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. 13541
    Prev
    Next
    INDEX
    Explanations

    pronouns related to individuals or groups

    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
    167
    +0.23
    1.7%
    7
    +0.17
    1.2%
    178
    +0.17
    1.2%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    224
    +0.23
    0.02
    148
    +0.17
    0.02
    235
    +0.17
    0.02
    Negative Logits
    phrine
    -0.83
     livest
    -0.76
     Fourth
    -0.76
    âĶĢâĶĢâĶĢâĶĢ
    -0.75
    align
    -0.75
     illum
    -0.71
     besie
    -0.69
     collapses
    -0.67
     horr
    -0.66
     Caucasus
    -0.65
    POSITIVE LOGITS
    krit
    0.88
    keyes
    0.87
    igned
    0.83
    zynski
    0.79
    erie
    0.76
    gling
    0.75
    eer
    0.74
    abilia
    0.73
    inea
    0.72
    inois
    0.72
    Activations Density 0.116%

    No Known Activations