INDEX

Explanations

narrow or specific terms

oai_token-act-pair · gpt-3.5-turbo

instances of the word "narrow" or related terms that convey narrowness

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

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

Not in Any Lists

No Comments

Negative Logits

ICAN

-0.78

CF

-0.75

YL

-0.74

FORE

-0.73

IRED

-0.73

natureconservancy

-0.71

 Destruction

-0.70

 Mobil

-0.69

è¦ļéĨĴ

-0.68

NSA

-0.68

POSITIVE LOGITS

 minded

0.98

 narrow

0.94

sided

0.93

creen

0.93

sighted

0.85

 confines

0.84

minded

0.83

cut

0.82

band

0.82

 strokes

0.81

Activations Density 0.019%

narrow or specific terms

instances of the word "narrow" or related terms that convey narrowness

No Comments

No Known Activations