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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports 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. 4-RES-JB
    11. 12615
    Prev
    Next
    INDEX
    Explanations

    phrases related to a specific person named "Harald."

    oai_token-act-pair · gpt-3.5-turbo

    references to a specific name or figure

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

    No Comments

    Negative Logits
    PER
    -0.83
    displayText
    -0.76
    puter
    -0.74
    ORTS
    -0.70
    PLAY
    -0.69
    former
    -0.69
    soDeliveryDate
    -0.68
    GROUP
    -0.67
    Plex
    -0.66
    mercial
    -0.65
    POSITIVE LOGITS
    sburg
    1.15
    ald
    0.91
    rums
    0.90
    sson
    0.90
    orf
    0.90
    ry
    0.88
    emort
    0.86
    oran
    0.84
    ewater
    0.81
    stadt
    0.81
    Activations Density 0.014%

    No Known Activations