© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    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. 3-RES-JB
    11. 20236
    Prev
    Next
    INDEX
    Explanations

    words related to imitation, false representation, or something not genuine

    oai_token-act-pair · gpt-3.5-turbo

    terms related to pseudo or quasi concepts, often in a political or social context

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 3-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    3
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    lain
    -0.76
    ribly
    -0.68
     Qiao
    -0.67
     Norn
    -0.67
    utherland
    -0.66
    hips
    -0.65
    IAL
    -0.65
    rified
    -0.65
    士
    -0.64
    REAM
    -0.63
    POSITIVE LOGITS
    actively
    0.88
    educ
    0.88
    chal
    0.87
    isd
    0.87
    entimes
    0.87
    spective
    0.86
    cele
    0.85
    judicial
    0.85
    -)
    0.82
    jit
    0.81
    Activations Density 0.020%

    No Known Activations