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

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI 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. 4229
    Prev
    Next
    INDEX
    Explanations

    instances of transitional phrases indicating contradiction or exception

    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
    224
    +0.11
    0.5%
    547
    +0.11
    0.5%
    244
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    561
    +0.11
    0.07
    393
    +0.11
    0.09
    547
    +0.11
    0.05
    Negative Logits
    official
    -0.66
    Ü
    -0.65
     ðŁĺ
    -0.64
     Massacre
    -0.63
     ILCS
    -0.63
     Notice
    -0.62
    MAS
    -0.60
     EntityItem
    -0.60
    arantine
    -0.59
    fuck
    -0.58
    POSITIVE LOGITS
     added
    1.07
     noting
    1.07
     adds
    1.03
     cautioned
    1.02
     urged
    0.99
     stressed
    0.94
     adding
    0.93
     contends
    0.93
     believes
    0.93
     concedes
    0.92
    Activations Density 0.888%

    No Known Activations