© 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. 11-RES-JB
    11. 100
    Prev
    Next
    INDEX
    Explanations

    words related to criminal activities or actions that break the law

    oai_token-act-pair · gpt-3.5-turbo

    occurrences of the word "commit" and its variations, particularly in relation to harmful or illegal actions

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 11-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    11
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    framework
    -0.73
    jp
    -0.70
     Remastered
    -0.70
    anned
    -0.68
    gypt
    -0.67
    angular
    -0.64
    iewicz
    -0.63
    è£ħ
    -0.62
    cn
    -0.62
    CN
    -0.61
    POSITIVE LOGITS
     suicide
    1.51
     adultery
    1.37
     perjury
    1.34
     atrocities
    1.27
     crimes
    1.26
     heinous
    1.21
     treason
    1.20
     offences
    1.16
     acts
    1.11
     misdem
    1.10
    Activations Density 0.047%

    No Known Activations