INDEX

Explanations

verbs related to causing harm or damage

oai_token-act-pair · gpt-3.5-turbo

terms related to destruction or annihilation

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

DragonMagazine

-0.77

soType

-0.76

POL

-0.71

liner

-0.67

ETF

-0.66

Clar

-0.65

matic

-0.65

cone

-0.63

travel

-0.63

twitch

-0.63

POSITIVE LOGITS

 havoc

1.09

roying

0.86

ukong

0.71

IED

0.68

 wrought

0.68

 arte

0.68

ados

0.68

adoes

0.67

adelphia

0.67

 rubble

0.66

Activations Density 0.049%

verbs related to causing harm or damage

terms related to destruction or annihilation

No Comments

No Known Activations