INDEX

Explanations

references to news stories and headlines

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

references to news articles and stories

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 12-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.hook_resid_post

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.11.hook_resid_post

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

awa

-0.78

 guiActiveUnfocused

-0.77

%%

-0.74

mble

-0.73

xual

-0.72

anca

-0.71

ably

-0.68

 Seym

-0.68

Ã¨re

-0.66

ucket

-0.66

POSITIVE LOGITS

 stories

1.01

 headlines

0.82

 picks

0.74

 tales

0.73

 articles

0.73

 news

0.73

 tips

0.69

 Stories

0.69

 Opinion

0.67

 tech

0.66

Activations Density 0.062%

references to news stories and headlines

references to news articles and stories

No Comments

No Known Activations