© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI 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. 5751
    Prev
    Next
    INDEX
    Explanations

    evaluative adjectives expressing positive or negative feelings towards situations or events

    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
    479
    +0.12
    0.5%
    506
    +0.09
    0.4%
    602
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    271
    +0.12
    0.06
    737
    +0.09
    0.07
    42
    +0.09
    0.08
    Negative Logits
    otte
    -0.81
    lete
    -0.80
    racuse
    -0.78
    Downloadha
    -0.77
    ividual
    -0.75
    iler
    -0.74
    awed
    -0.74
    velop
    -0.72
    aneers
    -0.71
    eatured
    -0.71
    POSITIVE LOGITS
     territory
    1.07
     advice
    1.01
     madness
    1.01
     insanity
    0.98
     stuff
    0.96
     journalism
    0.93
     misinformation
    0.93
     stupidity
    0.91
     terminology
    0.90
     behavior
    0.89
    Activations Density 0.735%

    No Known Activations