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. 13553
    Prev
    Next
    INDEX
    Explanations

    expressions of affection or emotional connection

    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
    276
    +0.20
    1.5%
    274
    +0.15
    1.2%
    62
    +0.15
    1.1%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    326
    +0.20
    0.01
    343
    +0.15
    0.02
    447
    +0.15
    0.01
    Negative Logits
    quartered
    -0.78
     angered
    -0.66
     trapped
    -0.65
    podcast
    -0.64
     attracted
    -0.64
     enraged
    -0.64
     Opera
    -0.64
     hol
    -0.63
     stricken
    -0.61
     Tart
    -0.61
    POSITIVE LOGITS
     lapt
    0.94
    hovah
    0.92
    amins
    0.91
    inkle
    0.90
    ï¸ı
    0.90
     pse
    0.89
     cannabin
    0.87
    elong
    0.82
    icka
    0.81
    trak
    0.81
    Activations Density 0.029%

    No Known Activations