© 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. 0-RES-JB
    11. 11935
    Prev
    Next
    INDEX
    Explanations

    the word "Never" followed by positive or neutral statements

    oai_token-act-pair · gpt-3.5-turbo

    the repetition of the word "Never" in various contexts

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 0-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    0
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    urat
    -0.74
     preliminary
    -0.64
    abus
    -0.61
    amination
    -0.61
     territorial
    -0.59
     prelim
    -0.58
    ina
    -0.58
    iya
    -0.58
     recap
    -0.58
    ordinate
    -0.58
    POSITIVE LOGITS
     Never
    3.58
    Never
    2.85
    never
    2.23
     NEVER
    2.03
     Always
    1.81
    Always
    1.57
     never
    1.53
     Ever
    1.47
     ALWAYS
    1.27
     Forever
    1.19
    Activations Density 0.023%

    No Known Activations