INDEX

Explanations

phrases related to a specific software or service named "Waze"

oai_token-act-pair · gpt-3.5-turbo

variations of the word "maze."

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

orders

-0.67

 ammon

-0.64

nan

-0.64

ials

-0.64

mbuds

-0.63

tons

-0.63

 autonomy

-0.60

dated

-0.58

talk

-0.58

ietal

-0.57

POSITIVE LOGITS

aze

1.33

azes

0.93

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

0.78

ecake

0.78

 Flavoring

0.76

edo

0.74

eer

0.70

ADRA

0.70

HOU

0.69

lda

0.69

Activations Density 0.009%

phrases related to a specific software or service named "Waze"

variations of the word "maze."

No Comments

No Known Activations