INDEX

Explanations

the word "ten" with varying intensities

oai_token-act-pair · gpt-3.5-turbo

the term "ten."

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

 swoop

-0.65

CRIPTION

-0.62

exh

-0.61

bl

-0.60

 Checks

-0.60

QR

-0.60

 flares

-0.58

 Galile

-0.58

BST

-0.57

 junk

-0.57

POSITIVE LOGITS

ten

4.63

Ten

1.64

ts

1.45

te

1.36

Ten

1.32

ted

1.31

ter

1.30

ting

1.28

sen

1.28

ton

1.27

Activations Density 0.011%

the word "ten" with varying intensities

the term "ten."

No Comments

No Known Activations