© 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. 4904
    Prev
    Next
    INDEX
    Explanations

    negative content that includes threats, violence, and abuse towards others

    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
    51
    +0.10
    0.5%
    86
    +0.09
    0.4%
    711
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    373
    +0.10
    -0.02
    211
    +0.09
    0.01
    13
    +0.09
    0.03
    Negative Logits
    terday
    -0.89
     awa
    -0.68
     rehearsal
    -0.68
     envy
    -0.67
     hump
    -0.64
     resil
    -0.62
     gren
    -0.62
    atform
    -0.61
     wip
    -0.61
    ãĥīãĥ©ãĤ´ãĥ³
    -0.61
    POSITIVE LOGITS
    raising
    0.71
    adas
    0.70
    endar
    0.68
     MFT
    0.65
    ouse
    0.64
    channelAvailability
    0.62
    assadors
    0.62
     DEFENSE
    0.61
    cellaneous
    0.61
     totaled
    0.61
    Activations Density 0.230%

    No Known Activations