© 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 EvalsExportsinterp-engineNEWAPI 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. 11-RES-JB
    11. 6063
    Prev
    Next
    INDEX
    Explanations

    verbs related to achieving or accomplishing something, often in the form of "can + verb"

    oai_token-act-pair · gpt-3.5-turbo

    instances of the word "can" and its variations indicating ability or possibility

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 11-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    11
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    Choice
    -0.64
     Sod
    -0.63
     Maur
    -0.63
    senal
    -0.62
    DERR
    -0.62
    ULTS
    -0.61
    PDATE
    -0.61
     behavi
    -0.61
     Reich
    -0.59
     Sisters
    -0.59
    POSITIVE LOGITS
    't
    1.75
     afford
    1.41
     convince
    1.18
     persuade
    1.10
     manage
    1.03
     muster
    1.01
     tolerate
    0.98
     locate
    0.94
    NOT
    0.92
     find
    0.91
    Activations Density 0.098%

    No Known Activations