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. 4451
    Prev
    Next
    INDEX
    Explanations

    expressions of frustration or annoyance, such as phew, ugh, oh, and sighs

    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
    441
    +0.07
    0.5%
    485
    +0.06
    0.5%
    166
    +0.06
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    284
    +0.07
    0.09
    8
    +0.06
    0.10
    326
    +0.06
    0.18
    Negative Logits
    iers
    -0.42
    icts
    -0.41
    iership
    -0.39
    jri
    -0.39
    ":[{"
    -0.39
    ifles
    -0.39
    jug
    -0.38
     sanctioned
    -0.38
    forming
    -0.38
    ensis
    -0.38
    POSITIVE LOGITS
    HHHH
    0.55
    hhhh
    0.54
    hhh
    0.52
    hh
    0.49
     goodbye
    0.44
     Clockwork
    0.43
    athe
    0.43
    awk
    0.42
    ouls
    0.41
    yeah
    0.41
    Activations Density 5.800%

    No Known Activations