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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 49152
    5. 8-RES_FS49152-JB
    6. 25664
    Prev
    Next
    INDEX
    Explanations

    references to scientific studies and their results

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs49152-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    49,152
    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
    373
    +0.14
    0.6%
    97
    +0.11
    0.5%
    134
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    97
    +0.14
    0.02
    126
    +0.11
    0.02
    368
    +0.10
    0.02
    Negative Logits
    gallery
    -0.84
     theirs
    -0.83
    Ire
    -0.78
    \/\/
    -0.72
    Company
    -0.71
    Beast
    -0.70
    DERR
    -0.69
    ALK
    -0.67
    Cooldown
    -0.66
     hers
    -0.66
    POSITIVE LOGITS
     Evaluation
    0.96
     Evidence
    0.96
     Strategies
    0.90
     Statistical
    0.87
     Trends
    0.86
     Predict
    0.85
     Behavioral
    0.85
    âĢIJ
    0.83
     polarization
    0.83
     inferred
    0.82
    Activations Density 0.114%

    No Known Activations