INDEX

Explanations

phrases related to people and interactions

oai_token-act-pair · gpt-3.5-turbo

references to individuals, particularly women and friends, in various contexts

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

 è£ıç

-0.74

;;;;;;;;;;;;

-0.71

SPONSORED

-0.69

 olds

-0.68

ĸļ

-0.68

assies

-0.66

aeda

-0.66

estyles

-0.66

ologies

-0.63

ernels

-0.62

POSITIVE LOGITS

alyst

0.84

 named

0.82

 sergeant

0.77

 spokeswoman

0.77

 staffer

0.77

 wrote

0.75

 reviewer

0.74

 responded

0.74

 spokesman

0.72

 spokesperson

0.72

Activations Density 0.302%

phrases related to people and interactions

references to individuals, particularly women and friends, in various contexts

No Comments

No Known Activations