© 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. 5304
    Prev
    Next
    INDEX
    Explanations

    future tense auxiliary verbs and phrases indicating intention or prediction

    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
    581
    +0.13
    0.6%
    606
    +0.12
    0.6%
    415
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    551
    +0.13
    0.06
    49
    +0.12
    0.06
    566
    +0.10
    0.05
    Negative Logits
    arin
    -0.69
    nery
    -0.62
    iful
    -0.61
     Bird
    -0.61
    jriwal
    -0.60
    ertodd
    -0.60
    guards
    -0.59
     airs
    -0.58
     hawk
    -0.57
     Tok
    -0.57
    POSITIVE LOGITS
     notice
    0.94
    onde
    0.84
     encount
    0.83
    noticed
    0.83
     noticed
    0.82
     encounter
    0.82
     need
    0.79
     hear
    0.78
     see
    0.75
     encountering
    0.75
    Activations Density 0.200%

    No Known Activations