© 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
    2. OpenAI · Sparse Autoencoder for GPT2-Small - v5 2024
    3. GPT2-Small
    4. Residual Post - 32k
    5. 8-RES_POST_32K-OAI
    6. 5337
    Prev
    Next
    INDEX
    Explanations

    phrases indicating needs or desires for specific attributes in products or people

    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
    556
    +0.09
    0.4%
    307
    +0.09
    0.4%
    396
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    551
    +0.09
    0.08
    139
    +0.09
    0.08
    300
    +0.09
    0.07
    Negative Logits
     doubtless
    -0.67
     occurring
    -0.66
     unexplained
    -0.64
     Origins
    -0.64
    NetMessage
    -0.62
     Parenthood
    -0.62
     wondering
    -0.61
     batches
    -0.61
     beginnings
    -0.59
    */(
    -0.58
    POSITIVE LOGITS
     conform
    0.86
     compliant
    0.86
     adequate
    0.84
     adequately
    0.83
    clud
    0.82
     equitable
    0.81
     suitable
    0.81
     realistic
    0.80
     truthful
    0.80
    atible
    0.79
    Activations Density 0.821%

    No Known Activations