© 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. 2517
    Prev
    Next
    INDEX
    Explanations

    communication content elements in the context of a New York Times subscription

    oai_token-act-pair · gpt-3.5-turbo

    phrases related to subscription services and user interactions

    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
    aq
    -0.58
    IRE
    -0.58
    uation
    -0.58
    Wan
    -0.58
    isks
    -0.57
    mat
    -0.57
    NZ
    -0.56
     [+
    -0.56
    pex
    -0.55
     rehearsal
    -0.54
    POSITIVE LOGITS
    interstitial
    0.67
     subscribing
    0.67
     Carbuncle
    0.66
    entimes
    0.65
    scribe
    0.63
     });
    0.60
    DragonMagazine
    0.56
    andise
    0.56
    reddits
    0.56
    )</
    0.56
    Activations Density 0.037%

    No Known Activations