INDEX

Explanations

phrases related to comparison or contrast

oai_token-act-pair · gpt-3.5-turbo

instances of the word "and" or similar conjunctions in lists

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

cies

-0.68

itionally

-0.68

itions

-0.67

itives

-0.66

orer

-0.65

que

-0.65

hens

-0.64

coat

-0.63

enta

-0.62

eed

-0.62

POSITIVE LOGITS

rongh

0.65

 somew

0.62

 recomm

0.59

 diminishing

0.59

Aph

0.59

 Ares

0.58

uana

0.58

 disqual

0.58

 grav

0.57

accompanied

0.57

Activations Density 0.206%

phrases related to comparison or contrast

instances of the word "and" or similar conjunctions in lists

No Comments

No Known Activations