INDEX

Explanations

phrases related to depth or layers of information or ranking

oai_token-act-pair · gpt-3.5-turbo

references to varying levels of depth in contexts such as thoughts, strategies, or charts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 6-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.6.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.6.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

eer

-0.94

jee

-0.87

eers

-0.77

abad

-0.75

eering

-0.71

nee

-0.65

 Sabha

-0.62

rote

-0.61

ifiable

-0.60

 Allaah

-0.60

POSITIVE LOGITS

 perception

0.92

 depth

0.91

 penetration

0.86

 Depth

0.85

depth

0.74

galitarian

0.72

colour

0.71

 breadth

0.70

acies

0.69

 theorem

0.69

Activations Density 0.019%

phrases related to depth or layers of information or ranking

references to varying levels of depth in contexts such as thoughts, strategies, or charts

No Comments

No Known Activations