INDEX

Explanations

news-related terms or information such as articles, posts, stories, and agreements

oai_token-act-pair · gpt-3.5-turbo

references to original articles or sources of information

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

ileged

-0.72

perty

-0.72

rom

-0.68

ensitive

-0.68

cipled

-0.67

idious

-0.65

cellaneous

-0.63

indust

-0.62

among

-0.62

rome

-0.62

POSITIVE LOGITS

ãĤ¨ãĥ«

0.78

ItemImage

0.78

SpaceEngineers

0.72

appa

0.66

æī

0.64

 guiActiveUnfocused

0.63

ãĥĬ

0.63

ãĥ¼ãĤ¯

0.62

 Dunham

0.61

 apologise

0.60

Activations Density 0.138%

news-related terms or information such as articles, posts, stories, and agreements

references to original articles or sources of information

No Comments

No Known Activations