© 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. 4899
    Prev
    Next
    INDEX
    Explanations

    instances of being taken or made to do something, often in a context that implies coercion or obligation

    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
    241
    +0.12
    0.5%
    208
    +0.10
    0.5%
    624
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    665
    +0.12
    0.10
    469
    +0.10
    0.06
    70
    +0.10
    0.07
    Negative Logits
     patented
    -0.70
     Forge
    -0.68
     itself
    -0.65
     SetTextColor
    -0.65
     pecul
    -0.60
    dri
    -0.59
     occurring
    -0.59
     ours
    -0.59
     customary
    -0.58
     uttered
    -0.57
    POSITIVE LOGITS
    involved
    0.99
    leep
    0.86
     acquainted
    0.83
     asleep
    0.82
     homeless
    0.81
     interested
    0.81
    opausal
    0.81
     victimized
    0.80
    interested
    0.80
    angu
    0.79
    Activations Density 0.596%

    No Known Activations