INDEX

Explanations

phrases related to political discussions or debate

oai_token-act-pair · gpt-3.5-turbo

references to "the" in various contexts, particularly related to subjects in discussions

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

icone

-0.76

riott

-0.73

thouse

-0.73

VERT

-0.72

strom

-0.70

esters

-0.68

rays

-0.68

ãĥīãĥ©

-0.68

appointed

-0.67

AppData

-0.67

POSITIVE LOGITS

 topic

1.42

 subject

1.40

 matter

1.23

eve

1.19

 merits

1.07

 issue

0.99

 sidelines

0.99

 basis

0.97

 behalf

0.95

 heels

0.94

Activations Density 0.120%

phrases related to political discussions or debate

references to "the" in various contexts, particularly related to subjects in discussions

No Comments

No Known Activations