INDEX

Explanations

mentions of being informed or knowledgeable about certain information

oai_token-act-pair · gpt-3.5-turbo

references to awareness or knowledge of particular situations or facts

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

IFrame

Link

Negative Logits

Textures

-0.64

amen

-0.63

cell

-0.63

wrap

-0.63

 Warriors

-0.62

irrel

-0.61

nic

-0.60

 averages

-0.59

avored

-0.59

pen

-0.59

POSITIVE LOGITS

ledged

1.22

 Lauder

0.88

ledge

0.82

ness

0.79

 Aware

0.78

 aware

0.77

isance

0.74

itaire

0.74

 Dragonbound

0.73

eering

0.73

Activations Density 0.026%

mentions of being informed or knowledgeable about certain information

references to awareness or knowledge of particular situations or facts

No Known Activations