© 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. Apollo Research · Taylor · Identifying Functionally Important Features with End-to-End Sparse Dictionary Learning
    5. GPT2-Small
    7. Residual Similar CE Local
    9. 6-RES_SCL-AJT
    11. 34046
    Prev
    Next
    INDEX
    Explanations

    unexpected outcomes or events in a narrative

    oai_token-act-pair · gpt-3.5-turbo
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    neuronpedia/gpt2-small__res_scl-ajt/6-res_scl-ajt
    Prompts (Dashboard)
    12,288 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    46,080
    Data Type
    torch.float32
    Hook Point
    blocks.6.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    apollo-research/Skylion007-openwebtext-tokenizer-gpt2
    Hook Point Layer
    6
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    675
    +0.13
    0.6%
    447
    +0.12
    0.5%
    228
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    393
    +0.13
    0.27
    608
    +0.12
    0.20
    283
    +0.11
    0.03
    Negative Logits
    onics
    -0.81
    anti
    -0.72
    mob
    -0.66
    ore
    -0.66
    onomy
    -0.66
    aim
    -0.65
    orean
    -0.65
    ove
    -0.64
    oug
    -0.64
    ismo
    -0.62
    POSITIVE LOGITS
     refill
    0.78
     reapp
    0.74
     beg
    0.73
     realize
    0.73
     realise
    0.72
    LESS
    0.72
     abruptly
    0.71
     remind
    0.70
    ersen
    0.70
     subsequently
    0.69
    Activations Density 9.797%

    No Known Activations