NLP Lineage

29 April 2022 · Multimodality

Flamingo

Flamingo: a Visual Language Model for Few-Shot Learning

J. Alayrac, J. Donahue, P. Luc, et al. (27 authors)

Date
29 April 2022
What the date marks
arXiv submission
Kind
Model
Organisation
DeepMind
Access
Not applicable
Other dates
NeurIPS 2022
Threads
In the guided tour
Chapter 11. Seeing and hearing
Learning phases
4. Embeddings and RAG

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

Vision-language models were trained end to end for specific tasks like captioning.

The problem

Retraining a language model for vision throws away its language ability and costs a fortune.

What it introduced

Freeze a pretrained vision encoder and a pretrained language model, and train only cross-attention layers to bridge them.

Why it matters

The bridging recipe that made multimodality affordable, and the template LLaVA and others followed.

What modern models inherit

Bolting vision onto a language model by training an adapter between frozen parts.

What came next

LLaVA did it cheaply and openly; GPT-4V and Gemini made it a product.

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

Flamingo has 28 documented ancestors across 6 chapters, reaching back to Recurrent networks in 1990. Showing its direct parents and the 10 turning points among them.

Flamingo has 28 documented ancestors across 6 chapters, reaching back to Recurrent networks in 1990. Showing all of them.

Flamingo has 30 recorded ancestors, including indirect ones, across 6 chapters, reaching back to Recurrent networks in 1990. Showing its direct parents and the 10 turning points among them.

Flamingo has 30 recorded ancestors, including indirect ones, across 6 chapters, reaching back to Recurrent networks in 1990. Showing all of them.

