INDEX

Explanations

contractions

oai_token-act-pair · gpt-3.5-turbo

negative statements or contractions indicating something is not the case

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

çļ

-0.81

onym

-0.72

ERG

-0.70

 surrounds

-0.68

è»

-0.67

è¦ļéĨĴ

-0.66

obo

-0.66

stre

-0.66

andowski

-0.65

çĶŁ

-0.65

POSITIVE LOGITS

 necessarily

1.14

 exactly

1.11

 quite

0.93

 gonna

0.93

 even

0.87

 bluff

0.86

 bothering

0.85

 EVEN

0.84

 really

0.82

 gotta

0.81

Activations Density 0.090%

contractions

negative statements or contractions indicating something is not the case

No Comments

No Known Activations