INDEX

Explanations

phrases related to things being of a low quality or level

oai_token-act-pair · gpt-3.5-turbo

references to low socioeconomic status or low quality

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

tnc

-0.85

andise

-0.79

 Mahjong

-0.75

ophers

-0.74

itutional

-0.71

 Orient

-0.70

thus

-0.70

cientious

-0.69

Pok

-0.68

arium

-0.68

POSITIVE LOGITS

 enough

0.92

 pitched

0.83

est

0.80

 hanging

0.78

 dipping

0.78

 priced

0.76

 altitude

0.74

 doses

0.74

 ceilings

0.73

 paced

0.72

Activations Density 0.028%

phrases related to things being of a low quality or level

references to low socioeconomic status or low quality

No Comments

No Known Activations