INDEX

Explanations

personal responsibility and action-related words and phrases

oai_token-act-pair · gpt-3.5-turbo

references to "this" or "that" in a document

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

sts

-1.01

²¾

-0.93

anches

-0.92

lations

-0.92

arest

-0.92

pps

-0.91

quarters

-0.91

idelines

-0.90

warts

-0.88

cakes

-0.84

POSITIVE LOGITS

 sucker

1.12

 opportunity

1.00

 stuff

1.00

 kind

1.00

 same

0.99

 thing

0.99

guy

0.98

 information

0.96

 particular

0.96

 trope

0.94

Activations Density 0.143%

personal responsibility and action-related words and phrases

references to "this" or "that" in a document

No Comments

No Known Activations