INDEX

Explanations

phrases related to breaking or dividing something into chunks or pieces

oai_token-act-pair · gpt-3.5-turbo

terms related to component breakdown and categorization in systems

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

atem

-0.71

apest

-0.69

kered

-0.64

MIA

-0.61

 alike

-0.61

stice

-0.59

Brand

-0.59

erity

-0.59

athi

-0.58

quickShipAvailable

-0.57

POSITIVE LOGITS

livion

1.23

 bloodstream

0.90

gil

0.82

 fray

0.75

 mode

0.72

 realms

0.71

ongh

0.71

 orbit

0.70

safe

0.70

 realm

0.69

Activations Density 0.354%

phrases related to breaking or dividing something into chunks or pieces

terms related to component breakdown and categorization in systems

No Comments

No Known Activations