© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports 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. 3045
    Prev
    Next
    INDEX
    Explanations

    phrases related to avoiding consequences or breaking rules

    oai_token-act-pair · gpt-3.5-turbo

    instances of the phrase "get away" with various numerical values indicating frequency and context

    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
     Yamato
    -0.63
     Merrill
    -0.61
     urgency
    -0.59
    Interstitial
    -0.59
     Bulg
    -0.57
     succession
    -0.55
     Takeru
    -0.55
     Zeit
    -0.54
     Laksh
    -0.53
    GMT
    -0.52
    POSITIVE LOGITS
     safely
    0.73
    oned
    0.68
     unsc
    0.64
    uced
    0.64
    pun
    0.62
    door
    0.61
    ped
    0.60
    yp
    0.59
    =~=~
    0.59
    ixed
    0.58
    Activations Density 0.047%

    No Known Activations