INDEX

Explanations

phrases containing 'turned', indicating a change in role or status

oai_token-act-pair · gpt-3.5-turbo

words related to transformations or changes in state

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

Zup

-0.72

 Archdemon

-0.67

 tuple

-0.66

IRD

-0.66

 ital

-0.66

 listeners

-0.65

ICAL

-0.64

 WARN

-0.63

IBLE

-0.63

STR

-0.62

POSITIVE LOGITS

cum

3.62

turned

2.70

bum

1.94

enture

1.01

cele

0.97

Ã¦

0.93

lum

0.91

}{

0.90

mud

0.86

ensis

0.85

Activations Density 0.040%

phrases containing 'turned', indicating a change in role or status

words related to transformations or changes in state

No Comments

No Known Activations