INDEX

Explanations

phrases related to a specific brand or company named "Land"

oai_token-act-pair · gpt-3.5-turbo

references to "Land."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

inian

-0.70

staking

-0.65

 erection

-0.65

 ovarian

-0.63

 dating

-0.62

inus

-0.62

 unexplained

-0.60

ometimes

-0.60

hene

-0.59

ologically

-0.58

POSITIVE LOGITS

Land

2.12

Sub

0.83

iors

0.81

anka

0.67

Else

0.67

 Skies

0.65

 subs

0.65

Gra

0.65

usha

0.64

Bee

0.62

Activations Density 0.001%

phrases related to a specific brand or company named "Land"

references to "Land."

No Comments

No Known Activations