INDEX

Explanations

words related to national entities or activities

oai_token-act-pair · gpt-3.5-turbo

references to national issues or themes

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

IFrame

Link

Negative Logits

omething

-0.82

xual

-0.77

eva

-0.72

ÄŁ

-0.71

kick

-0.69

Ware

-0.68

ÃŁ

-0.67

ever

-0.67

hops

-0.66

riction

-0.66

POSITIVE LOGITS

 Geographic

0.84

wide

0.83

natureconservancy

0.78

ized

0.76

 anthem

0.75

istic

0.74

 outcry

0.74

izations

0.73

 championship

0.72

ities

0.71

Activations Density 0.039%

words related to national entities or activities

references to national issues or themes

No Known Activations