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

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. OpenAI · Sparse Autoencoder for GPT2-Small - v5 2024
    3. GPT2-Small
    4. Attention Out - 128k
    5. 0-ATT_128K-OAI
    6. 13539
    Prev
    Next
    INDEX
    Explanations

    references to drug overdoses and related experiences

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    jbloom/GPT2-Small-OAI-v5-128k-attn-out-SAEs/v5_128k_layer_0
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    131,072
    Data Type
    torch.float32
    Hook Name
    blocks.0.hook_attn_out
    Hook Layer
    0
    Architecture
    standard
    Context Size
    64
    Dataset
    Skylion007/openwebtext
    Activation Function
    topk
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    659
    +0.23
    1.6%
    503
    +0.17
    1.2%
    160
    +0.17
    1.2%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    729
    +0.23
    0.01
    630
    +0.17
    0.01
    454
    +0.17
    0.01
    Negative Logits
    ividual
    -0.79
     Hots
    -0.78
    aneously
    -0.78
    Reviewer
    -0.78
    wcs
    -0.75
    icial
    -0.74
    ³³³
    -0.73
    icz
    -0.72
    netflix
    -0.72
    imester
    -0.72
    POSITIVE LOGITS
    llah
    0.92
    pace
    0.81
    uper
    0.74
     subsequ
    0.73
    requires
    0.73
     systems
    0.72
    bars
    0.72
    ysis
    0.72
    EMS
    0.70
    erv
    0.69
    Activations Density 0.008%

    No Known Activations