© 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 49152
    9. 8-RES_FS49152-JB
    11. 37345
    Prev
    Next
    INDEX
    Explanations

    the presence of first-person plural and singular pronouns alongside specific action verbs and phrases indicating criticism or expectations related to social or political issues

    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
    660
    +0.10
    0.5%
    185
    +0.10
    0.5%
    483
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    662
    +0.10
    0.03
    762
    +0.10
    0.03
    49
    +0.10
    0.03
    Negative Logits
     Emin
    -0.57
     moreover
    -0.57
     Sigma
    -0.56
    bp
    -0.56
    rame
    -0.54
     Ack
    -0.54
    Chain
    -0.54
    bench
    -0.53
     AAA
    -0.53
    olate
    -0.52
    POSITIVE LOGITS
    erest
    0.69
     counterparts
    0.67
    zek
    0.65
     predecessors
    0.63
    resent
    0.61
    ayn
    0.60
    adesh
    0.60
    igr
    0.60
    thinkable
    0.59
    én
    0.58
    Activations Density 0.132%

    No Known Activations