INDEX

Explanations

mentions urging the reader to check out specific content

oai_token-act-pair · gpt-3.5-turbo

references to selected recommendations or curated content

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

eg

-0.71

lich

-0.70

ATIONAL

-0.68

 interf

-0.68

OPS

-0.67

adr

-0.65

REL

-0.64

lit

-0.64

million

-0.64

lia

-0.64

POSITIVE LOGITS

 Picks

1.95

ļéĨĴ

1.13

ertodd

0.97

pocket

0.92

enser

0.91

 Annotations

0.87

 Pick

0.85

 picks

0.83

oldown

0.80

daq

0.79

Activations Density 0.004%

mentions urging the reader to check out specific content

references to selected recommendations or curated content

No Comments

No Known Activations