INDEX

Explanations

links or references within text

oai_token-act-pair · gpt-3.5-turbo

items or links to additional content or media sources

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

":[

-0.66

gag

-0.64

 attendant

-0.64

 easing

-0.64

uyomi

-0.62

 dash

-0.62

 centrif

-0.61

 locality

-0.58

 regul

-0.56

 alike

-0.56

POSITIVE LOGITS

Interstitial

0.80

ACTED

0.78

ixel

0.74

alky

0.72

Unknown

0.68

NetMessage

0.67

ructose

0.65

 Failed

0.65

ptin

0.65

 Requires

0.63

Activations Density 0.743%

links or references within text

items or links to additional content or media sources

No Comments

No Known Activations