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

    mentions of heightened intensity or escalation

    oai_token-act-pair · gpt-3.5-turbo

    phrases related to intensity or severity, particularly involving the notion of something being "full-blown."

    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
    irez
    -0.74
    ierre
    -0.73
    wa
    -0.64
     tripod
    -0.63
     indo
    -0.62
    unte
    -0.61
     shorts
    -0.60
     vulner
    -0.60
     bulldo
    -0.59
     knees
    -0.59
    POSITIVE LOGITS
    blown
    2.37
    fledged
    1.70
     Sprint
    1.55
    ruly
    1.18
    productive
    0.89
     Ebola
    0.83
    vind
    0.78
    rage
    0.74
    formed
    0.73
    lance
    0.71
    Activations Density 0.009%

    No Known Activations