Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsBlogSlackPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 3072
    5. 8-RES_FS3072-JB
    6. 1520
    Prev
    Next
    INDEX
    Explanations

    phrases related to completion or addition

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs3072-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    3,072
    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
    726
    +0.10
    0.4%
    259
    +0.09
    0.4%
    552
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    245
    +0.10
    0.05
    47
    +0.09
    0.04
    1
    +0.09
    0.04
    Negative Logits
    recogn
    -0.72
    conduct
    -0.69
    nature
    -0.69
    pract
    -0.68
    isation
    -0.65
    rf
    -0.64
    peer
    -0.64
    Collins
    -0.63
    won
    -0.63
    utan
    -0.63
    POSITIVE LOGITS
     coffers
    1.02
     shelves
    1.01
     vacancies
    0.99
     gaps
    0.94
     arenas
    0.91
     stadiums
    0.89
     rooms
    0.86
     notebooks
    0.84
    oresc
    0.82
     pores
    0.81
    Activations Density 0.989%

    No Known Activations