© 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. 22387
    Prev
    Next
    INDEX
    Explanations

    words related to opinions or evaluations of actions, such as deserved, looks, cannot, and would

    oai_token-act-pair · gpt-3.5-turbo

    instances of verbs and subjects related to sports and performance

    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
    rina
    -0.59
    ependent
    -0.58
    ergy
    -0.56
    ormonal
    -0.53
    ento
    -0.52
     millenn
    -0.49
    redits
    -0.49
    undy
    -0.49
    eni
    -0.48
    ccording
    -0.48
    POSITIVE LOGITS
    rawdownloadcloneembedreportprint
    0.70
     himself
    0.69
    SourceFile
    0.68
     Himself
    0.66
     Paste
    0.65
    Redditor
    0.63
     herself
    0.61
    ombat
    0.57
    NetMessage
    0.56
    ĵĺ
    0.55
    Activations Density 1.674%

    No Known Activations