INDEX

Explanations

words related to different levels of hierarchy or achievement

oai_token-act-pair · gpt-3.5-turbo

references to different tiers or ranks of achievement or authority

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Predators

-0.81

 Andromeda

-0.75

 Twins

-0.71

lez

-0.70

 Autob

-0.65

vous

-0.65

 Scor

-0.64

 Blessing

-0.64

 Pengu

-0.62

 Gamble

-0.62

POSITIVE LOGITS

chart

0.82

 levels

0.82

 heights

0.81

 level

0.80

level

0.80

 crossings

0.78

xual

0.76

 certific

0.73

atical

0.68

 attainment

0.68

Activations Density 0.022%

words related to different levels of hierarchy or achievement

references to different tiers or ranks of achievement or authority

No Comments

No Known Activations