© 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. 8-RES-JB
    11. 1344
    Prev
    Next
    INDEX
    Explanations

    references to physical actions involving grabbing or holding someone

    oai_token-act-pair · gpt-3.5-turbo

    instances of the phrase "by [someone/something]," typically indicating an action performed by a person

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    8
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    actly
    -0.65
    TB
    -0.64
    itives
    -0.61
    atem
    -0.60
    llor
    -0.60
    ati
    -0.57
    BSD
    -0.56
    GI
    -0.56
    SPONSORED
    -0.56
    YA
    -0.55
    POSITIVE LOGITS
     virtue
    1.21
    products
    1.20
    akuya
    1.08
     default
    1.02
    laws
    0.99
    catch
    0.99
    gone
    0.97
     leaps
    0.94
    product
    0.91
    pass
    0.91
    Activations Density 0.132%

    No Known Activations