INDEX

Explanations

references to interviews or statements made in a specific timeframe

oai_token-act-pair · gpt-3.5-turbo

references to interviews or discussions surrounding specific topics or individuals

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 repeal

-0.84

 unconstitutional

-0.81

 deport

-0.80

 Rodham

-0.80

 needles

-0.78

 Klux

-0.77

SPONSORED

-0.76

 nonviolent

-0.76

 aspirin

-0.75

 granite

-0.75

POSITIVE LOGITS

 Razer

1.19

 devs

1.18

 Verge

1.13

 Ubisoft

1.10

 Geek

1.07

 Wired

1.06

 Nvidia

1.05

IGN

1.03

 Developer

1.00

 Developers

1.00

Activations Density 0.871%

references to interviews or statements made in a specific timeframe

references to interviews or discussions surrounding specific topics or individuals

No Comments

No Known Activations