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

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI 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. 5084
    Prev
    Next
    INDEX
    Explanations

    questions or instructions starting with "HOW", "WHAT", and "WHERE"

    oai_token-act-pair · gpt-3.5-turbo

    interrogative phrases that seek to identify locations or situations

    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
    zag
    -0.70
    onda
    -0.63
     Smile
    -0.61
    acci
    -0.61
    gru
    -0.60
    rics
    -0.59
     Lore
    -0.58
    stan
    -0.58
    aph
    -0.57
     nerv
    -0.56
    POSITIVE LOGITS
    WHERE
    3.15
    HOW
    2.21
    WHAT
    1.79
     WHERE
    1.67
     WHAT
    1.29
    Whereas
    1.23
    ONDON
    1.17
    eenth
    1.17
     ALSO
    1.14
     HOW
    1.08
    Activations Density 0.040%

    No Known Activations