© 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. 4-RES-JB
    11. 16091
    Prev
    Next
    INDEX
    Explanations

    names of states or regions and discuss their conditions

    oai_token-act-pair · gpt-3.5-turbo

    references to states

    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
     Cosponsors
    -0.83
    <?
    -0.82
     reperto
    -0.81
    */(
    -0.80
    uristic
    -0.75
    wcs
    -0.72
    bley
    -0.69
    poon
    -0.67
    reddits
    -0.66
    ertodd
    -0.66
    POSITIVE LOGITS
     course
    0.79
     affairs
    0.75
     justice
    0.72
     Darkness
    0.69
     Mour
    0.68
    rency
    0.67
     separation
    0.67
     Decay
    0.66
     Saud
    0.65
    ortunately
    0.65
    Activations Density 0.198%

    No Known Activations