INDEX

Explanations

words related to separation or division of different entities

oai_token-act-pair · gpt-3.5-turbo

references to the concept of separation in various contexts

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

nz

-0.69

 wonders

-0.66

bye

-0.64

mA

-0.62

ãĥ¥

-0.61

enegger

-0.61

psc

-0.60

unc

-0.60

roll

-0.60

rouse

-0.59

POSITIVE LOGITS

 between

0.84

 separating

0.83

 sexes

0.83

 hairs

0.82

 Between

0.79

 from

0.72

icut

0.72

 apart

0.71

aration

0.70

 FROM

0.70

Activations Density 0.045%

words related to separation or division of different entities

references to the concept of separation in various contexts

No Comments

No Known Activations