INDEX

Explanations

the concept of going "back" to a previous state or location

oai_token-act-pair · gpt-3.5-turbo

phrases indicating a return to a previous state or reference point

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

anooga

-0.65

orst

-0.63

olulu

-0.58

uracy

-0.56

Us

-0.56

Specifically

-0.55

 Donation

-0.54

 ideal

-0.54

 whiff

-0.53

rowd

-0.53

POSITIVE LOGITS

stab

1.07

packing

1.06

packs

0.96

dated

0.92

gam

0.89

doors

0.88

wards

0.86

stairs

0.83

door

0.82

tracking

0.82

Activations Density 0.035%

the concept of going "back" to a previous state or location

phrases indicating a return to a previous state or reference point

No Comments

No Known Activations