INDEX

Explanations

titles or sections of academic texts or articles

oai_token-act-pair · gpt-3.5-turbo

instances of the word "Edit" 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 @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

milo

-0.88

ciating

-0.84

Interstitial

-0.67

IRD

-0.66

wagen

-0.65

 pigeon

-0.64

soDeliveryDate

-0.61

jee

-0.61

 behavi

-0.61

knife

-0.59

POSITIVE LOGITS

edit

0.77

 edit

0.72

itals

0.70

ipedia

0.69

orship

0.68

edia

0.68

ril

0.67

arus

0.65

iton

0.64

spring

0.64

Activations Density 0.016%

titles or sections of academic texts or articles

instances of the word "Edit" in various contexts

No Comments

No Known Activations