© 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. 10-RES_POST_32K-OAI
    11. 32072
    Prev
    Next
    INDEX
    Explanations

    descriptive adjectives and phrases that evoke vivid imagery

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 10-res_post_32k-oai
    Configuration
    jbloom/GPT2-Small-OAI-v5-32k-resid-post-SAEs/v5_32k_layer_10.pt
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    32,768
    Data Type
    torch.float32
    Hook Name
    blocks.10.hook_resid_post
    Hook Layer
    10
    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
    748
    +0.12
    0.6%
    321
    +0.12
    0.5%
    290
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.12
    0.13
    374
    +0.12
    -0.02
    251
    +0.10
    0.05
    Negative Logits
    Ability
    -0.78
    TION
    -0.76
     alot
    -0.74
    channelAvailability
    -0.74
    Individual
    -0.74
    Specific
    -0.69
    rosso
    -0.69
     whilst
    -0.66
     persons
    -0.66
     undesirable
    -0.66
    POSITIVE LOGITS
     mansion
    0.82
     Gothic
    0.81
     wooden
    0.79
     Plymouth
    0.78
    ilded
    0.78
     granite
    0.77
     limestone
    0.76
     Colonial
    0.75
     mah
    0.74
     paperback
    0.74
    Activations Density 1.853%

    No Known Activations