INDEX

Explanations

locations or objects positioned on top of something else

oai_token-act-pair · gpt-3.5-turbo

instances of the word "atop" and related spatial positioning terms

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

nery

-0.81

gm

-0.79

details

-0.74

cci

-0.72

uld

-0.72

uum

-0.70

ibr

-0.69

deal

-0.69

lich

-0.69

len

-0.69

POSITIVE LOGITS

 atop

0.86

ĺħ

0.86

İĭ

0.76

ĪĴ

0.72

ĨĴ

0.70

 heights

0.67

mast

0.65

 podium

0.64

hap

0.64

ĸļ

0.63

Activations Density 0.004%

locations or objects positioned on top of something else

instances of the word "atop" and related spatial positioning terms

No Comments

No Known Activations