© 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 · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 12288
    9. 8-RES_FS12288-JB
    11. 1381
    Prev
    Next
    INDEX
    Explanations

    phrases related to making changes or adaptations

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs12288-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    12,288
    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
    481
    +0.11
    0.5%
    714
    +0.11
    0.5%
    66
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    300
    +0.11
    0.06
    690
    +0.11
    0.04
    714
    +0.09
    0.04
    Negative Logits
    wine
    -0.68
    english
    -0.66
     Nass
    -0.65
    ا
    -0.63
     Dahl
    -0.62
    utherland
    -0.61
    Topics
    -0.60
     Seymour
    -0.59
    essor
    -0.59
    NAS
    -0.59
    POSITIVE LOGITS
     effected
    1.15
     drastic
    1.12
     gradual
    1.01
     incremental
    0.99
     undone
    0.93
     occur
    0.90
     Enhance
    0.82
     ripple
    0.82
     accelerate
    0.81
     postpone
    0.81
    Activations Density 0.619%

    No Known Activations