INDEX

Explanations

the prefix "ne-" followed by a specific sequence of letters

oai_token-act-pair · gpt-3.5-turbo

occurrences of the prefix "ne" related to scientific or technical terms

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

ãĤ·ãĥ£

-0.92

DragonMagazine

-0.90

ãĥ¼ãĥĨãĤ£

-0.80

MODE

-0.78

Ú

-0.76

æ©Ł

-0.75

å§«

-0.72

inho

-0.72

xual

-0.71

hips

-0.69

POSITIVE LOGITS

arest

0.88

ne

0.86

 neut

0.85

uter

0.83

arthed

0.80

braska

0.77

verend

0.74

yth

0.72

ccess

0.71

uron

0.70

Activations Density 0.007%

the prefix "ne-" followed by a specific sequence of letters

occurrences of the prefix "ne" related to scientific or technical terms

No Comments

No Known Activations