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. 13550
    Prev
    Next
    INDEX
    Explanations

    references to conditions, particularly in a mathematical or technical context

    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
    311
    +0.20
    1.3%
    679
    +0.20
    1.3%
    128
    +0.17
    1.1%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    116
    +0.20
    0.01
    739
    +0.20
    0.01
    700
    +0.17
    0.01
    Negative Logits
    çīĪ
    -0.79
    cles
    -0.78
    avering
    -0.74
    GES
    -0.74
    NING
    -0.73
    erenn
    -0.73
     Alc
    -0.73
    hner
    -0.73
    angler
    -0.72
    azz
    -0.72
    POSITIVE LOGITS
     Palestin
    1.03
    xual
    1.01
    emort
    0.97
    namese
    0.96
    icipated
    0.92
    anwhile
    0.88
    terday
    0.86
    comings
    0.85
     disadvant
    0.84
     coerc
    0.84
    Activations Density 0.009%

    No Known Activations