© 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. 11-RES-JB
    11. 10083
    Prev
    Next
    INDEX
    Explanations

    words or phrases expressing strong emotional evaluations like "egregious," "saddening," and "laughable."

    oai_token-act-pair · gpt-3.5-turbo

    words that indicate negative or distressing qualities and emphasize their significance

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 11-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    11
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     warr
    -0.63
     incorpor
    -0.63
     Nope
    -0.61
     comr
    -0.61
     invincible
    -0.59
     shenan
    -0.59
     hemor
    -0.58
     bye
    -0.58
     wills
    -0.57
     broom
    -0.57
    POSITIVE LOGITS
     considering
    1.12
     given
    1.06
    given
    1.04
     because
    0.97
    because
    0.87
     when
    0.87
     insofar
    0.86
     owing
    0.81
     since
    0.81
    today
    0.78
    Activations Density 0.204%

    No Known Activations