INDEX

Explanations

numeric comparisons, such as 'more than'

oai_token-act-pair · gpt-3.5-turbo

comparative phrases indicating magnitude or quantity

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

ALE

-0.76

 direction

-0.67

Alz

-0.65

ALSE

-0.63

Interstitial

-0.62

umm

-0.62

ModLoader

-0.61

 modulation

-0.61

Accessory

-0.60

oris

-0.60

POSITIVE LOGITS

whelming

0.82

etheless

0.67

whel

0.67

isans

0.65

grid

0.64

atos

0.63

 SetFontSize

0.62

 Hours

0.62

lando

0.60

 than

0.59

Activations Density 0.063%

numeric comparisons, such as 'more than'

comparative phrases indicating magnitude or quantity

No Comments

No Known Activations