© 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. 6496
    Prev
    Next
    INDEX
    Explanations

    words related to wrongful actions or misconduct

    oai_token-act-pair · gpt-3.5-turbo

    references to wrongdoing or misconduct

    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
     Diplom
    -0.69
     Archives
    -0.64
     Churchill
    -0.64
     Novel
    -0.63
    amph
    -0.63
    gae
    -0.63
     Interstellar
    -0.63
     Fleet
    -0.61
     Drone
    -0.61
     Dynam
    -0.61
    POSITIVE LOGITS
     wrongdoing
    2.37
    large
    1.10
     Fro
    1.09
     harmed
    0.85
    ppings
    0.84
     innocence
    0.83
    thought
    0.81
     graft
    0.80
     scrimmage
    0.78
    terday
    0.78
    Activations Density 0.036%

    No Known Activations