INDEX

Explanations

instances of the pronoun "we."

oai_token-act-pair · gpt-3.5-turbo

instances of the word "we."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

him

-0.76

ragon

-0.62

yna

-0.58

emon

-0.55

incial

-0.54

ensed

-0.53

igun

-0.52

Override

-0.52

ounding

-0.51

him

-0.50

POSITIVE LOGITS

we

2.67

we

1.75

We

1.71

We

1.67

our

1.52

 ourselves

1.51

WE

1.40

 ours

1.39

Our

1.24

us

1.18

Activations Density 0.170%

instances of the pronoun "we."

instances of the word "we."

No Comments

No Known Activations