INDEX

Explanations

word pairs related to speculation and assumption

oai_token-act-pair · gpt-3.5-turbo

references to technical or regulatory details

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

ividual

-0.85

irlf

-0.77

ogun

-0.77

ancial

-0.73

uilt

-0.69

othal

-0.69

itably

-0.68

escription

-0.67

ecause

-0.67

Ã©e

-0.66

POSITIVE LOGITS

 antioxid

0.85

Balt

0.76

CLASSIFIED

0.74

natureconservancy

0.73

 behavi

0.69

NetMessage

0.68

GoldMagikarp

0.67

 Seym

0.67

Interstitial

0.66

WAYS

0.66

Activations Density 4.811%

word pairs related to speculation and assumption

references to technical or regulatory details

No Comments

No Known Activations