INDEX

Explanations

words ending in 'ian'

oai_token-act-pair · gpt-3.5-turbo

the letter 'n'

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

xual

-0.73

EStream

-0.71

 penetrating

-0.69

 CLSID

-0.68

EStreamFrame

-0.67

MODE

-0.67

 acron

-0.64

 fractured

-0.62

 screwed

-0.61

 duplication

-0.61

POSITIVE LOGITS

aturally

1.05

ihil

1.05

umeric

1.05

ucle

1.03

uggets

1.02

onsense

1.01

imble

0.98

oun

0.97

ovation

0.97

aught

0.97

Activations Density 0.024%

words ending in 'ian'

the letter 'n'

No Comments

No Known Activations