INDEX

Explanations

occurrences of the word "left" followed by a number

oai_token-act-pair · gpt-3.5-turbo

instances of the word "left" indicating abandonment or lack of support

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

andise

-0.69

alez

-0.69

%]

-0.68

insula

-0.65

CoC

-0.64

Temperature

-0.62

alist

-0.62

acity

-0.61

Wan

-0.61

uracy

-0.61

POSITIVE LOGITS

overs

1.09

 undone

0.93

wing

0.90

 untreated

0.86

ward

0.81

wich

0.81

fing

0.75

handed

0.75

 Dise

0.73

 handed

0.71

Activations Density 0.029%

occurrences of the word "left" followed by a number

instances of the word "left" indicating abandonment or lack of support

No Comments

No Known Activations