INDEX

Explanations

words containing the string "ils" with varying activation values

oai_token-act-pair · gpt-3.5-turbo

repeated mentions of a specific entity or subject matter, likely related to the context of listings or mentions of properties

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

¥ŀ

-0.89

ĵĺ

-0.85

ĳå£«

-0.83

etheless

-0.81

©¶æ

-0.80

©¶æ¥µ

-0.80

İĭ

-0.79

xus

-0.74

 unden

-0.72

ļé

-0.70

POSITIVE LOGITS

iblings

0.92

uminati

0.83

ands

0.82

waukee

0.82

pread

0.80

aints

0.79

inki

0.79

inx

0.78

gio

0.78

pace

0.78

Activations Density 0.013%

words containing the string "ils" with varying activation values

repeated mentions of a specific entity or subject matter, likely related to the context of listings or mentions of properties

No Comments

No Known Activations