© 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. 7-RES-JB
    11. 2884
    Prev
    Next
    INDEX
    Explanations

    words related to news events, such as "hack" and "prize-winning journalist"

    oai_token-act-pair · gpt-3.5-turbo

    references to cybersecurity incidents and related reports

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 7-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    7
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    ependent
    -0.58
     CFL
    -0.57
    -+-+
    -0.56
     Adin
    -0.54
    omething
    -0.54
    luster
    -0.54
    sembly
    -0.53
     fixtures
    -0.52
    verage
    -0.51
    eneg
    -0.51
    POSITIVE LOGITS
     notwithstanding
    0.76
     (@
    0.71
    !).
    0.66
    ="/
    0.62
    ?).
    0.62
    ãģ®å
    0.62
     adv
    0.61
    Redditor
    0.61
     replied
    0.60
    }.
    0.60
    Activations Density 0.966%

    No Known Activations