Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    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. 5131
    Prev
    Next
    INDEX
    Explanations

    identifiers and actions related to subscription and engagement with content

    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
    365
    +0.10
    0.4%
    126
    +0.09
    0.4%
    666
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    759
    +0.10
    0.06
    182
    +0.09
    0.06
    563
    +0.09
    0.06
    Negative Logits
     hence
    -0.79
     whereas
    -0.78
     secondly
    -0.78
     permitting
    -0.72
     otherwise
    -0.68
     thereby
    -0.64
     allowing
    -0.62
    Footnote
    -0.61
     thus
    -0.60
     urgently
    -0.59
    POSITIVE LOGITS
    ¹
    0.68
    tek
    0.65
    text
    0.62
    land
    0.60
    notation
    0.58
    rah
    0.58
    dy
    0.57
    ileged
    0.56
    ³
    0.56
    lett
    0.56
    Activations Density 0.519%

    No Known Activations