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

    Neuronpedia

    APIAssistant AxisNEWCircuit 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. 505
    Prev
    Next
    INDEX
    Explanations

    terms related to post-war circumstances and issues

    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
    433
    +0.10
    0.5%
    710
    +0.10
    0.5%
    321
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    710
    +0.10
    0.07
    433
    +0.10
    0.07
    540
    +0.10
    0.07
    Negative Logits
     Qiao
    -0.79
    Ø©
    -0.71
     Bett
    -0.65
     Orchestra
    -0.65
    ï¸
    -0.64
    IBLE
    -0.64
     Sabha
    -0.63
    owship
    -0.63
    onge
    -0.63
     Brill
    -0.62
    POSITIVE LOGITS
    mortem
    1.42
    gres
    1.33
    doctoral
    1.26
    modern
    1.20
    natal
    1.12
    hum
    1.10
    greSQL
    1.10
    req
    1.09
    industrial
    1.07
     facto
    1.06
    Activations Density 1.129%

    No Known Activations