INDEX

Explanations

phrases related to military operations or units

oai_token-act-pair · gpt-3.5-turbo

references to "Bomber Command" and related military terminology

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

 Bundy

-0.89

Utah

-0.74

itably

-0.72

æł

-0.71

ibly

-0.70

LY

-0.69

Redditor

-0.69

-0.68

POSITIVE LOGITS

Bom

1.02

bsite

0.92

umin

0.85

veyard

0.81

zeb

0.79

cock

0.75

cham

0.75

endon

0.75

aim

0.75

alore

0.74

Activations Density 0.022%

phrases related to military operations or units

references to "Bomber Command" and related military terminology

No Comments

No Known Activations