INDEX

Explanations

unrelated incidents or topics

oai_token-act-pair · gpt-3.5-turbo

references to situations or incidents that are deemed unrelated

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

oise

-0.73

aneers

-0.68

hene

-0.67

addafi

-0.65

asure

-0.64

 Stre

-0.63

ige

-0.63

Wah

-0.63

=-=-=-=-=-=-=-=-

-0.62

Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯

-0.61

POSITIVE LOGITS

 unrelated

1.26

 minded

0.91

wise

0.84

worldly

0.81

lihood

0.81

 related

0.80

 thereto

0.78

merce

0.78

 unaffected

0.78

NESS

0.77

Activations Density 0.006%

**unrelated** incidents or topics

references to situations or incidents that are deemed unrelated

No Comments

No Known Activations

unrelated incidents or topics