INDEX

Explanations

specific characters or sequences ("Q" followed by a number) within the text

oai_token-act-pair · gpt-3.5-turbo

instances of the term "Q" with varying frequencies, likely indicating a focus on specific technical terminology or concepts

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

ufact

-0.85

fulness

-0.84

 Lauder

-0.73

coni

-0.70

 Magicka

-0.69

 fullest

-0.65

 immobil

-0.63

fitting

-0.62

 afore

-0.62

milo

-0.61

POSITIVE LOGITS

UE

1.16

ues

1.02

uably

0.94

ued

0.93

WER

0.91

atari

0.89

atar

0.89

addafi

0.87

naire

0.87

DN

0.87

Activations Density 0.013%

specific characters or sequences ("Q" followed by a number) within the text

instances of the term "Q" with varying frequencies, likely indicating a focus on specific technical terminology or concepts

No Comments

No Known Activations