© 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. 22450
    Prev
    Next
    INDEX
    Explanations

    words related to negative impacts or harm

    oai_token-act-pair · gpt-3.5-turbo

    instances of the word "damaging" and related terms expressing adverse impacts

    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
    tu
    -0.68
    bert
    -0.64
     Fut
    -0.62
     simplicity
    -0.61
    ees
    -0.61
     Coh
    -0.61
    live
    -0.61
    âĦ¢
    -0.61
     Holmes
    -0.59
    ek
    -0.59
    POSITIVE LOGITS
     damaging
    3.57
     harmful
    1.94
     detrimental
    1.83
     harming
    1.74
     compromising
    1.71
     destructive
    1.70
     devastating
    1.60
     crippling
    1.60
     damage
    1.55
     corros
    1.52
    Activations Density 0.016%

    No Known Activations