INDEX

Explanations

references to the word "potential."

oai_token-act-pair · gpt-3.5-turbo

the word "potential" 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 @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 textbooks

-0.70

 Whitney

-0.68

CDs

-0.62

 Sleeping

-0.61

 textbook

-0.61

Tig

-0.60

bows

-0.59

 Deng

-0.58

tex

-0.58

 Pirates

-0.58

POSITIVE LOGITS

ential

1.14

entially

1.13

loo

0.96

 proble

0.96

yip

0.94

ategory

0.91

cffff

0.90

hester

0.90

etheless

0.89

ividual

0.87

Activations Density 0.010%

references to the word "potential."

the word "potential" and its variations

No Comments

No Known Activations