© 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. 4770
    Prev
    Next
    INDEX
    Explanations

    mentions of wildlife and regulations regarding dangerous animals

    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
    398
    +0.11
    0.5%
    264
    +0.11
    0.5%
    483
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    393
    +0.11
    0.10
    294
    +0.11
    0.04
    264
    +0.10
    0.02
    Negative Logits
    inventoryQuantity
    -0.86
     reinvest
    -0.80
     Degree
    -0.76
     resign
    -0.74
     refund
    -0.73
    antage
    -0.72
     Savings
    -0.72
     Investment
    -0.71
     prest
    -0.70
    Timeout
    -0.70
    POSITIVE LOGITS
     animals
    1.62
     reptiles
    1.60
     humans
    1.60
     mammals
    1.60
     apes
    1.57
     primates
    1.51
     rodents
    1.50
    humans
    1.49
     goats
    1.42
     Humans
    1.41
    Activations Density 1.659%

    No Known Activations