INDEX

Explanations

phrases related to discussing or commenting on different topics

oai_token-act-pair · gpt-3.5-turbo

references to quantities or amounts of things

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

ngth

-0.97

gerald

-0.77

Pwr

-0.74

icone

-0.71

viks

-0.70

aurus

-0.66

strom

-0.65

xp

-0.65

eers

-0.63

umers

-0.62

POSITIVE LOGITS

 topics

1.14

 topic

1.11

 upcoming

0.95

 impending

0.95

 hypothetical

0.94

 aspects

0.92

 aspect

0.89

 specifics

0.88

 matters

0.84

 obscure

0.84

Activations Density 0.408%

phrases related to discussing or commenting on different topics

references to quantities or amounts of things

No Comments

No Known Activations