INDEX

Explanations

terms indicating causality or logical consequences

oai_token-act-pair · gpt-3.5-turbo

connective words that indicate causality or consequences

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

terms indicating causality preceded by "and"

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

•conditional activation

only activates when word is preceded by "and"

No Comments

Negative Logits

igger

-0.78

ocene

-0.72

POST

-0.70

IGHT

-0.70

MM

-0.69

 Golf

-0.69

MM

-0.67

ISION

-0.65

IG

-0.63

UD

-0.63

POSITIVE LOGITS

forth

1.11

 guiActiveUn

0.80

é¾įåĸļå£«

0.78

 constitute

0.78

 accounted

0.76

 necess

0.76

 behold

0.75

 prevented

0.74

 subjected

0.74

 proceeded

0.73

Activations Density 0.025%

terms indicating causality or logical consequences

connective words that indicate causality or consequences

terms indicating causality preceded by "and"

No Comments

No Known Activations