INDEX

Explanations

occurrences of the word "post."

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

URLs or email addresses ending with "post.com"

oai_token-act-pair · gpt-3.5-turbo

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

 Harm

-0.75

estial

-0.72

BILITY

-0.71

aughs

-0.70

BILITIES

-0.69

 Coch

-0.68

 Brill

-0.68

 Minutes

-0.68

ggle

-0.67

BUS

-0.64

POSITIVE LOGITS

gres

1.19

post

1.12

mortem

0.97

greSQL

0.93

natal

0.91

posts

0.85

doctoral

0.83

aldehyde

0.82

etheless

0.81

fix

0.81

Activations Density 0.006%

occurrences of the word "post."

URLs or email addresses ending with "post.com"

No Comments

No Known Activations