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 768
    5. 8-RES_FS768-JB
    6. 228
    Prev
    Next
    INDEX
    Explanations

    legal terms and names related to lawsuits and cases

    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
    110
    +0.11
    0.5%
    431
    +0.10
    0.5%
    680
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.11
    0.26
    680
    +0.10
    0.10
    110
    +0.10
    0.13
    Negative Logits
    rupulous
    -0.84
    theless
    -0.80
    uria
    -0.79
    riel
    -0.78
    terday
    -0.76
    aws
    -0.74
    olas
    -0.74
    [_
    -0.73
    rily
    -0.73
    ensibly
    -0.73
    POSITIVE LOGITS
     lineup
    1.13
     debacle
    1.04
     era
    1.02
     program
    1.01
     initiative
    1.01
     expedition
    1.00
     playbook
    0.98
     facility
    0.98
     fiasco
    0.98
     bandwagon
    0.96
    Activations Density 6.658%

    No Known Activations