INDEX

Explanations

instances where something is being treated or perceived differently

oai_token-act-pair · gpt-3.5-turbo

instances of the word "differently" to indicate varying perspectives or treatments

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

Not in Any Lists

No Comments

Negative Logits

advertising

-0.61

 Encyclopedia

-0.61

-0.60

eer

-0.59

 Nationals

-0.58

 reader

-0.58

 Zion

-0.58

 Prospect

-0.58

 Ammunition

-0.57

 Copenhagen

-0.57

POSITIVE LOGITS

iating

1.03

 situated

0.85

wcs

0.85

styles

0.84

iates

0.83

iator

0.81

 colored

0.80

colored

0.80

 coloured

0.79

 differentiated

0.77

Activations Density 0.006%

instances where something is being treated or perceived differently

instances of the word "differently" to indicate varying perspectives or treatments

No Comments

No Known Activations