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 3072
    5. 8-RES_FS3072-JB
    6. 289
    Prev
    Next
    INDEX
    Explanations

    quotes and references from various texts and articles

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs3072-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    3,072
    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
    87
    +0.46
    2.5%
    64
    +0.14
    0.8%
    29
    +0.09
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    87
    +0.46
    -0.06
    756
    +0.14
    -0.10
    283
    +0.09
    -0.00
    Negative Logits
     *)
    -0.55
     ?)
    -0.51
    !)
    -0.50
    ffen
    -0.49
     depends
    -0.49
    )|
    -0.49
     Roose
    -0.48
    -)
    -0.48
    ?)
    -0.47
    iaries
    -0.45
    POSITIVE LOGITS
     respectively
    0.98
    etheless
    0.73
    *.
    0.61
    collect
    0.54
    anwhile
    0.50
     culminating
    0.50
    åĮ
    0.50
    inguished
    0.50
    NetMessage
    0.48
    ãĢĤ
    0.48
    Activations Density 7.993%

    No Known Activations