© 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. 5542
    Prev
    Next
    INDEX
    Explanations

    conditional phrases or statements

    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
    527
    +0.12
    0.6%
    479
    +0.11
    0.5%
    200
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    551
    +0.12
    0.07
    139
    +0.11
    0.07
    49
    +0.11
    0.07
    Negative Logits
    soon
    -0.99
     eventually
    -0.97
     soon
    -0.89
     continue
    -0.86
     Soon
    -0.86
     be
    -0.85
     retaliate
    -0.85
     tolerate
    -0.82
     someday
    -0.82
     disapprove
    -0.81
    POSITIVE LOGITS
    HQ
    0.69
     Meaning
    0.66
     Worth
    0.64
     sucks
    0.60
     suggests
    0.60
    ENDED
    0.59
     emphasizes
    0.58
    ractical
    0.58
     indicates
    0.58
     fits
    0.58
    Activations Density 0.130%

    No Known Activations