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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Apollo Research · Taylor · Identifying Functionally Important Features with End-to-End Sparse Dictionary Learning
    3. GPT2-Small
    4. Residual Similar CE Local
    5. 6-RES_SCL-AJT
    6. 45233
    Prev
    Next
    INDEX
    Explanations

    the verb 'stumble' and variations, as well as related concepts like 'surges' and 'havoc'

    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
    447
    +0.22
    1.1%
    418
    +0.10
    0.5%
    267
    +0.09
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    326
    +0.22
    0.24
    288
    +0.10
    0.22
    464
    +0.09
    0.09
    Negative Logits
    venants
    -0.62
    ators
    -0.60
    thood
    -0.59
    tan
    -0.59
    hips
    -0.59
    essee
    -0.59
    scl
    -0.58
    selage
    -0.58
    SHIP
    -0.58
    urers
    -0.58
    POSITIVE LOGITS
    weed
    0.83
     inco
    0.76
     scramble
    0.66
     shuffle
    0.65
    bum
    0.64
    buck
    0.63
    dale
    0.61
    uffle
    0.61
     err
    0.61
    own
    0.61
    Activations Density 10.730%

    No Known Activations