© 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. 9029
    Prev
    Next
    INDEX
    Explanations

    phrases related to takedowns, legal actions, or aggressive maneuvers

    oai_token-act-pair · gpt-3.5-turbo

    occurrences of the word "takedown" and related names or terms associated with actions taken against content, especially in a legal or copyright context

    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
     cryst
    -0.75
    Flor
    -0.70
    istically
    -0.69
    317
    -0.62
    OST
    -0.62
    ril
    -0.60
    UME
    -0.60
    ãĥ¥
    -0.60
    ATE
    -0.60
    ORPG
    -0.59
    POSITIVE LOGITS
    mission
    0.82
    amn
    0.79
    quished
    0.79
    halla
    0.77
     GOODMAN
    0.77
    akedown
    0.75
    missions
    0.71
    heses
    0.71
    aline
    0.71
     notices
    0.70
    Activations Density 0.037%

    No Known Activations