INDEX

Explanations

phrases indicating extremes or superlatives, such as "the most," "the best," or "the worst."

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate a ranking or comparison among entities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

tera

-1.05

krit

-0.83

kamp

-0.80

miah

-0.78

Ô

-0.76

matter

-0.75

gow

-0.74

ueller

-0.74

riber

-0.72

 unfocusedRange

-0.70

POSITIVE LOGITS

 highs

0.85

 sorts

0.84

 evils

0.83

 lows

0.79

 earners

0.77

 pursuits

0.77

 entrants

0.73

kin

0.73

 manners

0.72

 partners

0.72

Activations Density 0.069%

phrases indicating extremes or superlatives, such as "the most," "the best," or "the worst."

phrases that indicate a ranking or comparison among entities

No Comments

No Known Activations