INDEX

Explanations

location-related phrases, especially involving streets

oai_token-act-pair · gpt-3.5-turbo

references to violence occurring in urban areas, particularly focusing on "streets."

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

igham

-0.79

emort

-0.72

opsis

-0.72

ITED

-0.72

itors

-0.71

itor

-0.68

isman

-0.67

inite

-0.64

 Blanc

-0.63

ibilities

-0.62

POSITIVE LOGITS

cape

1.30

fare

1.05

walker

0.94

 streets

0.92

walk

0.92

ffiti

0.84

wear

0.82

ways

0.80

walking

0.78

cars

0.78

Activations Density 0.020%

location-related phrases, especially involving streets

references to violence occurring in urban areas, particularly focusing on "streets."

No Comments

No Known Activations