INDEX

Explanations

words related to different sectors such as legal, financial, public health, and digital

oai_token-act-pair · gpt-3.5-turbo

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

milo

-0.79

 Yorkers

-0.75

atism

-0.72

Americans

-0.65

ensibly

-0.63

 Telesc

-0.63

orsi

-0.63

Cath

-0.62

 Tactics

-0.61

imus

-0.60

POSITIVE LOGITS

 realm

1.34

 vicinity

1.29

 arena

1.21

 sphere

1.15

era

1.09

 context

1.08

 universe

1.07

osphere

1.07

 realms

1.06

 confines

1.06

Activations Density 0.343%

words related to different sectors such as legal, financial, public health, and digital

No Comments

No Known Activations