Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Open Source Sparse Autoencoders for all Residual Stream Layers of GPT2-Small
    3. GPT2-Small
    4. Residual Stream
    5. 4-RES-JB
    6. 8309
    Prev
    Next
    INDEX
    Explanations

    reference to criminal activities and individuals involved in illegal actions

    oai_token-act-pair · gpt-3.5-turbo

    references to criminals and criminality

    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
    HAHA
    -0.67
    arp
    -0.65
    urgical
    -0.65
    Ħ¢
    -0.64
    chell
    -0.64
    chron
    -0.63
    ŃĶ
    -0.63
    orse
    -0.62
    yip
    -0.62
    rolog
    -0.62
    POSITIVE LOGITS
     mastermind
    0.95
     prey
    0.88
     gangs
    0.88
     trafficking
    0.81
     criminals
    0.79
     robbing
    0.78
     offenders
    0.78
     stealing
    0.78
     trespass
    0.77
     smugglers
    0.77
    Activations Density 0.015%

    No Known Activations