© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    3. OpenAI · Sparse Autoencoder for GPT2-Small - v5 2024
    5. GPT2-Small
    7. Residual Post - 32k
    9. 8-RES_POST_32K-OAI
    11. 4760
    Prev
    Next
    INDEX
    Explanations

    phrases pertaining to social issues and their impacts on communities

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_post_32k-oai
    Configuration
    jbloom/GPT2-Small-OAI-v5-32k-resid-post-SAEs/v5_32k_layer_8.pt
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    32,768
    Data Type
    torch.float32
    Hook Name
    blocks.8.hook_resid_post
    Hook Layer
    8
    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
    86
    +0.12
    0.5%
    396
    +0.11
    0.5%
    400
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    362
    +0.12
    0.14
    608
    +0.11
    0.12
    393
    +0.10
    0.10
    Negative Logits
     toast
    -0.85
     farewell
    -0.69
     exclaim
    -0.68
     Ended
    -0.66
    married
    -0.64
     Minutes
    -0.62
     cakes
    -0.62
    ittal
    -0.62
     cured
    -0.61
     depended
    -0.60
    POSITIVE LOGITS
     backlog
    0.99
     scourge
    0.93
     pervasive
    0.88
     misinformation
    0.88
     rampant
    0.83
     imbalance
    0.83
     misconceptions
    0.82
     disparities
    0.82
     dele
    0.81
     escalating
    0.81
    Activations Density 1.313%

    No Known Activations