INDEX

Explanations

terms related to mazes

oai_token-act-pair · gpt-3.5-turbo

references to mazes or labyrinths

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

OTT

-0.74

riter

-0.74

fixed

-0.74

ulf

-0.74

arers

-0.71

rities

-0.68

monds

-0.67

ding

-0.66

arer

-0.66

ded

-0.65

POSITIVE LOGITS

 maze

1.05

yrinth

1.05

 labyrinth

0.99

 Maze

0.92

ĸļ

0.79

ingly

0.76

 Ples

0.71

xon

0.66

 crawling

0.64

 cavern

0.64

Activations Density 0.019%

terms related to mazes

references to mazes or labyrinths

No Comments

No Known Activations