INDEX

Explanations

names of individuals in text

oai_token-act-pair · gpt-3.5-turbo

mentions of specific individuals, likely authority figures or experts, in the context of providing quotes or insights

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 revenge

-0.67

tumblr

-0.64

 vigilante

-0.63

 diaper

-0.62

KKK

-0.62

 racist

-0.61

ĸļ

-0.59

 reincarn

-0.59

 youtube

-0.59

*/(

-0.58

POSITIVE LOGITS

endor

0.74

ansky

0.71

hani

0.71

lett

0.71

mann

0.71

rup

0.70

enda

0.69

owsky

0.69

elli

0.68

patrick

0.68

Activations Density 0.518%

names of individuals in text

mentions of specific individuals, likely authority figures or experts, in the context of providing quotes or insights

No Comments

No Known Activations