NLP Lineage

21 February 2024 · Open weights

Gemma

Gemma: Open Models Based on Gemini Research and Technology

Gemma Team

Date
21 February 2024
What the date marks
Model release
Kind
Model
Organisation
Google DeepMind
Access
Gemma Terms of Use
Other dates
arXiv 2403.08295
Threads
In the guided tour
Chapter 8. The weights get out
Learning phases
6. Local models and deployment

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

Google published research and shipped products, but released no competitive open weights.

The problem

The open ecosystem was consolidating around Meta and Mistral.

What it introduced

Small open models built from the Gemini work, under Google's own use-restricted terms.

Why it matters

It brought the third major lab into open weights and intensified the small-model competition.

What modern models inherit

Frontier labs releasing small open models alongside closed flagships.

What came next

Gemma 2 and 3, and continued pressure at the small end.

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

Gemma has 53 documented ancestors across 11 chapters, reaching back to Recurrent networks in 1990. Showing its direct parents and the 22 turning points among them.

Gemma has 53 documented ancestors across 11 chapters, reaching back to Recurrent networks in 1990. Showing all of them.

Gemma has 54 recorded ancestors, including indirect ones, across 11 chapters, reaching back to Recurrent networks in 1990. Showing its direct parents and the 22 turning points among them.

Gemma has 54 recorded ancestors, including indirect ones, across 11 chapters, reaching back to Recurrent networks in 1990. Showing all of them.

