INDEX

Explanations

references to things that are not genuine, especially related to news or items

oai_token-act-pair · gpt-3.5-turbo

terms related to misinformation or false narratives

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

xual

-0.93

ands

-0.80

arching

-0.75

bard

-0.73

hens

-0.73

}}}

-0.71

APTER

-0.70

pins

-0.70

UNCH

-0.69

ishops

-0.69

POSITIVE LOGITS

pas

0.90

 fake

0.74

IDs

0.72

ument

0.71

ulously

0.68

 sounding

0.67

gery

0.65

ulent

0.65

tan

0.65

 reef

0.65

Activations Density 0.030%

references to things that are not genuine, especially related to news or items

terms related to misinformation or false narratives

No Comments

No Known Activations