INDEX

Explanations

phrases related to actions or behaviors being done by a specific subject

oai_token-act-pair · gpt-3.5-turbo

phrases related to media criticism and accountability

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ometer

-0.75

é»Ĵ

-0.73

externalActionCode

-0.71

hiba

-0.71

ŃĶ

-0.67

isl

-0.67

neau

-0.66

ģ«

-0.66

wagen

-0.66

EPA

-0.66

POSITIVE LOGITS

 itself

1.02

its

0.91

Its

0.87

me

0.84

us

0.84

 dudes

0.83

ITS

0.79

 gays

0.76

 kids

0.74

 thugs

0.74

Activations Density 0.725%

phrases related to actions or behaviors being done by a specific subject

phrases related to media criticism and accountability

No Comments

No Known Activations