INDEX

Explanations

names of people or characters

oai_token-act-pair · gpt-3.5-turbo

proper nouns, particularly names and titles

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

oute

-0.83

ledged

-0.71

ably

-0.70

 tremend

-0.64

Dug

-0.61

 Showdown

-0.61

gian

-0.59

alities

-0.59

 Cassidy

-0.59

Ĥª

-0.59

POSITIVE LOGITS

pta

0.85

owship

0.85

gren

0.82

istics

0.78

minist

0.76

seek

0.76

heed

0.76

borg

0.75

quist

0.73

qv

0.73

Activations Density 0.112%

names of people or characters

proper nouns, particularly names and titles

No Comments

No Known Activations