INDEX

Explanations

instances of the abbreviation "DO" in the text

oai_token-act-pair · gpt-3.5-turbo

instances of the acronym "DO" and variations of the word "do"

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

Reviewer

-0.85

 Handling

-0.74

Pwr

-0.73

wagen

-0.72

ulations

-0.68

ulated

-0.68

ONSORED

-0.68

 Subtle

-0.67

ulating

-0.66

itatively

-0.66

POSITIVE LOGITS

ppel

1.07

omsday

1.00

herty

0.99

oms

0.96

zens

0.94

ctr

0.93

ctors

0.90

omed

0.86

lez

0.85

Í

0.85

Activations Density 0.042%

instances of the abbreviation "DO" in the text

instances of the acronym "DO" and variations of the word "do"

No Comments

No Known Activations