INDEX

Explanations

words or phrases containing the substring "wa"

oai_token-act-pair · gpt-3.5-turbo

forms of the verb "to be" 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 @ 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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Publisher

-0.68

 Luther

-0.68

 Soccer

-0.67

displayText

-0.67

croft

-0.66

rd

-0.65

toc

-0.64

 scales

-0.62

 Tables

-0.59

Catal

-0.59

POSITIVE LOGITS

velength

1.55

wa

1.30

aii

1.15

ÅĤ

1.02

Wa

1.01

isoft

1.00

pees

0.93

veland

0.93

anty

0.93

ascus

0.92

Activations Density 0.005%

words or phrases containing the substring "wa"

forms of the verb "to be" in various contexts

No Comments

No Known Activations