INDEX

Explanations

a specific word with the letters 'qu' followed by a high activation value

oai_token-act-pair · gpt-3.5-turbo

the word "qu" in various contexts

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

 Norn

-0.68

 Heidi

-0.66

ASED

-0.64

 Dayton

-0.64

 foreskin

-0.63

 handlers

-0.63

ATURE

-0.61

SEA

-0.61

 Bakr

-0.60

Aus

-0.59

POSITIVE LOGITS

qu

4.43

QU

2.32

quer

2.29

qua

2.14

quin

2.11

quit

2.11

ques

2.00

Qu

1.90

quet

1.82

que

1.80

Activations Density 0.021%

a specific word with the letters 'qu' followed by a high activation value

the word "qu" in various contexts

No Comments

No Known Activations