© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Open Source Sparse Autoencoders for all Residual Stream Layers of GPT2-Small
    5. GPT2-Small
    7. Residual Stream
    9. 0-RES-JB
    11. 20420
    Prev
    Next
    INDEX
    Explanations

    professional or competitive athletes

    oai_token-act-pair · gpt-3.5-turbo

    references to athletes and participants in various contexts

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 0-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.hook_resid_pre
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    24,576
    Data Type
    torch.float32
    Hook Point
    blocks.0.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    0
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     families
    -1.12
     iPads
    -1.01
     households
    -1.00
     Presidents
    -0.99
     customers
    -0.98
     presidents
    -0.97
     businesses
    -0.97
     Republicans
    -0.95
     Muslims
    -0.95
     MPs
    -0.95
    POSITIVE LOGITS
     participant
    1.74
     practitioner
    1.60
     believer
    1.57
     traveler
    1.56
     observer
    1.55
     athlete
    1.55
     performer
    1.52
     rider
    1.49
     survivor
    1.49
     dancer
    1.49
    Activations Density 0.105%

    No Known Activations