  1. 2022FlamingoThis milestone.
  2. 2021CLIPFlamingo built directly on thisChapter 11. Seeing and hearing
  3. 2020Vision TransformerFlamingo built directly on thisChapter 11. Seeing and hearing
  4. 2020GPT-3 Turning pointFlamingo was influenced by thisChapter 5. Scale becomes a science
  5. 2020Scaling laws Turning pointGPT-3 was influenced by thisChapter 5. Scale becomes a science
  6. 2019Sparse Transformer Turning pointGPT-3 built directly on thisChapter 4. Pretrain once, use everywhere
  7. 2019GPT-2 Turning pointGPT-3 built directly on this
    CLIP built directly on this    Chapter 4. Pretrain once, use everywhere
  8. 2019WebTextGPT-2 was influenced by thisChapter 4. Pretrain once, use everywhere
  9. 2018BERT Turning pointVision Transformer was influenced by thisChapter 4. Pretrain once, use everywhere
  10. 2018GPT-1BERT was influenced by this
    GPT-2 built directly on this    Chapter 4. Pretrain once, use everywhere
  11. 2018ELMoBERT was influenced by thisChapter 4. Pretrain once, use everywhere
  12. 2018ULMFiTBERT was influenced by thisChapter 4. Pretrain once, use everywhere
  13. 2017CoVeELMo was influenced by thisChapter 3. Attention, then only attention
  14. 2017Transformer Turning pointVision Transformer built directly on thisChapter 3. Attention, then only attention
  15. 2016Layer normTransformer built directly on this
    GPT-2 built directly on this    Chapter 3. Attention, then only attention
  16. 2015Residual connectionsTransformer was influenced by thisChapter 3. Attention, then only attention
  17. 2015LM pretraining Turning pointBERT was influenced by thisChapter 2. Words become vectors
  18. 2015BPE tokenizationGPT-2 built directly on thisChapter 2. Words become vectors
  19. 2015Luong attentionTransformer was influenced by thisChapter 3. Attention, then only attention
  20. 2014AdamTransformer was influenced by thisChapter 2. Words become vectors
  21. 2014GloVeELMo built directly on thisChapter 2. Words become vectors
  22. 2014Seq2seq Turning pointTransformer built directly on thisChapter 2. Words become vectors
  23. 2014Attention Turning pointTransformer built directly on thisChapter 3. Attention, then only attention
  24. 2014GRU / encoder-decoderAttention built directly on thisChapter 2. Words become vectors
  25. 2013word2vecELMo was influenced by thisChapter 2. Words become vectors
  26. 2011NLP from scratchELMo was influenced by thisChapter 1. Language as probability
  27. 2003Neural LMNLP from scratch was influenced by this
    word2vec was influenced by this    Chapter 1. Language as probability
  28. 1997LSTM Turning pointSeq2seq built directly on this
    ELMo built directly on this    Chapter 1. Language as probability
  29. 1990Recurrent networksLSTM was influenced by thisChapter 1. Language as probability
  1. 2022FlamingoThis milestone.
  2. 2021CLIPFlamingo built directly on thisChapter 11. Seeing and hearing
  3. 2020Vision TransformerFlamingo built directly on thisChapter 11. Seeing and hearing
  4. 2020GPT-3 Turning pointFlamingo was influenced by thisChapter 5. Scale becomes a science
  5. 2020Scaling laws Turning pointGPT-3 was influenced by thisChapter 5. Scale becomes a science
  6. 2019Sparse Transformer Turning pointGPT-3 built directly on thisChapter 4. Pretrain once, use everywhere
  7. 2019GPT-2 Turning pointGPT-3 built directly on this
    CLIP built directly on this    Chapter 4. Pretrain once, use everywhere
  8. 2019WebTextGPT-2 was influenced by thisChapter 4. Pretrain once, use everywhere
  9. 2019Transformer-XLSparse Transformer was influenced by this (indirect influence)Chapter 4. Pretrain once, use everywhere
  10. 2018BERT Turning pointVision Transformer was influenced by thisChapter 4. Pretrain once, use everywhere
  11. 2018GPT-1BERT was influenced by this
    GPT-2 built directly on this    Chapter 4. Pretrain once, use everywhere
  12. 2018ELMoBERT was influenced by thisChapter 4. Pretrain once, use everywhere
  13. 2018ULMFiTBERT was influenced by thisChapter 4. Pretrain once, use everywhere
  14. 2017CoVeELMo was influenced by thisChapter 3. Attention, then only attention
  15. 2017Transformer Turning pointVision Transformer built directly on thisChapter 3. Attention, then only attention
  16. 2017ConvS2STransformer was influenced by this (indirect influence)Chapter 3. Attention, then only attention
  17. 2016Layer normTransformer built directly on this
    GPT-2 built directly on this    Chapter 3. Attention, then only attention
  18. 2015Residual connectionsTransformer was influenced by thisChapter 3. Attention, then only attention
  19. 2015LM pretraining Turning pointBERT was influenced by thisChapter 2. Words become vectors
  20. 2015BPE tokenizationGPT-2 built directly on thisChapter 2. Words become vectors
  21. 2015Luong attentionTransformer was influenced by thisChapter 3. Attention, then only attention
  22. 2014AdamTransformer was influenced by thisChapter 2. Words become vectors
  23. 2014GloVeELMo built directly on thisChapter 2. Words become vectors
  24. 2014Seq2seq Turning pointTransformer built directly on thisChapter 2. Words become vectors
  25. 2014Attention Turning pointTransformer built directly on thisChapter 3. Attention, then only attention
  26. 2014GRU / encoder-decoderAttention built directly on thisChapter 2. Words become vectors
  27. 2013word2vecELMo was influenced by thisChapter 2. Words become vectors
  28. 2011NLP from scratchELMo was influenced by thisChapter 1. Language as probability
  29. 2003Neural LMNLP from scratch was influenced by this
    word2vec was influenced by this    Chapter 1. Language as probability
  30. 1997LSTM Turning pointSeq2seq built directly on this
    ELMo built directly on this    Chapter 1. Language as probability
  31. 1990Recurrent networksLSTM was influenced by thisChapter 1. Language as probability

After: what it led to

No documented descendants are recorded for Flamingo without weak links.

No documented descendants are recorded for Flamingo without weak links.

Flamingo has 1 recorded descendants, including indirect ones, across 1 chapter, reaching forward to LLaVA in 2023. Showing all of them.

Flamingo has 1 recorded descendants, including indirect ones, across 1 chapter, reaching forward to LLaVA in 2023. Showing all of them.

  1. 2022FlamingoThis milestone.
  2. 2023LLaVAwas influenced by Flamingo (indirect influence)Chapter 11. Seeing and hearing

Direct links and their sources

Where it came from

Built directly on

Influenced by

What it led to

Influenced

Terms