INDEX

Explanations

phrases related to hidden aspects or underlying factors

oai_token-act-pair · gpt-3.5-turbo

characters or letter sequences that are part of various words often related to negative or undesirable themes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

 Yankee

-0.63

sama

-0.61

 Oath

-0.60

busters

-0.58

EStream

-0.57

Wan

-0.57

Sym

-0.57

 famous

-0.56

Zi

-0.56

 Patriot

-0.56

POSITIVE LOGITS

ases

0.96

atch

0.96

asing

0.94

orde

0.91

ibling

0.89

arted

0.87

ased

0.87

ited

0.87

anked

0.87

ighty

0.87

Activations Density 0.089%

phrases related to hidden aspects or underlying factors

characters or letter sequences that are part of various words often related to negative or undesirable themes

No Comments

No Known Activations