INDEX

Explanations

terms related to finality or completion

oai_token-act-pair · gpt-3.5-turbo

references to conclusive or culminating events

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

stadt

-0.72

heid

-0.67

 Everywhere

-0.64

thens

-0.63

holes

-0.62

then

-0.62

LI

-0.62

shire

-0.62

kun

-0.61

velt

-0.59

POSITIVE LOGITS

 curtain

0.84

iminary

0.83

 installment

0.79

 showdown

0.78

 FANTASY

0.77

 touches

0.74

ebin

0.72

 resting

0.71

 exam

0.70

 countdown

0.70

Activations Density 0.019%

terms related to finality or completion

references to conclusive or culminating events

No Comments

No Known Activations