INDEX

Explanations

mentions of the word "tar" in the text

oai_token-act-pair · gpt-3.5-turbo

references to tar

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

nesota

-0.85

iasco

-0.82

etts

-0.81

alian

-0.81

ĨĴ

-0.80

assetsadobe

-0.78

shire

-0.77

 Startup

-0.76

ettings

-0.74

ļéĨĴ

-0.73

POSITIVE LOGITS

tar

1.06

 sands

1.01

balls

0.83

pee

0.82

 smear

0.81

anium

0.81

 blister

0.79

ried

0.74

ãĥ£

0.73

azo

0.73

Activations Density 0.016%

mentions of the word "tar" in the text

references to tar

No Comments

No Known Activations