INDEX

Explanations

references to iconic or well-known entities

oai_token-act-pair · gpt-3.5-turbo

references to the concept of "iconic" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

©¶æ

-0.82

vere

-0.78

gans

-0.75

sterdam

-0.75

THER

-0.72

ND

-0.72

ften

-0.72

abet

-0.68

hner

-0.68

ying

-0.68

POSITIVE LOGITS

 iconic

1.07

 landmarks

0.90

 emblem

0.80

wcs

0.80

 icons

0.77

 icon

0.76

 artwork

0.75

 brands

0.75

 personalities

0.75

 recognizable

0.74

Activations Density 0.016%

references to iconic or well-known entities

references to the concept of "iconic" in various contexts

No Comments

No Known Activations