© 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. 2-RES-JB
    11. 23698
    Prev
    Next
    INDEX
    Explanations

    phrases related to inappropriate behavior

    oai_token-act-pair · gpt-3.5-turbo

    instances of the word "inappropriate" in various contexts

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 2-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    2
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    ership
    -0.88
    rix
    -0.84
    abiding
    -0.81
    adr
    -0.78
    ript
    -0.78
    ingen
    -0.77
    ynthesis
    -0.77
    hyde
    -0.76
    bern
    -0.76
    uster
    -0.75
    POSITIVE LOGITS
     inappropriate
    1.01
     improper
    0.86
     undermin
    0.84
     inappropriately
    0.82
     awaken
    0.81
     interference
    0.80
     misuse
    0.80
     interfere
    0.79
     interfered
    0.77
     behaviour
    0.77
    Activations Density 0.011%

    No Known Activations