© 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. 1416
    Prev
    Next
    INDEX
    Explanations

    words related to international relations and diplomatic activities

    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
    483
    +0.13
    0.6%
    565
    +0.10
    0.5%
    294
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    203
    +0.13
    0.07
    483
    +0.10
    0.06
    208
    +0.10
    0.06
    Negative Logits
    ãĤ©
    -0.76
    äºĶ
    -0.72
    çĶ°
    -0.70
    QB
    -0.68
    MY
    -0.67
    ãĥĨ
    -0.66
    ãĥĩ
    -0.65
    iHUD
    -0.62
    chal
    -0.62
    çľ
    -0.62
    POSITIVE LOGITS
     nevertheless
    1.36
     nonetheless
    1.25
     anecd
    1.14
     suffice
    0.97
     luckily
    0.85
     hey
    0.78
     fortunately
    0.75
     guesses
    0.72
     theoretically
    0.69
     indications
    0.68
    Activations Density 0.338%

    No Known Activations