INDEX

Explanations

phrases related to education and school

oai_token-act-pair · gpt-3.5-turbo

references to attending or being involved in school

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

 genus

-0.72

ourke

-0.67

 species

-0.66

 Bundy

-0.62

 Ventura

-0.61

 Samar

-0.60

 Dominion

-0.60

 dispatch

-0.58

Za

-0.58

 Kham

-0.57

POSITIVE LOGITS

girls

0.92

pai

0.80

uates

0.78

VERTISEMENT

0.78

kids

0.78

work

0.76

neys

0.74

skirts

0.74

amiya

0.74

boy

0.73

Activations Density 0.054%

phrases related to education and school

references to attending or being involved in school

No Comments

No Known Activations