NLP Lineage

1 February 2024 · Language modelling

OLMo

OLMo: Accelerating the Science of Language Models

D. Groeneveld, I. Beltagy, P. Walsh, et al. (43 authors)

Date
1 February 2024
What the date marks
arXiv submission
Kind
Model
Organisation
Allen Institute for AI
Access
Apache 2.0
Other dates
ACL 2024
Threads
In the guided tour
Chapter 8. The weights get out
Learning phases
8. Training from scratch

Notable: no later milestone in the data is documented as descending from it. 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-weight models shipped weights but withheld data, code and logs.

The problem

Weights alone do not let you reproduce or diagnose how a model came to be.

What it introduced

Weights, data, training code, evaluation harness and intermediate checkpoints, all released.

Why it matters

The clearest example of genuinely open-source rather than merely open-weight, and the reference for that distinction.

What modern models inherit

The vocabulary distinction between open weights and open source.

What came next

OLMo 2 and 3, and continued pressure on what open should mean.

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

OLMo has 46 documented ancestors across 9 chapters, reaching back to Recurrent networks in 1990. Showing its direct parents and the 19 turning points among them.

OLMo has 46 documented ancestors across 9 chapters, reaching back to Recurrent networks in 1990. Showing all of them.

OLMo has 47 recorded ancestors, including indirect ones, across 9 chapters, reaching back to Recurrent networks in 1990. Showing its direct parents and the 19 turning points among them.

OLMo has 47 recorded ancestors, including indirect ones, across 9 chapters, reaching back to Recurrent networks in 1990. Showing all of them.

