INDEX

Explanations

words related to actions or tasks that are not being done or are forbidden

oai_token-act-pair · gpt-3.5-turbo

instances of the word "does" and its variations

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

 Dise

-0.74

)=(

-0.73

Tai

-0.67

ulative

-0.66

case

-0.65

 palms

-0.65

 Handling

-0.65

 Methods

-0.64

Reviewer

-0.64

 Replacement

-0.63

POSITIVE LOGITS

ppel

1.09

omsday

0.92

berman

0.91

ozy

0.90

herty

0.90

not

0.88

 indeed

0.88

pez

0.88

 nothing

0.85

 likewise

0.84

Activations Density 0.134%

words related to actions or tasks that are not being done or are forbidden

instances of the word "does" and its variations

No Comments

No Known Activations