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. 323
    Prev
    Next
    INDEX
    Explanations

    financial and economic terms and discussions

    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
    623
    +0.13
    0.6%
    60
    +0.10
    0.5%
    274
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    274
    +0.13
    0.08
    761
    +0.10
    0.09
    402
    +0.09
    0.08
    Negative Logits
    yah
    -0.68
    NING
    -0.63
    bis
    -0.62
    rpm
    -0.62
    clinton
    -0.61
    dog
    -0.61
    wave
    -0.59
    hare
    -0.59
    Ĥİ
    -0.58
    spread
    -0.58
    POSITIVE LOGITS
    endum
    1.12
     insult
    1.07
    itional
    0.93
    ressing
    0.93
    itionally
    0.92
     bonuses
    0.87
     bonus
    0.86
    resses
    0.86
    added
    0.84
     extras
    0.84
    Activations Density 3.993%

    No Known Activations