  1. 2024OLMoThis milestone.
  2. 2024DolmaOLMo built directly on thisChapter 8. The weights get out
  3. 2023Llama 2 Turning pointOLMo was influenced by thisChapter 8. The weights get out
  4. 2023RefinedWebDolma was influenced by thisChapter 8. The weights get out
  5. 2023Grouped-query attention Turning pointLlama 2 built directly on thisChapter 9. Making it run anywhere
  6. 2023PythiaOLMo was influenced by thisChapter 8. The weights get out
  7. 2023LLaMA Turning pointLlama 2 built directly on thisChapter 8. The weights get out
  8. 2022Chinchilla Turning pointLLaMA built directly on thisChapter 5. Scale becomes a science
  9. 2022InstructGPT Turning pointLlama 2 was influenced by thisChapter 6. Teaching models what people want
  10. 2021GopherChinchilla built directly on thisChapter 5. Scale becomes a science
  11. 2021FLANInstructGPT was influenced by thisChapter 6. Teaching models what people want
  12. 2021RoPE Turning pointLLaMA built directly on thisChapter 10. Reading more than you remember
  13. 2020The PilePythia built directly on this
    Dolma was influenced by this    Chapter 5. Scale becomes a science
  14. 2020RLHF for summarisation Turning pointInstructGPT built directly on thisChapter 6. Teaching models what people want
  15. 2020GPT-3 Turning pointInstructGPT built directly on this
    The Pile was influenced by this    Chapter 5. Scale becomes a science
  16. 2020Scaling laws Turning pointPythia was influenced by thisChapter 5. Scale becomes a science
  17. 2019Multi-query attentionGrouped-query attention built directly on thisChapter 9. Making it run anywhere
  18. 2019C4 Turning pointDolma was influenced by thisChapter 4. Pretrain once, use everywhere
  19. 2019T5 Turning pointFLAN was influenced by thisChapter 4. Pretrain once, use everywhere
  20. 2019RLHF for LMsRLHF for summarisation built directly on thisChapter 6. Teaching models what people want
  21. 2019Sparse Transformer Turning pointGPT-3 built directly on thisChapter 4. Pretrain once, use everywhere
  22. 2019GPT-2 Turning pointScaling laws was influenced by thisChapter 4. Pretrain once, use everywhere
  23. 2019WebTextGPT-2 was influenced by thisChapter 4. Pretrain once, use everywhere
  24. 2019Transformer-XLRoPE was influenced by thisChapter 4. Pretrain once, use everywhere
  25. 2018BERT Turning pointT5 was influenced by thisChapter 4. Pretrain once, use everywhere
  26. 2018SentencePieceT5 built directly on thisChapter 4. Pretrain once, use everywhere
  27. 2018GPT-1GPT-2 built directly on thisChapter 4. Pretrain once, use everywhere
  28. 2018ELMoBERT was influenced by thisChapter 4. Pretrain once, use everywhere
  29. 2018ULMFiTBERT was influenced by thisChapter 4. Pretrain once, use everywhere
  30. 2017CoVeELMo was influenced by thisChapter 3. Attention, then only attention
  31. 2017Transformer Turning pointScaling laws built directly on thisChapter 3. Attention, then only attention
  32. 2017RLHF originRLHF for summarisation built directly on thisChapter 6. Teaching models what people want
  33. 2016Layer normTransformer built directly on this
    GPT-2 built directly on this    Chapter 3. Attention, then only attention
  34. 2015Residual connectionsTransformer was influenced by thisChapter 3. Attention, then only attention
  35. 2015LM pretraining Turning pointGPT-1 was influenced by thisChapter 2. Words become vectors
  36. 2015BPE tokenizationLLaMA built directly on thisChapter 2. Words become vectors
  37. 2015Luong attentionTransformer was influenced by thisChapter 3. Attention, then only attention
  38. 2014AdamTransformer was influenced by thisChapter 2. Words become vectors
  39. 2014GloVeELMo built directly on thisChapter 2. Words become vectors
  40. 2014Seq2seq Turning pointTransformer built directly on thisChapter 2. Words become vectors
  41. 2014Attention Turning pointTransformer built directly on thisChapter 3. Attention, then only attention
  42. 2014GRU / encoder-decoderAttention built directly on thisChapter 2. Words become vectors
  43. 2013word2vecELMo was influenced by thisChapter 2. Words become vectors
  44. 2011NLP from scratchELMo was influenced by thisChapter 1. Language as probability
  45. 2003Neural LMNLP from scratch was influenced by this
    word2vec was influenced by this    Chapter 1. Language as probability
  46. 1997LSTM Turning pointSeq2seq built directly on thisChapter 1. Language as probability
  47. 1990Recurrent networksLSTM was influenced by thisChapter 1. Language as probability
  1. 2024OLMoThis milestone.
  2. 2024DolmaOLMo built directly on thisChapter 8. The weights get out
  3. 2023Llama 2 Turning pointOLMo was influenced by thisChapter 8. The weights get out
  4. 2023RefinedWebDolma was influenced by thisChapter 8. The weights get out
