Hugging Face's logo Hugging Face
  • Models
  • Datasets
  • Spaces
  • Buckets new
  • Docs
  • Enterprise
  • Pricing
    • Website
      • Tasks
      • HuggingChat
      • Collections
      • Languages
      • Organizations
    • Community
      • Blog
      • Posts
      • Daily Papers
      • Hardware
      • Learn
      • Discord
      • Forum
      • GitHub
    • Solutions
      • Team & Enterprise
      • Hugging Face PRO
      • Enterprise Support
      • Inference Providers
      • Inference Endpoints
      • Storage Buckets

  • Log In
  • Sign Up

AbstractPhil
/
captionbert-8192-v2-b

Feature Extraction
Transformers
TensorBoard
Safetensors
English
captionbert_v2
sentence-similarity
consensus-distillation
geometric-deep-learning
amoe
custom_code
Model card Files Files and versions
xet
Metrics Training metrics Community

Instructions to use AbstractPhil/captionbert-8192-v2-b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

  • Libraries
  • Transformers

    How to use AbstractPhil/captionbert-8192-v2-b with Transformers:

    # Use a pipeline as a high-level helper
    from transformers import pipeline
    
    pipe = pipeline("feature-extraction", model="AbstractPhil/captionbert-8192-v2-b", trust_remote_code=True)
    # Load model directly
    from transformers import AutoModel
    model = AutoModel.from_pretrained("AbstractPhil/captionbert-8192-v2-b", trust_remote_code=True, device_map="auto")
  • Notebooks
  • Google Colab
  • Kaggle
captionbert-8192-v2-b / amoe /b-collective
23 MB
Ctrl+K
Ctrl+K
  • 1 contributor
History: 4 commits
AbstractPhil's picture
AbstractPhil
card: the b-collective shipped with NO README. Adds the native 8-task table (.6031 -> .7294, BIOSSES inverts favourably vs v2), the trunk-bound decomposition (port -.0424, keys recover 29%, retrain recovers all), alignment health, and the final_model.pt-vs-best_model.pt root distinction between the two repos
fbd7a0d verified 3 days ago
  • README.md
    6.37 kB
    card: the b-collective shipped with NO README. Adds the native 8-task table (.6031 -> .7294, BIOSSES inverts favourably vs v2), the trunk-bound decomposition (port -.0424, keys recover 29%, retrain recovers all), alignment health, and the final_model.pt-vs-best_model.pt root distinction between the two repos 3 days ago
  • captionbert-b-arms-native.dispatch.pt
    1.59 MB
    xet
    final 3 days ago
  • captionbert-b-arms-realigned.dispatch.pt
    1.59 MB
    xet
    stage A: re-aligned dispatch 3 days ago
  • equiv.anchor.pt
    6.6 MB
    xet
    stage B: anchors retrained on -b 3 days ago
  • metrics.json
    13 kB
    final 3 days ago
  • paraphrase.anchor.pt
    6.61 MB
    xet
    stage B: anchors retrained on -b 3 days ago
  • simplify.anchor.pt
    6.61 MB
    xet
    stage B: anchors retrained on -b 3 days ago