© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Open Source Sparse Autoencoders for all Residual Stream Layers of GPT2-Small
    5. GPT2-Small
    7. Residual Stream
    9. 4-RES-JB
    11. 6716
    Prev
    Next
    INDEX
    Explanations

    words related to the process of becoming softer or more lenient

    oai_token-act-pair · gpt-3.5-turbo

    terms related to changes in intensity or firmness, particularly focusing on the concepts of softening and hardening

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 4-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    4
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    Discussion
    -0.75
    args
    -0.73
    Choice
    -0.73
    arded
    -0.70
    rast
    -0.70
    WHERE
    -0.70
    ination
    -0.69
    iop
    -0.69
    place
    -0.67
    Champ
    -0.67
    POSITIVE LOGITS
    xual
    0.92
     soften
    0.88
     misunder
    0.81
    terness
    0.79
     streng
    0.78
    ignt
    0.74
    laus
    0.73
     softened
    0.72
     bom
    0.72
     confir
    0.72
    Activations Density 0.030%

    No Known Activations