INDEX

Explanations

the word "entirely"

oai_token-act-pair · gpt-3.5-turbo

the phrase "entirely" followed by strong descriptors or modifiers in various contexts

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

 ANGEL

-0.72

hao

-0.69

liest

-0.68

former

-0.68

 commissioner

-0.67

 Commissioner

-0.65

LI

-0.65

 Millennium

-0.64

 Pressure

-0.64

ulators

-0.64

POSITIVE LOGITS

 reliant

0.93

 devoted

0.74

BuyableInstoreAndOnline

0.74

 dependent

0.74

 composed

0.72

 depends

0.71

 accounted

0.70

 comprised

0.68

obin

0.68

 separate

0.66

Activations Density 0.011%

the word "entirely"

the phrase "entirely" followed by strong descriptors or modifiers in various contexts

No Comments

No Known Activations