INDEX

Explanations

forms of the word "wit"

oai_token-act-pair · gpt-3.5-turbo

(the word) "wit" and its variations in different contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

deen

-0.76

ibaba

-0.73

antha

-0.71

igree

-0.70

transfer

-0.67

 obligated

-0.67

duction

-0.66

Ü

-0.64

avorite

-0.63

prone

-0.63

POSITIVE LOGITS

wit

1.33

chery

1.11

glers

0.96

ivities

0.88

hered

0.87

 gust

0.83

wit

0.80

nesses

0.79

 Beir

0.75

vity

0.74

Activations Density 0.005%

forms of the word "wit"

(the word) "wit" and its variations in different contexts

No Comments

No Known Activations