INDEX

Explanations

phrases related to a person's characteristics or actions

oai_token-act-pair · gpt-3.5-turbo

references to individuals, particularly in specific locations or situations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

DragonMagazine

-0.85

 selves

-0.76

 bonuses

-0.66

 broadcasters

-0.66

tackle

-0.62

 EVENTS

-0.61

atars

-0.61

 consumers

-0.59

 commanders

-0.59

 trillions

-0.58

POSITIVE LOGITS

 whom

1.15

 whose

0.82

whose

0.80

sole

0.78

nikov

0.73

 surn

0.73

eln

0.71

who

0.70

okin

0.70

itol

0.68

Activations Density 0.279%

phrases related to a person's characteristics or actions

references to individuals, particularly in specific locations or situations

No Comments

No Known Activations