© 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 24576
    9. 8-RES_FS24576-JB
    11. 18802
    Prev
    Next
    INDEX
    Explanations

    phrases related to legal and financial matters

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs24576-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    24,576
    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
    726
    +0.13
    0.6%
    190
    +0.12
    0.6%
    44
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    726
    +0.13
    0.05
    190
    +0.12
    0.05
    287
    +0.09
    0.03
    Negative Logits
    urches
    -0.65
    instead
    -0.63
     WRITE
    -0.61
    heid
    -0.59
     Sending
    -0.59
     Chains
    -0.59
     Agenda
    -0.58
     Submission
    -0.58
     Creed
    -0.57
     MARK
    -0.57
    POSITIVE LOGITS
     importantly
    1.28
    likely
    1.03
     likely
    1.00
     notably
    0.96
    afa
    0.93
     prominently
    0.90
     certainly
    0.88
     assured
    0.86
    rar
    0.83
     definitely
    0.81
    Activations Density 0.051%

    No Known Activations