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

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 1536
    5. 8-RES_FS1536-JB
    6. 423
    Prev
    Next
    INDEX
    Explanations

    words related to perceptions or impressions

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs1536-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    1,536
    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
    195
    +0.15
    0.7%
    603
    +0.12
    0.6%
    183
    +0.12
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    195
    +0.15
    0.07
    603
    +0.12
    0.08
    178
    +0.12
    0.08
    Negative Logits
    ests
    -0.89
    estern
    -0.84
    ilts
    -0.69
    ainers
    -0.68
    hal
    -0.67
    orem
    -0.66
    lished
    -0.66
    ribution
    -0.66
    cised
    -0.66
    restling
    -0.65
    POSITIVE LOGITS
     innocuous
    0.95
     awfully
    0.83
     oddly
    0.83
     poised
    0.83
     unstoppable
    0.80
     strangely
    0.79
     like
    0.77
     destined
    0.76
     promising
    0.76
    Pause
    0.74
    Activations Density 0.907%

    No Known Activations