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

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsBlogSlackPrivacy & 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. 6478
    Prev
    Next
    INDEX
    Explanations

    instances of the word "irrelevant" and its variations, as well as the name "Trevor" in various contexts

    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
    151
    +0.20
    1.6%
    651
    +0.18
    1.4%
    726
    +0.16
    1.3%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    151
    +0.20
    0.01
    326
    +0.18
    0.01
    343
    +0.16
    0.01
    Negative Logits
    anamo
    -1.11
    ertodd
    -0.99
     Flavoring
    -0.80
    peed
    -0.75
    imore
    -0.75
    ropolitan
    -0.74
    ongyang
    -0.74
    ãħĭ
    -0.73
    eredith
    -0.73
    hammad
    -0.72
    POSITIVE LOGITS
    soDeliveryDate
    0.94
    thumbnails
    0.92
    Interstitial
    0.84
    oÄŁ
    0.81
    terness
    0.79
    vous
    0.78
     corrid
    0.77
    Pont
    0.76
    uers
    0.73
     horm
    0.72
    Activations Density 0.016%

    No Known Activations