Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 24576
    5. 8-RES_FS24576-JB
    6. 19348
    Prev
    Next
    INDEX
    Explanations

    contexts related to legal cases and legal terminology

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs24576-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    24,576
    Data Type
    torch.float32
    Hook Point
    blocks.8.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    8
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    631
    +0.15
    0.7%
    668
    +0.10
    0.5%
    750
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    631
    +0.15
    0.05
    744
    +0.10
    0.04
    31
    +0.10
    0.04
    Negative Logits
     rumours
    -0.79
     ITV
    -0.72
    VERTISEMENT
    -0.72
    76561
    -0.72
     £
    -0.71
     feeds
    -0.69
     highs
    -0.68
    ources
    -0.67
     indo
    -0.67
    ;;;;;;;;;;;;
    -0.67
    POSITIVE LOGITS
     Scalia
    1.51
     justices
    1.39
    sup
    1.19
     Gins
    1.14
    Justice
    1.08
     upholding
    1.06
     appellate
    1.04
    Sup
    1.02
     ACLU
    1.01
     dissent
    1.01
    Activations Density 0.324%

    No Known Activations