© 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. 14432
    Prev
    Next
    INDEX
    Explanations

    words related to the lowest part or position, or extreme negative outcomes

    oai_token-act-pair · gpt-3.5-turbo

    the phrase "the bottom line" or variations indicating conclusions or essential points

    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
    thouse
    -0.80
    Äĩ
    -0.76
    ement
    -0.73
    uld
    -0.72
    kson
    -0.67
    encers
    -0.67
    asca
    -0.67
    inct
    -0.65
     Vu
    -0.65
    qua
    -0.64
    POSITIVE LOGITS
    bottom
    1.04
     bottom
    1.01
    Bottom
    0.99
     drawer
    0.89
     Bottom
    0.88
     srfAttach
    0.74
     shelf
    0.73
     bracket
    0.72
    most
    0.72
     toolbar
    0.70
    Activations Density 0.009%

    No Known Activations