© 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. 4136
    Prev
    Next
    INDEX
    Explanations

    phrases related to relaxed and informal settings or behaviors

    oai_token-act-pair · gpt-3.5-turbo

    references to the concept of casualness

    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
    CVE
    -0.73
    Rated
    -0.73
    burgh
    -0.71
     Frie
    -0.71
     GOODMAN
    -0.70
     Luk
    -0.69
     Tens
    -0.68
    hner
    -0.66
    Tube
    -0.66
     Roses
    -0.65
    POSITIVE LOGITS
     casual
    0.93
    cas
    0.85
    ual
    0.84
    ty
    0.81
     observer
    0.80
     minded
    0.80
    aneously
    0.79
     stroll
    0.79
    ization
    0.78
    istically
    0.74
    Activations Density 0.012%

    No Known Activations