INDEX

Explanations

phrases indicating that something has been created, established, or produced

oai_token-act-pair · gpt-3.5-turbo

instances of the word "made" 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

 stripes

-0.67

 stripe

-0.64

 Adin

-0.61

 Rutherford

-0.59

 conditioning

-0.58

IRE

-0.58

 herbs

-0.58

 ding

-0.57

Mub

-0.57

 prick

-0.57

POSITIVE LOGITS

ensibly

0.82

 sure

0.81

roid

0.79

solete

0.78

unal

0.78

ventus

0.76

omet

0.75

ogyn

0.75

acons

0.75

netflix

0.75

Activations Density 0.033%

phrases indicating that something has been created, established, or produced

instances of the word "made" in various contexts

No Comments

No Known Activations