© 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 12288
    9. 8-RES_FS12288-JB
    11. 730
    Prev
    Next
    INDEX
    Explanations

    medical-related terms or information, focusing on the risk of smoking for individuals with diabetes and the impact of certain external factors on health outcomes

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs12288-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    12,288
    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
    391
    +0.12
    0.6%
    495
    +0.11
    0.5%
    654
    +0.08
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.12
    0.09
    495
    +0.11
    0.05
    409
    +0.08
    0.04
    Negative Logits
     Canaver
    -0.70
    odcast
    -0.70
    icians
    -0.68
    kef
    -0.67
    »Ĵ
    -0.66
    oros
    -0.64
    parser
    -0.64
    £ı
    -0.63
    pecially
    -0.62
    ometimes
    -0.61
    POSITIVE LOGITS
     capacity
    1.15
     expectancy
    1.14
     density
    1.10
     intensity
    1.08
     rate
    1.07
     throughput
    1.06
     rates
    1.05
     capability
    1.04
     burden
    1.02
     efficiency
    1.02
    Activations Density 0.561%

    No Known Activations