© 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. 5448
    Prev
    Next
    INDEX
    Explanations

    expressions of emotional distress, particularly related to crying

    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
    546
    +0.11
    0.5%
    178
    +0.11
    0.5%
    215
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    730
    +0.11
    0.03
    70
    +0.11
    0.02
    178
    +0.10
    0.03
    Negative Logits
    POL
    -0.75
    OPA
    -0.70
    cius
    -0.69
    MG
    -0.69
    olitics
    -0.68
    escal
    -0.68
    rank
    -0.67
    nesota
    -0.66
    ratulations
    -0.66
    vernment
    -0.66
    POSITIVE LOGITS
     tears
    0.95
     hyster
    0.92
     cried
    0.88
     uncontroll
    0.83
     crying
    0.83
     Tears
    0.83
    baby
    0.81
     weeping
    0.80
     crocod
    0.75
    bed
    0.74
    Activations Density 0.125%

    No Known Activations