INDEX

Explanations

references to being the first time something has happened

oai_token-act-pair · gpt-3.5-turbo

instances of phrases indicating something is not the first occurrence

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

'[

-0.68

 ONLY

-0.67

 Tsukuyomi

-0.67

externalActionCode

-0.67

[[

-0.65

([

-0.65

resso

-0.64

maximum

-0.63

only

-0.62

glomer

-0.61

POSITIVE LOGITS

 woes

0.81

 lately

0.76

 plag

0.76

 controversy

0.74

 problematic

0.73

 controversies

0.73

 troubled

0.73

 troubling

0.72

 politic

0.70

 scandals

0.70

Activations Density 0.363%

references to being the first time something has happened

instances of phrases indicating something is not the first occurrence

No Comments

No Known Activations