© 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. 8114
    Prev
    Next
    INDEX
    Explanations

    phrases related to illegal activities or items

    oai_token-act-pair · gpt-3.5-turbo

    terms related to illegal activities or behavior

    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
    isson
    -0.77
     Brief
    -0.75
    warm
    -0.73
    igne
    -0.68
    izons
    -0.67
    ourse
    -0.67
    feed
    -0.66
    quin
    -0.66
    IST
    -0.66
     Bread
    -0.64
    POSITIVE LOGITS
     illegal
    3.58
    illegal
    3.08
     Illegal
    2.75
     illegally
    2.39
     unlawful
    2.30
     illeg
    2.12
     unauthorized
    1.96
     illicit
    1.96
     unconstitutional
    1.91
    legal
    1.65
    Activations Density 0.027%

    No Known Activations