© 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. 3-RES-JB
    11. 15684
    Prev
    Next
    INDEX
    Explanations

    words related to the literal interpretation of language and expressions

    oai_token-act-pair · gpt-3.5-turbo

    words that emphasize literal interpretations and metaphors

    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
    ramid
    -0.82
    pered
    -0.77
    ccoli
    -0.71
    reau
    -0.70
    adish
    -0.69
    Crash
    -0.67
     Buchanan
    -0.66
    ACP
    -0.65
    ierra
    -0.65
    resses
    -0.65
    POSITIVE LOGITS
     literal
    0.90
    TY
    0.83
    orical
    0.80
     interpretation
    0.79
     truth
    0.75
    istically
    0.73
     translation
    0.72
     analogue
    0.72
    aneous
    0.72
     incarnation
    0.71
    Activations Density 0.006%

    No Known Activations