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

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI 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. 4701
    Prev
    Next
    INDEX
    Explanations

    descriptors related to music and its emotional or stylistic qualities

    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
    750
    +0.08
    0.4%
    342
    +0.08
    0.4%
    18
    +0.08
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.08
    0.11
    151
    +0.08
    0.09
    453
    +0.08
    0.07
    Negative Logits
    osponsors
    -0.78
     Preferred
    -0.76
    tax
    -0.76
     Badge
    -0.75
    share
    -0.74
    gat
    -0.73
    eligible
    -0.72
    ewitness
    -0.71
    culosis
    -0.71
    legates
    -0.70
    POSITIVE LOGITS
     realism
    1.34
     thriller
    1.14
     satire
    1.14
     humor
    1.06
     comedy
    1.06
     humour
    1.04
     imagery
    1.00
     elegance
    1.00
     scenes
    0.99
     storytelling
    0.97
    Activations Density 0.992%

    No Known Activations