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. 4402
    Prev
    Next
    INDEX
    Explanations

    mentions of personal experiences and professional accomplishments

    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
    +0.28
    1.8%
    362
    +0.12
    0.7%
    393
    +0.11
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    15
    +0.28
    0.20
    745
    +0.12
    -0.15
    362
    +0.11
    0.15
    Negative Logits
    eers
    -0.59
    netflix
    -0.57
    river
    -0.55
     Uriel
    -0.53
    hyde
    -0.53
    ibaba
    -0.52
    phans
    -0.50
    ÑĮ
    -0.50
    Tree
    -0.50
    ahead
    -0.49
    POSITIVE LOGITS
     own
    1.10
     intentions
    0.85
     predicament
    0.82
     involvement
    0.80
     inability
    0.80
     accomplishments
    0.79
     efforts
    0.79
     plight
    0.79
     newfound
    0.78
     actions
    0.76
    Activations Density 14.996%

    No Known Activations