© 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. 333
    Prev
    Next
    INDEX
    Explanations

    names of places or people

    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
    123
    +0.14
    0.6%
    640
    +0.12
    0.6%
    616
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    123
    +0.14
    0.10
    22
    +0.12
    0.09
    29
    +0.11
    0.09
    Negative Logits
    Published
    -0.69
     therein
    -0.66
    llan
    -0.62
     %%
    -0.62
     lodged
    -0.60
    ATHER
    -0.58
    username
    -0.57
    awei
    -0.56
    linger
    -0.56
     flagged
    -0.55
    POSITIVE LOGITS
     spite
    1.45
     lieu
    1.34
     retrospect
    1.22
     accordance
    1.20
     terms
    1.18
     vain
    1.17
    escap
    1.16
    clus
    1.15
    humane
    1.13
     hindsight
    1.13
    Activations Density 0.380%

    No Known Activations