© 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. 4-RES-JB
    11. 6691
    Prev
    Next
    INDEX
    Explanations

    names or labels within text

    oai_token-act-pair · gpt-3.5-turbo

    repeated mentions of the word "name" in various contexts

    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
    romy
    -0.79
     gif
    -0.71
    imeo
    -0.71
    hap
    -0.67
    yrinth
    -0.65
    icult
    -0.65
    psey
    -0.65
    Js
    -0.64
     Safari
    -0.64
    ilib
    -0.63
    POSITIVE LOGITS
    plates
    1.22
    plate
    1.14
    paces
    1.06
    mark
    0.88
    ames
    0.85
    names
    0.85
     naming
    0.83
     names
    0.83
     suffix
    0.81
     aliases
    0.81
    Activations Density 0.044%

    No Known Activations