INDEX

Explanations

phrases indicating an unnecessary or obvious statement

oai_token-act-pair · gpt-3.5-turbo

phrases indicating redundancy or unnecessary information

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

soever

-0.82

wagon

-0.70

ansas

-0.66

fram

-0.65

raped

-0.65

vati

-0.64

raising

-0.63

backs

-0.62

 IMAGES

-0.61

bage

-0.60

POSITIVE LOGITS

lege

0.68

¨

0.67

 clichÃ©

0.62

otide

0.62

 note

0.61

 assume

0.60

 conclude

0.59

Ĳ

0.58

say

0.57

 speculate

0.56

Activations Density 0.059%

phrases indicating an unnecessary or obvious statement

phrases indicating redundancy or unnecessary information

No Comments

No Known Activations