INDEX

Explanations

phrases indicating contrasting or conflicting information

oai_token-act-pair · gpt-3.5-turbo

contrasting statements or qualifiers that introduce a counterargument

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

ãĥ³ãĤ¸

-0.83

 pione

-0.80

ountain

-0.75

intosh

-0.75

esm

-0.74

roy

-0.73

isexual

-0.72

nes

-0.72

gio

-0.72

vo

-0.71

POSITIVE LOGITS

it

1.06

 suffice

0.94

 beware

0.83

 imagine

0.82

 remember

0.81

 underscores

0.80

 nonetheless

0.79

 fortunately

0.77

 nevertheless

0.76

 luckily

0.74

Activations Density 0.223%

phrases indicating contrasting or conflicting information

contrasting statements or qualifiers that introduce a counterargument

No Comments

No Known Activations