INDEX

Explanations

contractions indicating negative sentiment or disbelief

oai_token-act-pair · gpt-3.5-turbo

phrases expressing uncertainty or conditionality

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

ishing

-0.69

arer

-0.64

«ĺ

-0.62

ÃŁ

-0.62

acca

-0.61

assing

-0.60

Shell

-0.59

active

-0.59

 Cosponsors

-0.58

Tik

-0.58

POSITIVE LOGITS

ĸļ

0.78

enance

0.73

uce

0.70

 tumble

0.69

clinton

0.68

tarians

0.67

 offend

0.65

arez

0.63

 sooner

0.63

yip

0.62

Activations Density 0.219%

contractions indicating negative sentiment or disbelief

phrases expressing uncertainty or conditionality

No Comments

No Known Activations