INDEX

Explanations

neural network-related terms

oai_token-act-pair · gpt-3.5-turbo

phrases indicating future goals or aspirations

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

xit

-0.71

Dial

-0.69

 cellar

-0.69

 supper

-0.67

 flask

-0.67

 aquarium

-0.66

ctions

-0.65

values

-0.63

Russ

-0.63

bard

-0.62

POSITIVE LOGITS

Ļ

1.35

©¶æ¥µ

1.00

aders

0.82

µ

0.75

Ģ

0.74

 Enemy

0.73

irlfriend

0.72

ç¥ŀ

0.70

ombat

0.70

ulse

0.69

Activations Density 0.000%

neural network-related terms

phrases indicating future goals or aspirations

No Comments

No Known Activations