  1. 2024GemmaThis milestone.
  2. 2023Gemini 1.0Gemma built directly on thisChapter 11. Seeing and hearing
  3. 2023Mistral 7BGemma was influenced by thisChapter 8. The weights get out
  4. 2023Llama 2 Turning pointGemma was influenced by thisChapter 8. The weights get out
  5. 2023Grouped-query attention Turning pointLlama 2 built directly on this
    Mistral 7B built directly on this    Chapter 9. Making it run anywhere
  6. 2023LLaMA Turning pointLlama 2 built directly on thisChapter 8. The weights get out
  7. 2022FlashAttentionMistral 7B was influenced by thisChapter 9. Making it run anywhere
  8. 2022PaLMGemini 1.0 succeeded thisChapter 5. Scale becomes a science
  9. 2022Chinchilla Turning pointLLaMA built directly on thisChapter 5. Scale becomes a science
  10. 2022InstructGPT Turning pointLlama 2 was influenced by thisChapter 6. Teaching models what people want
  11. 2022Chain of thought Turning pointPaLM was influenced by thisChapter 12. Thinking longer
  12. 2021GopherChinchilla built directly on thisChapter 5. Scale becomes a science
  13. 2021FLANInstructGPT was influenced by thisChapter 6. Teaching models what people want
  14. 2021RoPE Turning pointLLaMA built directly on thisChapter 10. Reading more than you remember
  15. 2020Vision TransformerGemini 1.0 was influenced by thisChapter 11. Seeing and hearing
  16. 2020RLHF for summarisation Turning pointInstructGPT built directly on thisChapter 6. Teaching models what people want
  17. 2020BigBirdFlashAttention succeeded thisChapter 10. Reading more than you remember
  18. 2020GShard Turning pointGemini 1.0 was influenced by thisChapter 5. Scale becomes a science
  19. 2020GPT-3 Turning pointInstructGPT built directly on thisChapter 5. Scale becomes a science
  20. 2020LongformerMistral 7B was influenced by thisChapter 10. Reading more than you remember
  21. 2020Scaling laws Turning pointGPT-3 was influenced by this
    Chinchilla built directly on this    Chapter 5. Scale becomes a science
  22. 2019Multi-query attentionGrouped-query attention built directly on thisChapter 9. Making it run anywhere
  23. 2019T5 Turning pointFLAN was influenced by thisChapter 4. Pretrain once, use everywhere
  24. 2019C4 Turning pointT5 was influenced by thisChapter 4. Pretrain once, use everywhere
  25. 2019RLHF for LMsRLHF for summarisation built directly on thisChapter 6. Teaching models what people want
  26. 2019Sparse Transformer Turning pointMistral 7B was influenced by thisChapter 4. Pretrain once, use everywhere
  27. 2019GPT-2 Turning pointGPT-3 built directly on thisChapter 4. Pretrain once, use everywhere
  28. 2019WebTextGPT-2 was influenced by thisChapter 4. Pretrain once, use everywhere
  29. 2019Transformer-XLRoPE was influenced by thisChapter 4. Pretrain once, use everywhere
  30. 2018BERT Turning pointVision Transformer was influenced by thisChapter 4. Pretrain once, use everywhere
  31. 2018SentencePieceT5 built directly on thisChapter 4. Pretrain once, use everywhere
  32. 2018GPT-1BERT was influenced by thisChapter 4. Pretrain once, use everywhere
  33. 2018ELMoBERT was influenced by thisChapter 4. Pretrain once, use everywhere
  34. 2018ULMFiTBERT was influenced by thisChapter 4. Pretrain once, use everywhere
  35. 2017CoVeELMo was influenced by thisChapter 3. Attention, then only attention
  36. 2017Transformer Turning pointSparse Transformer built directly on this
    Longformer built directly on this
    Vision Transformer built directly on this    Chapter 3. Attention, then only attention
  37. 2017RLHF originRLHF for summarisation built directly on thisChapter 6. Teaching models what people want
  38. 2017Sparse MoE Turning pointGShard built directly on thisChapter 3. Attention, then only attention
  39. 2016Layer normTransformer built directly on thisChapter 3. Attention, then only attention
  40. 2015Residual connectionsTransformer was influenced by thisChapter 3. Attention, then only attention
  41. 2015LM pretraining Turning pointBERT was influenced by thisChapter 2. Words become vectors
  42. 2015BPE tokenizationLLaMA built directly on thisChapter 2. Words become vectors
  43. 2015Luong attentionTransformer was influenced by thisChapter 3. Attention, then only attention
  44. 2014AdamTransformer was influenced by thisChapter 2. Words become vectors
  45. 2014GloVeELMo built directly on thisChapter 2. Words become vectors
  46. 2014Seq2seq Turning pointTransformer built directly on thisChapter 2. Words become vectors
  47. 2014Attention Turning pointTransformer built directly on thisChapter 3. Attention, then only attention
  48. 2014GRU / encoder-decoderAttention built directly on thisChapter 2. Words become vectors
  49. 2013word2vecELMo was influenced by thisChapter 2. Words become vectors
  50. 2011NLP from scratchELMo was influenced by thisChapter 1. Language as probability
  51. 2003Neural LMNLP from scratch was influenced by this
    word2vec was influenced by this    Chapter 1. Language as probability
  52. 1997LSTM Turning pointSparse MoE built directly on thisChapter 1. Language as probability
  53. 1991Mixture of expertsSparse MoE built directly on thisChapter 1. Language as probability
  54. 1990Recurrent networksLSTM was influenced by thisChapter 1. Language as probability
  1. 2024GemmaThis milestone.
  2. 2023Gemini 1.0Gemma built directly on thisChapter 11. Seeing and hearing
  3. 2023Mistral 7BGemma was influenced by thisChapter 8. The weights get out
  4. 2023Llama 2 Turning pointGemma was influenced by thisChapter 8. The weights get out
