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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports 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. 5533
    Prev
    Next
    INDEX
    Explanations

    elements of poetic structure and imagery

    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
    +0.12
    0.5%
    179
    +0.10
    0.5%
    48
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    160
    +0.12
    0.06
    107
    +0.10
    0.06
    442
    +0.10
    0.04
    Negative Logits
    ãĢij
    -0.68
    perty
    -0.66
     significantly
    -0.65
    ptions
    -0.64
     substantially
    -0.64
     stated
    -0.63
     favorably
    -0.63
    ability
    -0.62
     attrition
    -0.62
    ÄŁ
    -0.60
    POSITIVE LOGITS
    And
    1.00
    Everybody
    0.97
    Cause
    0.96
     Till
    0.88
     Where
    0.87
    Tell
    0.87
    Yeah
    0.87
    Where
    0.86
     And
    0.83
     Somebody
    0.83
    Activations Density 0.152%

    No Known Activations