  5. 2023Grouped-query attention Turning pointLlama 2 built directly on thisChapter 9. Making it run anywhere
  6. 2023PythiaOLMo was influenced by thisChapter 8. The weights get out
  7. 2023LLaMA Turning pointLlama 2 built directly on thisChapter 8. The weights get out
  8. 2022Chinchilla Turning pointLLaMA built directly on thisChapter 5. Scale becomes a science
  9. 2022InstructGPT Turning pointLlama 2 was influenced by thisChapter 6. Teaching models what people want
  10. 2021GopherChinchilla built directly on thisChapter 5. Scale becomes a science
  11. 2021FLANInstructGPT was influenced by thisChapter 6. Teaching models what people want
  12. 2021RoPE Turning pointLLaMA built directly on thisChapter 10. Reading more than you remember
  13. 2020The PilePythia built directly on this
    Dolma was influenced by this    Chapter 5. Scale becomes a science
  14. 2020RLHF for summarisation Turning pointInstructGPT built directly on thisChapter 6. Teaching models what people want
  15. 2020GPT-3 Turning pointInstructGPT built directly on this
    The Pile was influenced by this    Chapter 5. Scale becomes a science
  16. 2020Scaling laws Turning pointPythia was influenced by thisChapter 5. Scale becomes a science
  17. 2019Multi-query attentionGrouped-query attention built directly on thisChapter 9. Making it run anywhere
  18. 2019C4 Turning pointDolma was influenced by thisChapter 4. Pretrain once, use everywhere
  19. 2019T5 Turning pointFLAN was influenced by thisChapter 4. Pretrain once, use everywhere
  20. 2019RLHF for LMsRLHF for summarisation built directly on thisChapter 6. Teaching models what people want
  21. 2019Sparse Transformer Turning pointGPT-3 built directly on thisChapter 4. Pretrain once, use everywhere
  22. 2019WebTextC4 was influenced by this (indirect influence)Chapter 4. Pretrain once, use everywhere
  23. 2019GPT-2 Turning pointScaling laws was influenced by thisChapter 4. Pretrain once, use everywhere
  24. 2019Transformer-XLRoPE was influenced by thisChapter 4. Pretrain once, use everywhere
  25. 2018BERT Turning pointT5 was influenced by thisChapter 4. Pretrain once, use everywhere
  26. 2018SentencePieceT5 built directly on thisChapter 4. Pretrain once, use everywhere
  27. 2018GPT-1GPT-2 built directly on this
    WebText was influenced by this    Chapter 4. Pretrain once, use everywhere
  28. 2018ELMoGPT-1 was influenced by this (indirect influence)Chapter 4. Pretrain once, use everywhere
  29. 2018ULMFiTGPT-1 was influenced by this (indirect influence)Chapter 4. Pretrain once, use everywhere
  30. 2017CoVeELMo was influenced by thisChapter 3. Attention, then only attention
  31. 2017Transformer Turning pointScaling laws built directly on thisChapter 3. Attention, then only attention
  32. 2017RLHF originRLHF for summarisation built directly on thisChapter 6. Teaching models what people want
  33. 2017ConvS2STransformer was influenced by this (indirect influence)Chapter 3. Attention, then only attention
  34. 2016Layer normTransformer built directly on this
    GPT-2 built directly on this    Chapter 3. Attention, then only attention
  35. 2015Residual connectionsTransformer was influenced by thisChapter 3. Attention, then only attention
  36. 2015LM pretraining Turning pointGPT-1 was influenced by thisChapter 2. Words become vectors
  37. 2015BPE tokenizationLLaMA built directly on thisChapter 2. Words become vectors
  38. 2015Luong attentionTransformer was influenced by thisChapter 3. Attention, then only attention
  39. 2014AdamTransformer was influenced by thisChapter 2. Words become vectors
  40. 2014GloVeELMo built directly on thisChapter 2. Words become vectors
  41. 2014Seq2seq Turning pointTransformer built directly on thisChapter 2. Words become vectors
  42. 2014Attention Turning pointTransformer built directly on thisChapter 3. Attention, then only attention
  43. 2014GRU / encoder-decoderAttention built directly on thisChapter 2. Words become vectors
  44. 2013word2vecELMo was influenced by thisChapter 2. Words become vectors
  45. 2011NLP from scratchELMo was influenced by thisChapter 1. Language as probability
  46. 2003Neural LMNLP from scratch was influenced by this
    word2vec was influenced by this    Chapter 1. Language as probability
  47. 1997LSTM Turning pointSeq2seq built directly on thisChapter 1. Language as probability
  48. 1990Recurrent networksLSTM was influenced by thisChapter 1. Language as probability

After: what it led to

No documented descendants are recorded for OLMo without weak links.

No documented descendants are recorded for OLMo without weak links.

No recorded descendants are recorded for OLMo.

No recorded descendants are recorded for OLMo.

Direct links and their sources

Where it came from

Built directly on

Influenced by

What it led to

None recorded in the sources used.