INDEX

Explanations

acronyms and codes related to a specific topic or system, possibly related to a specific organization or field

oai_token-act-pair · gpt-3.5-turbo

references to numerical scores and identifiers in a competitive context

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

rand

-0.81

Wem

-0.76

lements

-0.76

etsk

-0.75

 guiActiveUn

-0.71

tein

-0.71

 Zimmer

-0.71

uton

-0.70

css

-0.70

Gro

-0.69

POSITIVE LOGITS

NA

2.99

TBA

2.03

TBD

1.82

 Daly

1.40

NR

1.39

NX

1.21

 Unknown

1.17

 TRUE

1.14

LC

1.11

NAS

1.11

Activations Density 0.048%

acronyms and codes related to a specific topic or system, possibly related to a specific organization or field

references to numerical scores and identifiers in a competitive context

No Comments

No Known Activations