INDEX

Explanations

corrections and inaccuracies in news articles

oai_token-act-pair · gpt-3.5-turbo

instances of corrections or updates to previously reported information

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

Mods

-0.74

SpaceEngineers

-0.70

Ń·

-0.69

ãģ®éŃĶ

-0.68

soDeliveryDate

-0.63

Ì¶

-0.63

 passively

-0.63

urses

-0.62

Frameworks

-0.62

 playbook

-0.62

POSITIVE LOGITS

 corrected

1.06

 incorrect

1.06

 typo

1.05

 incorrectly

1.03

 spelling

0.97

Clar

0.94

 clarified

0.93

 clarification

0.92

 mistakenly

0.90

orrect

0.89

Activations Density 0.133%

corrections and inaccuracies in news articles

instances of corrections or updates to previously reported information

No Comments

No Known Activations