INDEX

Explanations

references to 'drop' features or actions

oai_token-act-pair · gpt-3.5-turbo

references to "drop" in various contexts

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

paio

-0.89

gregation

-0.81

apo

-0.73

ILLE

-0.72

aldo

-0.71

urally

-0.70

 Gutenberg

-0.70

ancial

-0.69

Seg

-0.65

eur

-0.65

POSITIVE LOGITS

down

1.06

kick

1.05

downs

1.03

phrine

0.91

lights

0.89

bows

0.85

outs

0.85

drop

0.83

backs

0.82

 downed

0.82

Activations Density 0.019%

references to 'drop' features or actions

references to "drop" in various contexts

No Comments

No Known Activations