INDEX

Explanations

references to coding errors or mistakes in programming code

oai_token-act-pair · gpt-3.5-turbo

instances of the word "oops" and similar expressions indicating mistakes or errors

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 Sask

-0.72

STD

-0.66

CES

-0.66

ISON

-0.64

Downloadha

-0.64

assetsadobe

-0.63

penter

-0.62

 circumstance

-0.60

 wounds

-0.59

 Magicka

-0.58

POSITIVE LOGITS

oop

1.61

oops

0.99

olicy

0.82

leans

0.81

arser

0.79

inion

0.79

oot

0.78

opa

0.77

ogle

0.77

Raid

0.77

Activations Density 0.004%

references to coding errors or mistakes in programming code

instances of the word "oops" and similar expressions indicating mistakes or errors

No Comments

No Known Activations