Collections:
  - Name: BLIP
    Metadata:
      Training Data:
        - COCO
        - VG
        - Conceptual Captions
        - Conceptual 12M
        - SBU captions
      Architecture:
        - Transformer
      Training Resources: 8x A100 GPUs
    Paper:
      Title: 'BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language
        Understanding and Generation'
      URL: https://arxiv.org/abs/2201.12086
    README: configs/blip/README.md

Models:
  - Name: blip-base_8xb16_refcoco
    Metadata:
      FLOPs: null
      Parameters: 498488636
    In Collection: BLIP
    Results:
      - Task: Visual Grounding
        Dataset: RefCOCO
        Metrics:
          Accuracy (testA): 86.14
          Accuracy (testB): 77.33
    Weights: https://download.openmmlab.com/mmclassification/v1/blip/blip-base_8xb16_refcoco_20230508-d2d10f4c.pth
    Config: configs/blip/blip-base_8xb16_refcoco.py
  - Name: blip-base_3rdparty_caption
    Metadata:
      FLOPs: null
      Parameters: 223971644
    In Collection: BLIP
    Results:
      - Dataset: COCO
        Task: Image Caption
        Metrics:
          BLEU-4: 40.12
          CIDER: 132.82
    Weights: https://download.openmmlab.com/mmclassification/v1/blip/blip-base_3rdparty_coco-caption_20230419-a5b71af3.pth
    Config: configs/blip/blip-base_8xb32_caption.py
    Converted From:
      Weights: https://storage.googleapis.com/sfr-vision-language-research/LAVIS/models/BLIP/blip_coco_caption_base.pth
      Code: https://github.com/salesforce/LAVIS
  - Name: blip-base_3rdparty_nlvr
    Metadata:
      FLOPs: null
      Parameters: 259372034
    In Collection: BLIP
    Results:
      - Task: NLVR
        Dataset: NLVR2
        Metrics:
          Top 1 Accuracy: 82.33
    Weights: https://download.openmmlab.com/mmclassification/v1/blip/blip-base_3rdparty_nlvr_20230427-3b14d33f.pth
    Config: configs/blip/blip-base_8xb32_nlvr.py
    Converted From:
      Weights: https://storage.googleapis.com/sfr-vision-language-research/BLIP/models/model_base_nlvr.pth
      Code: https://github.com/salesforce/LAVIS
  - Name: blip-base_3rdparty_vqa
    Metadata:
      FLOPs: null
      Parameters: 361478972
    In Collection: BLIP
    Results:
      - Task: Visual Question Answering
        Dataset: VQAv2
        Metrics:
          Accuracy: 78.2
    Weights: https://download.openmmlab.com/mmclassification/v1/blip/blip-base_3rdparty-capflit_vqa_20230505-81488941.pth
    Config: configs/blip/blip-base_8xb32_vqa.py
    Converted From:
      Weights: https://storage.googleapis.com/sfr-vision-language-research/BLIP/models/model_base_vqa_capfilt_large.pth
      Code: https://github.com/salesforce/LAVIS
  - Name: blip-base_3rdparty_retrieval
    Metadata:
      FLOPs: null
      Parameters: 447486979
    In Collection: BLIP
    Results:
      - Task: Image-To-Text Retrieval
        Dataset: COCO
        Metrics:
          Recall@1: 82.52
          Recall@5: 95.34
      - Task: Text-To-Image Retrieval
        Dataset: COCO
        Metrics:
          Recall@1: 64.82
          Recall@5: 86.28
    Weights: https://download.openmmlab.com/mmclassification/v1/blip/blip-base_3rdparty_coco-retrieval_20230419-a1804d2c.pth
    Config: configs/blip/blip-base_8xb32_retrieval.py
    Converted From:
      Weights: https://storage.googleapis.com/sfr-vision-language-research/LAVIS/models/BLIP/blip_coco_retrieval.pth
      Code: https://github.com/salesforce/LAVIS