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

    Neuronpedia

    APICircuit 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. 34497
    Prev
    Next
    INDEX
    Explanations

    creators or authors of various works and projects

    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
    +0.16
    1.2%
    167
    +0.07
    0.5%
    360
    +0.06
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    679
    +0.16
    0.19
    294
    +0.07
    0.04
    320
    +0.06
    0.06
    Negative Logits
    ishops
    -0.42
    olit
    -0.40
     Devils
    -0.39
    ibaba
    -0.39
     seas
    -0.39
     Himal
    -0.38
    olk
    -0.38
     Downs
    -0.38
    phia
    -0.38
     Stain
    -0.37
    POSITIVE LOGITS
    paces
    0.56
     creator
    0.51
    ãĤº
    0.50
    creator
    0.50
    cember
    0.47
    forge
    0.46
    ãĥķãĤ¡
    0.45
     creators
    0.45
    caster
    0.45
    ãĤ´
    0.45
    Activations Density 9.426%

    No Known Activations