INDEX

Explanations

phrases related to scientific measurements or properties, particularly focusing on the concept of concentration gradients

oai_token-act-pair · gpt-3.5-turbo

terms related to concentration levels

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

ĪĴ

-0.91

mic

-0.82

\\\\\\\\

-0.74

sid

-0.71

pher

-0.66

 mans

-0.66

aired

-0.66

mer

-0.65

à¤¾

-0.63

 wives

-0.63

POSITIVE LOGITS

 concentration

0.98

 concentrations

0.87

uations

0.77

 quantities

0.75

encies

0.73

avorite

0.72

anguage

0.71

 Concent

0.71

 amounts

0.69

itated

0.69

Activations Density 0.013%

phrases related to scientific measurements or properties, particularly focusing on the concept of concentration gradients

terms related to concentration levels

No Comments

No Known Activations