INDEX

Explanations

phrases related to hidden or secretive activities happening in the background

oai_token-act-pair · gpt-3.5-turbo

phrases related to "behind-the-scenes" content

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

QL

-0.73

 attm

-0.68

abama

-0.65

pha

-0.63

 msec

-0.63

 Cohn

-0.63

ashtra

-0.62

 warmed

-0.61

iless

-0.61

RIC

-0.60

POSITIVE LOGITS

 scenes

0.81

 ropes

0.76

 walls

0.74

 rails

0.73

 fences

0.73

 fence

0.73

 curve

0.71

agate

0.71

 backs

0.67

 doors

0.67

Activations Density 0.103%

phrases related to hidden or secretive activities happening in the background

phrases related to "behind-the-scenes" content

No Comments

No Known Activations