INDEX

Explanations

words related to endorsements and recommendations

oai_token-act-pair · gpt-3.5-turbo

references to endorsements

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

istics

-0.71

istic

-0.69

rooms

-0.68

folk

-0.66

©¶æ¥µ

-0.66

 nerv

-0.65

assic

-0.63

fare

-0.63

rug

-0.63

IRO

-0.62

POSITIVE LOGITS

 endorsements

1.12

 endorsing

1.10

 endorsement

1.00

 endorse

0.99

 endorsed

0.91

 endors

0.84

 candidate

0.80

hovah

0.78

 loyalty

0.78

iliate

0.78

Activations Density 0.025%

words related to endorsements and recommendations

references to endorsements

No Comments

No Known Activations