INDEX

Explanations

words related to preferences or popularity

oai_token-act-pair · gpt-3.5-turbo

terms related to favorites or preferences

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

inas

-0.84

heed

-0.83

amping

-0.82

ural

-0.81

informed

-0.81

ijk

-0.80

ufact

-0.78

aping

-0.75

awed

-0.75

ldon

-0.75

POSITIVE LOGITS

 favorite

0.94

 Favorite

0.93

Favorite

0.91

 favorites

0.88

 Favor

0.78

 favourite

0.78

favorite

0.75

 haunt

0.74

 amongst

0.74

 favourites

0.73

Activations Density 0.012%

words related to preferences or popularity

terms related to favorites or preferences

No Comments

No Known Activations