INDEX

Explanations

specific mentions of the acronym "DS" at different activation levels

oai_token-act-pair · gpt-3.5-turbo

references to "DS" and prominent figures, especially in discussions about punditry

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 torches

-0.80

 Scotia

-0.79

ting

-0.72

 Australians

-0.71

aneous

-0.70

rals

-0.68

 vessels

-0.66

 waters

-0.66

tering

-0.65

 captcha

-0.64

POSITIVE LOGITS

ense

0.87

ocial

0.84

ql

0.83

ocrates

0.82

etics

0.80

ilon

0.80

Ã©t

0.80

imon

0.78

kers

0.76

pod

0.76

Activations Density 0.021%

specific mentions of the acronym "DS" at different activation levels

references to "DS" and prominent figures, especially in discussions about punditry

No Comments

No Known Activations