© 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. 8-RES-JB
    11. 7707
    Prev
    Next
    INDEX
    Explanations

    terms related to attacks on Western and US interests

    oai_token-act-pair · gpt-3.5-turbo

    references to terrorist activities and attacks against the United States or its interests

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    8
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    oult
    -0.75
    lopp
    -0.66
    ammy
    -0.66
    adjust
    -0.65
    FU
    -0.64
    rosc
    -0.64
    ISTORY
    -0.63
    cend
    -0.63
    apter
    -0.62
    join
    -0.62
    POSITIVE LOGITS
     targets
    1.02
     unarmed
    0.89
     innocent
    0.85
     civilians
    0.81
     helpless
    0.79
     innoc
    0.77
     unsuspecting
    0.77
     target
    0.76
     unprotected
    0.75
     suspected
    0.75
    Activations Density 0.745%

    No Known Activations