© 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. 3-RES-JB
    11. 4075
    Prev
    Next
    INDEX
    Explanations

    phrases related to social media platforms, especially Reddit and Imgur

    oai_token-act-pair · gpt-3.5-turbo

    references to the social media platform Reddit

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 3-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    3
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     Joyce
    -0.78
     Gw
    -0.77
     Lag
    -0.74
     Pf
    -0.72
     Ferr
    -0.71
    NetMessage
    -0.71
     Thompson
    -0.68
    ////////////////////////////////
    -0.66
     OD
    -0.66
     Witness
    -0.65
    POSITIVE LOGITS
    reddit
    1.15
    reddits
    1.14
     subreddits
    0.94
     subreddit
    0.94
    Reddit
    0.91
    icum
    0.85
    versive
    0.85
    unal
    0.83
    alion
    0.83
    ality
    0.83
    Activations Density 0.014%

    No Known Activations