INDEX

Explanations

contractions ending in 't

oai_token-act-pair · gpt-3.5-turbo

negative statements or prohibitions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

imity

-0.71

CI

-0.69

ULTS

-0.66

 è£ıè

-0.64

yk

-0.63

è¦ļéĨĴ

-0.61

ENDED

-0.60

 sidx

-0.59

ELD

-0.59

senal

-0.59

POSITIVE LOGITS

 afford

1.36

 rely

0.95

 ignore

0.95

 possibly

0.94

 deny

0.93

 underestimate

0.90

 imagine

0.89

 resist

0.89

 blame

0.86

 accuse

0.86

Activations Density 0.068%

contractions ending in 't

negative statements or prohibitions

No Comments

No Known Activations