© 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. 8-RES-JB
    11. 8257
    Prev
    Next
    INDEX
    Explanations

    phrases indicating confusion or interest in understanding a situation

    oai_token-act-pair · gpt-3.5-turbo

    phrases indicating ongoing events or situations

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    8
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    eah
    -0.76
    onso
    -0.75
    ĵ
    -0.74
    attery
    -0.74
    ritical
    -0.73
    ply
    -0.72
    ©¶æ
    -0.71
    prise
    -0.71
    ursed
    -0.71
    oning
    -0.70
    POSITIVE LOGITS
     inside
    0.85
     unfolding
    0.85
     behind
    0.84
    between
    0.82
     unnoticed
    0.81
    behind
    0.80
     underneath
    0.78
     elsewhere
    0.77
     unfold
    0.74
    inside
    0.74
    Activations Density 0.047%

    No Known Activations