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 24576
    5. 8-RES_FS24576-JB
    6. 368
    Prev
    Next
    INDEX
    Explanations

    dialogue and storytelling involving interpersonal conversations

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs24576-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    24,576
    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
    +0.15
    0.7%
    447
    +0.13
    0.6%
    680
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    380
    +0.15
    0.03
    680
    +0.13
    0.03
    399
    +0.10
    0.03
    Negative Logits
     inflic
    -0.65
     topp
    -0.65
     remains
    -0.64
     subord
    -0.62
     vessels
    -0.62
    })
    -0.62
    SIGN
    -0.62
     protections
    -0.61
    forming
    -0.60
     outward
    -0.60
    POSITIVE LOGITS
     lately
    0.84
     reddit
    0.83
     cowork
    0.82
    inav
    0.81
     acquaintance
    0.78
     Dmitry
    0.77
     recently
    0.77
     commenter
    0.73
    nesday
    0.72
     discussing
    0.72
    Activations Density 0.474%

    No Known Activations