© 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. 5270
    Prev
    Next
    INDEX
    Explanations

    negations and expressions of uncertainty or impossibility

    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
    720
    +0.12
    0.6%
    690
    +0.10
    0.5%
    393
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    643
    +0.12
    0.05
    383
    +0.10
    0.06
    307
    +0.10
    0.06
    Negative Logits
    incent
    -0.80
    sic
    -0.73
    Russ
    -0.73
    massive
    -0.72
    vacc
    -0.72
    kees
    -0.70
    razy
    -0.70
    chlor
    -0.69
    iox
    -0.69
    ccess
    -0.68
    POSITIVE LOGITS
     scorn
    0.77
     disse
    0.71
     imagination
    0.71
     legislator
    0.70
     deceived
    0.69
     foreigner
    0.67
     pleasures
    0.67
     pity
    0.67
     leisure
    0.67
     imag
    0.66
    Activations Density 0.402%

    No Known Activations