INDEX

Explanations

phrases related to returning or regaining things or activities

oai_token-act-pair · gpt-3.5-turbo

references to returning or going back to a previous state or condition

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

uously

-0.73

 understatement

-0.71

ashi

-0.70

uary

-0.65

 Hots

-0.62

 Guardiola

-0.60

cture

-0.60

"]=>

-0.59

ities

-0.59

idis

-0.57

POSITIVE LOGITS

dated

0.93

doors

0.89

packs

0.87

GROUND

0.87

tracking

0.84

door

0.83

trace

0.83

stab

0.82

fired

0.80

haul

0.78

Activations Density 0.030%

phrases related to returning or regaining things or activities

references to returning or going back to a previous state or condition

No Comments

No Known Activations