© 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. 4-RES-JB
    11. 19439
    Prev
    Next
    INDEX
    Explanations

    phrases related to governmental or bureaucratic activities

    oai_token-act-pair · gpt-3.5-turbo

    occurrences of the term "Affairs" in various contexts

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 4-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    4
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    gling
    -0.89
    emade
    -0.75
    iple
    -0.75
    ãĤ©
    -0.74
    bring
    -0.70
    packed
    -0.69
    own
    -0.69
    udging
    -0.68
    ilts
    -0.67
    ifts
    -0.67
    POSITIVE LOGITS
     Affairs
    0.93
     Correspond
    0.87
     Bureau
    0.83
    pport
    0.82
     Agency
    0.79
     Studies
    0.78
     affairs
    0.77
     Profession
    0.77
     Policy
    0.74
     Counsel
    0.74
    Activations Density 0.014%

    No Known Activations