© 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. 410
    Prev
    Next
    INDEX
    Explanations

    phrases related to the concept of something being the final or most recent

    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
    373
    +0.15
    0.7%
    223
    +0.11
    0.5%
    38
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    223
    +0.15
    0.05
    458
    +0.11
    0.05
    183
    +0.10
    0.04
    Negative Logits
    ingen
    -0.77
    selves
    -0.74
    mbuds
    -0.71
    olas
    -0.64
    amber
    -0.63
    cil
    -0.63
    Kay
    -0.62
    Rect
    -0.61
    aband
    -0.61
    nel
    -0.61
    POSITIVE LOGITS
     gasp
    1.13
     ditch
    1.03
     vest
    1.00
     rites
    0.97
     straw
    0.94
     installment
    0.94
     decade
    0.93
     bast
    0.93
     remnant
    0.89
     few
    0.86
    Activations Density 0.089%

    No Known Activations