INDEX

Explanations

words related to proper nouns in news and media contexts

oai_token-act-pair · gpt-3.5-turbo

repeated instances of the substring "ze"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

paced

-0.88

rooms

-0.75

 glim

-0.72

acles

-0.72

tered

-0.71

stri

-0.71

runners

-0.67

ordable

-0.66

folk

-0.65

woman

-0.64

POSITIVE LOGITS

zinski

1.37

lda

1.17

cki

1.04

ppelin

1.04

ichen

0.93

uble

0.84

ÅĤ

0.83

ppe

0.82

ihu

0.81

zx

0.81

Activations Density 0.036%

words related to proper nouns in news and media contexts

repeated instances of the substring "ze"

No Comments

No Known Activations