© 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. 0-RES-JB
    11. 10320
    Prev
    Next
    INDEX
    Explanations

    words related to negative attributes like evil and sinister

    oai_token-act-pair · gpt-3.5-turbo

    words related to evil or malevolence

    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
     Photograph
    -0.75
    Supplement
    -0.74
    kamp
    -0.73
     Sport
    -0.72
    adel
    -0.68
    vet
    -0.68
    ournals
    -0.66
    azines
    -0.65
    won
    -0.65
    usk
    -0.64
    POSITIVE LOGITS
     evil
    3.53
     Evil
    2.60
    evil
    2.59
    Evil
    2.50
     evils
    2.20
     wicked
    1.61
     nefarious
    1.60
     sinister
    1.57
     demonic
    1.50
     vile
    1.48
    Activations Density 0.041%

    No Known Activations