© 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. 24347
    Prev
    Next
    INDEX
    Explanations

    phrases related to a specific location, likely associated with a person named "Le"

    oai_token-act-pair · gpt-3.5-turbo

    frequent references to a specific name or entity

    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
    ãĥ¼ãĥĨãĤ£
    -1.05
    ãĥ¼ãĥĨ
    -0.97
    aneers
    -0.94
    ividual
    -0.93
    etheless
    -0.92
    ãģĨ
    -0.90
    âĶģ
    -0.86
    raints
    -0.84
    ĸļ
    -0.81
     teasp
    -0.81
    POSITIVE LOGITS
    isure
    1.26
    lean
    1.10
    icester
    1.05
    agues
    1.05
    ukemia
    0.98
    opard
    0.96
    ather
    0.95
    aching
    0.91
    azy
    0.90
    ven
    0.86
    Activations Density 0.019%

    No Known Activations