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

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Open Source Sparse Autoencoders for all Residual Stream Layers of GPT2-Small
    3. GPT2-Small
    4. Residual Stream
    5. 2-RES-JB
    6. 175
    Prev
    Next
    INDEX
    Explanations

    the word "aggravated" in a legal or criminal context

    oai_token-act-pair · gpt-4-turbo

    strong negative actions or crimes

    oai_token-act-pair · gpt-3.5-turbo

    terms related to violent or serious crimes

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 2-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.hook_resid_pre
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    24,576
    Data Type
    torch.float32
    Hook Point
    blocks.2.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    2
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    atch
    -0.91
    orrow
    -0.90
    tower
    -0.90
    sonian
    -0.86
    oi
    -0.83
    oppers
    -0.81
    icians
    -0.80
    ether
    -0.80
    tz
    -0.79
    enta
    -0.79
    POSITIVE LOGITS
     aggrav
    1.05
     aggravated
    0.93
     indecent
    0.71
     offenses
    0.71
     offence
    0.69
     offenders
    0.69
     assaults
    0.68
     exacerb
    0.68
     offences
    0.68
     offending
    0.67
    Activations Density 0.018%

    No Known Activations