INDEX

Explanations

phrases or words enclosed in square brackets, possibly related to quotations or attributions

oai_token-act-pair · gpt-3.5-turbo

various closing brackets or punctuation marks

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Ļ½

-0.77

ĻĤ

-0.75

Ń·

-0.75

ĪĴ

-0.74

©¶æ¥µ

-0.73

acies

-0.73

ãĥŃ

-0.70

İĭ

-0.69

±

-0.68

mbuds

-0.66

POSITIVE LOGITS

selves

0.72

,"

0.67

},"

0.66

cape

0.63

alone

0.62

thumbnails

0.62

."[

0.62

...]

0.60

".[

0.60

],

0.58

Activations Density 0.056%

phrases or words enclosed in square brackets, possibly related to quotations or attributions

various closing brackets or punctuation marks

No Comments

No Known Activations