INDEX

Explanations

the word "Let" at a high activation level

oai_token-act-pair · gpt-3.5-turbo

phrases that begin with "Let."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

PLIED

-0.77

 holiest

-0.76

?????-

-0.67

existent

-0.66

POSE

-0.66

 cumbers

-0.65

CLASSIFIED

-0.65

Zen

-0.64

oppable

-0.62

INESS

-0.61

POSITIVE LOGITS

icia

1.02

itia

0.87

ting

0.79

tering

0.78

sheet

0.75

hetically

0.75

tered

0.74

us

0.72

iton

0.72

cheon

0.72

Activations Density 0.024%

the word "Let" at a high activation level

phrases that begin with "Let."

No Comments

No Known Activations