INDEX

Explanations

titles and references to movies

oai_token-act-pair · gpt-3.5-turbo

references to movies

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

raham

-0.71

urst

-0.68

²¾

-0.67

ilities

-0.67

utter

-0.66

existent

-0.66

bleacher

-0.66

 Haitian

-0.66

ĵĺ

-0.66

İĭ

-0.65

POSITIVE LOGITS

 theaters

1.08

 theater

1.06

goers

1.04

 movies

1.03

 movie

0.96

 theatre

0.96

 Movies

0.89

 theat

0.87

 buffs

0.86

eers

0.86

Activations Density 0.025%

titles and references to movies

references to movies

No Comments

No Known Activations