© 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 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. 6922
    Prev
    Next
    INDEX
    Explanations

    references to specific geographical locations and conflicts in the Middle East

    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
    373
    +0.16
    0.7%
    452
    +0.11
    0.5%
    185
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.16
    0.05
    201
    +0.11
    0.04
    757
    +0.11
    0.03
    Negative Logits
    velength
    -0.76
    cohol
    -0.73
    xon
    -0.73
    asy
    -0.72
    ynt
    -0.72
    DNA
    -0.71
    oin
    -0.70
    zsche
    -0.69
    Magikarp
    -0.69
    cham
    -0.68
    POSITIVE LOGITS
     Province
    0.96
     peninsula
    0.95
     ravaged
    0.91
     devastated
    0.90
     Peninsula
    0.87
     province
    0.84
     Strip
    0.84
     liberated
    0.84
     neighbourhoods
    0.78
     countryside
    0.78
    Activations Density 0.237%

    No Known Activations