directly on this
    GPT-2 built directly on this    Chapter 3. Attention, then only attention
  17. 2015Residual connectionsTransformer was influenced by thisChapter 3. Attention, then only attention
  18. 2015LM pretraining Turning pointGPT-1 was influenced by thisChapter 2. Words become vectors
  19. 2015BPE tokenizationGPT-2 built directly on thisChapter 2. Words become vectors
  20. 2015Luong attentionTransformer was influenced by thisChapter 3. Attention, then only attention
  21. 2014AdamTransformer was influenced by thisChapter 2. Words become vectors
  22. 2014GloVeELMo built directly on thisChapter 2. Words become vectors
  23. 2014Seq2seq Turning pointTransformer built directly on thisChapter 2. Words become vectors
  24. 2014Attention Turning pointTransformer built directly on thisChapter 3. Attention, then only attention
  25. 2014GRU / encoder-decoderAttention built directly on thisChapter 2. Words become vectors
  26. 2013word2vecELMo was influenced by thisChapter 2. Words become vectors
  27. 2011NLP from scratchELMo was influenced by thisChapter 1. Language as probability
  28. 2003Neural LMNLP from scratch was influenced by this
    word2vec was influenced by this    Chapter 1. Language as probability
  29. 1997LSTM Turning pointSeq2seq built directly on thisChapter 1. Language as probability
  30. 1990Recurrent networksLSTM was influenced by thisChapter 1. Language as probability

After: what it led to

Pythia has 1 documented descendants across 1 chapter, reaching forward to OLMo in 2024. Showing all of them.

Pythia has 1 documented descendants across 1 chapter, reaching forward to OLMo in 2024. Showing all of them.

Pythia has 1 recorded descendants, including indirect ones, across 1 chapter, reaching forward to OLMo in 2024. Showing all of them.

Pythia has 1 recorded descendants, including indirect ones, across 1 chapter, reaching forward to OLMo in 2024. Showing all of them.

  1. 2023PythiaThis milestone.
  2. 2024OLMowas influenced by PythiaChapter 8. The weights get out
  1. 2023PythiaThis milestone.
  2. 2024OLMowas influenced by PythiaChapter 8. The weights get out

Direct links and their sources

Where it came from

Built directly on

Influenced by

What it led to

Influenced