INDEX

Explanations

mentions of specific acronyms 'AD' with varying importance levels

oai_token-act-pair · gpt-3.5-turbo

references to "AD" (Anno Domini), indicating historical and temporal contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

ĸļ

-0.87

ãĤ¡

-0.83

htt

-0.74

awaru

-0.69

href

-0.68

drop

-0.67

 Kubrick

-0.65

 Sturgeon

-0.65

Wem

-0.65

ãĤ©

-0.61

POSITIVE LOGITS

VERTISEMENT

1.32

vantage

1.06

MIN

1.01

VERTIS

0.98

venture

0.95

minist

0.91

visor

0.90

olf

0.88

vance

0.87

olph

0.87

Activations Density 0.009%

mentions of specific acronyms 'AD' with varying importance levels

references to "AD" (Anno Domini), indicating historical and temporal contexts

No Comments

No Known Activations