INDEX

Explanations

phrases indicating errors or issues and providing alternative actions

oai_token-act-pair · gpt-3.5-turbo

the word "or" and related phrases indicating alternatives or options

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 Mane

-0.65

beit

-0.65

ocracy

-0.64

ochond

-0.57

Mech

-0.57

 Modes

-0.56

eval

-0.56

urized

-0.55

oret

-0.54

ynes

-0.54

POSITIVE LOGITS

lando

0.84

Else

0.79

acle

0.72

chard

0.70

acles

0.66

 subscribe

0.65

leans

0.65

idad

0.65

chid

0.63

 bookmark

0.63

Activations Density 0.027%

phrases indicating errors or issues and providing alternative actions

the word "or" and related phrases indicating alternatives or options

No Comments

No Known Activations