© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI 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. 1907
    Prev
    Next
    INDEX
    Explanations

    mentions of online platforms and social media postings

    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
    649
    +0.11
    0.5%
    487
    +0.10
    0.4%
    313
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    269
    +0.11
    0.05
    313
    +0.10
    0.05
    654
    +0.09
    0.05
    Negative Logits
    ¯¯
    -0.65
    acter
    -0.61
    rious
    -0.60
    rats
    -0.59
    amus
    -0.59
    Abstract
    -0.59
    waters
    -0.57
    arnaev
    -0.57
    ynt
    -0.57
     ens
    -0.57
    POSITIVE LOGITS
     behalf
    1.58
    site
    1.08
     Pastebin
    1.08
     Youtube
    1.08
     YouTube
    1.06
    demand
    1.04
     youtube
    1.02
     Twitter
    1.01
    screen
    1.00
     Facebook
    0.96
    Activations Density 0.217%

    No Known Activations