  5. 2023Grouped-query attention Turning pointLlama 2 built directly on this
    Mistral 7B built directly on this    Chapter 9. Making it run anywhere
  6. 2023LLaMA Turning pointLlama 2 built directly on thisChapter 8. The weights get out
  7. 2022FlashAttentionMistral 7B was influenced by thisChapter 9. Making it run anywhere
  8. 2022PaLMGemini 1.0 succeeded thisChapter 5. Scale becomes a science
  9. 2022Chinchilla Turning pointLLaMA built directly on thisChapter 5. Scale becomes a science
  10. 2022InstructGPT Turning pointLlama 2 was influenced by thisChapter 6. Teaching models what people want
  11. 2022Chain of thought Turning pointPaLM was influenced by thisChapter 12. Thinking longer
  12. 2021GopherChinchilla built directly on thisChapter 5. Scale becomes a science
  13. 2021FLANInstructGPT was influenced by thisChapter 6. Teaching models what people want
  14. 2021RoPE Turning pointLLaMA built directly on thisChapter 10. Reading more than you remember
  15. 2020Vision TransformerGemini 1.0 was influenced by thisChapter 11. Seeing and hearing
  16. 2020RLHF for summarisation Turning pointInstructGPT built directly on thisChapter 6. Teaching models what people want
  17. 2020BigBirdFlashAttention succeeded thisChapter 10. Reading more than you remember
  18. 2020GShard Turning pointGemini 1.0 was influenced by thisChapter 5. Scale becomes a science
  19. 2020GPT-3 Turning pointInstructGPT built directly on thisChapter 5. Scale becomes a science
  20. 2020LongformerMistral 7B was influenced by thisChapter 10. Reading more than you remember
  21. 2020Scaling laws Turning pointGPT-3 was influenced by this
    Chinchilla built directly on this    Chapter 5. Scale becomes a science
  22. 2019Multi-query attentionGrouped-query attention built directly on thisChapter 9. Making it run anywhere
  23. 2019T5 Turning pointFLAN was influenced by thisChapter 4. Pretrain once, use everywhere
  24. 2019C4 Turning pointT5 was influenced by thisChapter 4. Pretrain once, use everywhere
  25. 2019RLHF for LMsRLHF for summarisation built directly on thisChapter 6. Teaching models what people want
  26. 2019Sparse Transformer Turning pointMistral 7B was influenced by thisChapter 4. Pretrain once, use everywhere
  27. 2019GPT-2 Turning pointGPT-3 built directly on thisChapter 4. Pretrain once, use everywhere
  28. 2019WebTextGPT-2 was influenced by thisChapter 4. Pretrain once, use everywhere
  29. 2019Transformer-XLSparse Transformer was influenced by this (indirect influence)Chapter 4. Pretrain once, use everywhere
  30. 2018BERT Turning pointVision Transformer was influenced by thisChapter 4. Pretrain once, use everywhere
  31. 2018SentencePieceT5 built directly on thisChapter 4. Pretrain once, use everywhere
  32. 2018GPT-1BERT was influenced by thisChapter 4. Pretrain once, use everywhere
  33. 2018ELMoBERT was influenced by thisChapter 4. Pretrain once, use everywhere
  34. 2018ULMFiTBERT was influenced by thisChapter 4. Pretrain once, use everywhere
  35. 2017CoVeELMo was influenced by thisChapter 3. Attention, then only attention
  36. 2017Transformer Turning pointSparse Transformer built directly on this
    Longformer built directly on this
    Vision Transformer built directly on this    Chapter 3. Attention, then only attention
  37. 2017RLHF originRLHF for summarisation built directly on thisChapter 6. Teaching models what people want
  38. 2017ConvS2STransformer was influenced by this (indirect influence)Chapter 3. Attention, then only attention
  39. 2017Sparse MoE Turning pointGShard built directly on thisChapter 3. Attention, then only attention
  40. 2016Layer normTransformer built directly on thisChapter 3. Attention, then only attention
  41. 2015Residual connectionsTransformer was influenced by thisChapter 3. Attention, then only attention
  42. 2015LM pretraining Turning pointBERT was influenced by thisChapter 2. Words become vectors
  43. 2015BPE tokenizationLLaMA built directly on thisChapter 2. Words become vectors
  44. 2015Luong attentionTransformer was influenced by thisChapter 3. Attention, then only attention
  45. 2014AdamTransformer was influenced by thisChapter 2. Words become vectors
  46. 2014GloVeELMo built directly on thisChapter 2. Words become vectors
  47. 2014Seq2seq Turning pointTransformer built directly on thisChapter 2. Words become vectors
  48. 2014Attention Turning pointTransformer built directly on thisChapter 3. Attention, then only attention
  49. 2014GRU / encoder-decoderAttention built directly on thisChapter 2. Words become vectors
  50. 2013word2vecELMo was influenced by thisChapter 2. Words become vectors
  51. 2011NLP from scratchELMo was influenced by thisChapter 1. Language as probability
  52. 2003Neural LMNLP from scratch was influenced by this
    word2vec was influenced by this    Chapter 1. Language as probability
  53. 1997LSTM Turning pointSparse MoE built directly on thisChapter 1. Language as probability
  54. 1991Mixture of expertsSparse MoE built directly on thisChapter 1. Language as probability
  55. 1990Recurrent networksLSTM was influenced by thisChapter 1. Language as probability

After: what it led to

No documented descendants are recorded for Gemma without weak links.

No documented descendants are recorded for Gemma without weak links.

No recorded descendants are recorded for Gemma.

No recorded descendants are recorded for Gemma.

Direct links and their sources

Where it came from

Built directly on

Influenced by

What it led to

None recorded in the sources used.