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

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI 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. 9-RES-JB
    11. 13715
    Prev
    Next
    INDEX
    Explanations

    the word "guilty" with varying levels of activation

    oai_token-act-pair · gpt-3.5-turbo

    references to guilt and guilty findings in a legal context

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 9-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    9
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    psey
    -0.81
    Dub
    -0.73
    edia
    -0.72
    quickShipAvailable
    -0.69
     Remastered
    -0.68
    pora
    -0.67
    andel
    -0.66
    oldown
    -0.66
    accompan
    -0.65
    Downloadha
    -0.65
    POSITIVE LOGITS
     verdict
    1.04
     plea
    0.89
     conscience
    0.87
     guilt
    0.83
     unfocusedRange
    0.79
     innocence
    0.78
     pleas
    0.78
     wrongdoing
    0.75
     guilty
    0.73
    alty
    0.72
    Activations Density 0.040%

    No Known Activations