© 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. 4-RES-JB
    11. 6405
    Prev
    Next
    INDEX
    Explanations

    references related to the term "Har" with varying levels of specificity, such as "Harpo," "Harley," names including "Har," or locations like "Harbin."

    oai_token-act-pair · gpt-3.5-turbo

    repeated mentions of the term "Har" possibly indicating a focus on a specific subject or entity related to that word

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

    No Comments

    Negative Logits
    éĹĺ
    -1.03
    ablishment
    -0.75
    uality
    -0.75
     semantics
    -0.72
    xual
    -0.71
    REDACTED
    -0.70
    lihood
    -0.70
    PsyNetMessage
    -0.69
     Giuliani
    -0.68
    ĸļ
    -0.67
    POSITIVE LOGITS
    rod
    1.13
    vard
    1.06
    rim
    1.05
    rier
    1.03
    riers
    1.03
    rah
    1.02
    mel
    0.97
    assment
    0.97
    rowing
    0.94
    rigan
    0.91
    Activations Density 0.009%

    No Known Activations