© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW
    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    2. Joseph Bloom · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 12288
    5. 8-RES_FS12288-JB
    6. 1505
    Prev
    Next
    INDEX
    Explanations

    information about film crew setups and movie production details

    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
    315
    +0.11
    0.5%
    84
    +0.10
    0.5%
    747
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    760
    +0.11
    0.03
    767
    +0.10
    0.03
    698
    +0.10
    0.03
    Negative Logits
     Cologne
    -0.80
     Hamburg
    -0.79
    AMS
    -0.73
    AMA
    -0.72
     Shepherd
    -0.71
    CCC
    -0.71
     TAG
    -0.69
     Adventures
    -0.69
     Clubs
    -0.69
     Lac
    -0.69
    POSITIVE LOGITS
    not
    1.00
     unob
    0.91
    nor
    0.88
    una
    0.85
    nob
    0.84
    no
    0.83
    oru
    0.81
    inyl
    0.80
    å¦
    0.79
    or
    0.78
    Activations Density 0.224%

    No Known Activations