INDEX

Explanations

the word "genre" with high activation values

oai_token-act-pair · gpt-3.5-turbo

references to different genres in media

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

urion

-0.85

riel

-0.84

 Lumpur

-0.80

amen

-0.71

ermanent

-0.69

erald

-0.66

 Wilhelm

-0.64

Uz

-0.64

 Lama

-0.63

administ

-0.63

POSITIVE LOGITS

genre

0.82

 genres

0.81

 fiction

0.76

ologies

0.75

 genre

0.74

allo

0.73

¥µ

0.71

 juices

0.69

adelphia

0.68

ĸļ

0.68

Activations Density 0.018%

the word "genre" with high activation values

references to different genres in media

No Comments

No Known Activations