© 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. 5485
    Prev
    Next
    INDEX
    Explanations

    references to specific characters, identities, or groups within a narrative, particularly in gaming or political contexts

    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
    563
    +0.11
    0.5%
    410
    +0.10
    0.5%
    423
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    383
    +0.11
    0.05
    77
    +0.10
    0.04
    410
    +0.09
    0.04
    Negative Logits
    bye
    -0.70
     Dragonbound
    -0.69
    omy
    -0.67
    },{"
    -0.66
    uts
    -0.65
    maxwell
    -0.65
    lass
    -0.63
    death
    -0.63
    NK
    -0.60
    });
    -0.60
    POSITIVE LOGITS
     being
    1.36
     exception
    1.23
     preferring
    1.22
     seeming
    1.21
     exceptions
    1.17
     needing
    1.14
     becoming
    1.13
     having
    1.12
     appearing
    1.10
     costing
    1.09
    Activations Density 1.901%

    No Known Activations