INDEX

Explanations

the word "arc" with varying activations, potentially indicating a focus on specific contexts or subjects related to "arc."

oai_token-act-pair · gpt-3.5-turbo

references to "arc" 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 @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

fulness

-0.86

tty

-0.77

lihood

-0.75

ãģŁ

-0.75

 à¤

-0.72

 Vive

-0.69

FUL

-0.69

mble

-0.67

liness

-0.65

gets

-0.65

POSITIVE LOGITS

otics

1.11

adian

0.98

adia

0.94

issa

0.93

aine

0.92

asm

0.92

inia

0.92

arc

0.91

itect

0.85

acia

0.83

Activations Density 0.007%

the word "arc" with varying activations, potentially indicating a focus on specific contexts or subjects related to "arc."

references to "arc" in various contexts

No Comments

No Known Activations