© 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. 4700
    Prev
    Next
    INDEX
    Explanations

    questions that seek clarification or inquiry about situations and concepts

    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
    374
    +0.11
    0.5%
    250
    +0.10
    0.5%
    329
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    60
    +0.11
    0.06
    287
    +0.10
    0.06
    263
    +0.09
    0.06
    Negative Logits
    oxide
    -0.72
    isse
    -0.69
    agents
    -0.67
    umin
    -0.67
    unte
    -0.67
    Interview
    -0.66
    ocard
    -0.65
    esm
    -0.64
    ruction
    -0.64
    online
    -0.64
    POSITIVE LOGITS
     feasible
    0.81
     feas
    0.79
     appropriate
    0.73
     recourse
    0.73
     footing
    0.72
     embr
    0.72
     best
    0.72
     prudent
    0.70
    Should
    0.68
     adequately
    0.67
    Activations Density 0.581%

    No Known Activations