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 1536
    5. 8-RES_FS1536-JB
    6. 953
    Prev
    Next
    INDEX
    Explanations

    instances of "one by one" mentioned in the text

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs1536-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    1,536
    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
    +0.32
    4.1%
    87
    +0.31
    3.9%
    756
    +0.09
    1.2%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    373
    +0.32
    -0.27
    87
    +0.31
    -0.15
    756
    +0.09
    -0.15
    Negative Logits
     Quote
    -0.21
     Âł
    -0.18
    )?
    -0.18
    ¶
    -0.18
     fixme
    -0.17
     tonight
    -0.17
     Originally
    -0.17
     yesterday
    -0.16
     fucking
    -0.15
     Transcript
    -0.15
    POSITIVE LOGITS
     advoc
    0.20
    undreds
    0.19
    Versions
    0.18
    vice
    0.18
    =~=~
    0.18
    BuyableInstoreAndOnline
    0.18
    yond
    0.18
    eredith
    0.18
    ãĥŁ
    0.18
    ãĥ¼ãĥ³
    0.18
    Activations Density 21.299%

    No Known Activations