© 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 768
    9. 8-RES_FS768-JB
    11. 648
    Prev
    Next
    INDEX
    Explanations

    descriptive texts about technical tasks or procedures

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs768-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    768
    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
    314
    +0.16
    0.7%
    157
    +0.12
    0.5%
    160
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    157
    +0.16
    0.19
    675
    +0.12
    0.18
    327
    +0.11
    0.17
    Negative Logits
     JPEG
    -0.73
     silhou
    -0.69
     Mobil
    -0.66
     Gmail
    -0.65
     decomp
    -0.63
     Mic
    -0.63
     bun
    -0.63
     Democracy
    -0.61
     BBC
    -0.61
     Axel
    -0.59
    POSITIVE LOGITS
    s
    1.34
    ski
    1.16
    sure
    1.10
    tal
    1.08
    tis
    1.08
    t
    1.04
    ship
    1.03
    til
    1.00
    thy
    1.00
    tre
    0.99
    Activations Density 2.147%

    No Known Activations