INDEX

Explanations

computer-related instructions and processes

oai_token-act-pair · gpt-3.5-turbo

structures emphasizing important information or instructions

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

ansas

-0.73

nered

-0.71

 Negro

-0.70

 Europeans

-0.66

 Indians

-0.66

 Plaint

-0.65

Jews

-0.62

 fathers

-0.61

 appell

-0.61

 Kenneth

-0.60

POSITIVE LOGITS

 optional

1.06

 handy

0.94

 easiest

0.94

 easier

0.91

advant

0.84

 FREE

0.83

 usually

0.83

optional

0.83

 easy

0.82

 advisable

0.82

Activations Density 0.731%

computer-related instructions and processes

structures emphasizing important information or instructions

No Comments

No Known Activations