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. 438
    Prev
    Next
    INDEX
    Explanations

    information and discussion about legal contracts, governmental involvement in marriages, and technology-related topics

    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
    264
    +0.14
    0.7%
    171
    +0.13
    0.6%
    726
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    171
    +0.14
    0.08
    81
    +0.13
    0.08
    264
    +0.10
    0.06
    Negative Logits
    ============
    -0.58
     unde
    -0.57
    ————————
    -0.55
     =================================================================
    -0.54
    Īè
    -0.54
     tension
    -0.53
    ...?
    -0.52
    âĶĢâĶĢâĶĢâĶĢ
    -0.52
    ****************
    -0.51
    />
    -0.51
    POSITIVE LOGITS
    foundland
    0.77
    etheless
    0.71
    intosh
    0.69
    ween
    0.68
    rite
    0.68
    arte
    0.68
    raine
    0.63
    llor
    0.62
    esses
    0.61
    etter
    0.60
    Activations Density 2.035%

    No Known Activations