© 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. 7895
    Prev
    Next
    INDEX
    Explanations

    references to health-related issues and their consequences

    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
    648
    +0.13
    0.6%
    654
    +0.10
    0.5%
    567
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    362
    +0.13
    0.12
    608
    +0.10
    0.10
    393
    +0.10
    0.08
    Negative Logits
    quer
    -0.83
    ipop
    -0.81
     Powder
    -0.59
    soever
    -0.58
     Glover
    -0.58
     speeches
    -0.57
     Scream
    -0.57
     whichever
    -0.57
     guesses
    -0.57
     answering
    -0.57
    POSITIVE LOGITS
     importance
    1.56
     impossibility
    1.22
     necessity
    1.20
     usefulness
    1.19
     superiority
    1.17
     significance
    1.16
     dangers
    1.15
     limitations
    1.08
     virtues
    1.08
     shortcomings
    1.07
    Activations Density 0.854%

    No Known Activations