Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 49152
    5. 8-RES_FS49152-JB
    6. 41248
    Prev
    Next
    INDEX
    Explanations

    references to news outlets or media affiliations

    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
    127
    +0.09
    0.4%
    623
    +0.08
    0.4%
    373
    +0.08
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    767
    +0.09
    0.02
    586
    +0.08
    0.02
    28
    +0.08
    0.02
    Negative Logits
     guiName
    -0.85
    etheless
    -0.78
     tyr
    -0.68
    conservancy
    -0.67
    radical
    -0.67
    luster
    -0.63
    inav
    -0.61
    icular
    -0.58
     proced
    -0.57
    onential
    -0.56
    POSITIVE LOGITS
    )—
    1.56
    )"
    1.56
    )
    1.56
    )--
    1.52
    ):
    1.51
    ),"
    1.48
    )'
    1.41
    )|
    1.38
    !)
    1.37
    )(
    1.37
    Activations Density 0.073%

    No Known Activations