© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW 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. 9066
    Prev
    Next
    INDEX
    Explanations

    phrases emphasizing key points or main ideas in a discourse

    oai_token-act-pair · gpt-4o-mini
    Triggered 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
    Neuron Alignment
    Index
    Value
    % of L₁
    616
    +0.10
    0.5%
    364
    +0.09
    0.4%
    418
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    362
    +0.10
    0.07
    608
    +0.09
    0.06
    393
    +0.09
    0.05
    Negative Logits
    OWS
    -0.63
    ividual
    -0.62
    ylon
    -0.61
    incinn
    -0.60
    efer
    -0.60
     Leilan
    -0.60
    IENT
    -0.60
    CENT
    -0.58
    rn
    -0.58
    isco
    -0.57
    POSITIVE LOGITS
     implication
    1.10
     gist
    1.06
     takeaway
    1.03
     lesson
    0.98
     reason
    0.95
     extent
    0.94
     nature
    0.92
     hallmark
    0.91
     irony
    0.91
     motto
    0.87
    Activations Density 0.369%

    No Known Activations