© 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. 2-RES-JB
    11. 15643
    Prev
    Next
    INDEX
    Explanations

    descriptions related to different groups, cultural backgrounds, and noble families

    oai_token-act-pair · gpt-3.5-turbo

    references to family and ethnic identity in various contexts

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 2-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    2
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     Supplementary
    -0.47
    activated
    -0.46
     notified
    -0.44
    say
    -0.43
    veyard
    -0.43
    ahon
    -0.42
     nearest
    -0.42
     Rosenthal
    -0.41
     amended
    -0.41
     Nicholson
    -0.39
    POSITIVE LOGITS
    Reviewer
    0.60
    etheless
    0.55
    cellence
    0.53
    terness
    0.51
     affair
    0.50
    izen
    0.49
    çľ
    0.48
    imity
    0.47
    Redditor
    0.46
     behavi
    0.44
    Activations Density 3.927%

    No Known Activations