© 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. 838
    Prev
    Next
    INDEX
    Explanations

    information related to current events and news, including details about taxi regulations, political commentary, medical emergencies, and technological advancements such as Amazon's products

    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
    373
    +0.30
    1.7%
    314
    +0.14
    0.8%
    627
    +0.07
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    373
    +0.30
    -0.06
    242
    +0.14
    0.04
    666
    +0.07
    0.04
    Negative Logits
     Solitaire
    -0.53
     Schn
    -0.47
     redacted
    -0.44
     Medal
    -0.44
     Bleach
    -0.44
     Balanced
    -0.43
     Zen
    -0.43
     Mult
    -0.43
     Hipp
    -0.43
     Schr
    -0.42
    POSITIVE LOGITS
    ®
    0.71
    Ń
    0.69
    ¸
    0.67
    º
    0.65
    ¯
    0.64
    ventures
    0.63
    »
    0.62
    eline
    0.61
    ãĤ§
    0.60
    ãĤ¦ãĤ¹
    0.60
    Activations Density 1.674%

    No Known Activations