INDEX

Explanations

phrases related to softening or easing something

oai_token-act-pair · gpt-3.5-turbo

terms associated with mitigating negative effects or concerns

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

recated

-0.70

arro

-0.70

ogg

-0.70

itars

-0.68

supported

-0.68

 Pett

-0.66

owned

-0.65

ALT

-0.65

 è£ıç

-0.64

 Fairfax

-0.64

POSITIVE LOGITS

 airflow

0.81

 wearer

0.79

 downward

0.78

 considerably

0.77

 wrinkles

0.74

lid

0.73

facade

0.73

 cred

0.72

 flow

0.71

 coupling

0.70

Activations Density 0.298%

phrases related to softening or easing something

terms associated with mitigating negative effects or concerns

No Comments

No Known Activations